COM 2022 Proceedings

Proceedings
of the
61st Conference
of Metallurgists,
COM 2022
Proceedings of the 61st Conference of Metallurgists,
COM 2022
Proceedings of the 61st
Conference of Metallurgists,
COM 2022
Metallurgy and Materials Society of CIM
Metallurgy and Materials Society of CIM
Westmount, QC, Canada
ISBN 978-3-031-17424-7 ISBN 978-3-031-17425-4 (eBook)

https://doi.org/10.1007/978-3-031-17425-4
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the
whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, expressed or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Thank you to our proceedings sponsors.
COM 2022 Organizing Committee
and Symposium Organizers
Conference Chair
Janice Zinck, Geoscience and Mines, Government of Nova Scotia
Secretary
Roki Fukuzawa, Hatch
Student Activities
Sidney Omelon, McGill University
“Big Idea” Transformational R&D Plenary Chair
Sumanth Shankar, McMaster University
Symposium Organizers
Advances in Materials Manufacturing VI – Existing and Emerging Materials

Chair: Peng Peng, University of Waterloo
Yu Zou, University of Toronto
Daolun Chen, Ryerson University
Abu Syed Kabir, Carleton University
Xialiang Jin, University of British Columbia
Priti Wanjara, National Research Council Canada
viii COM 2022 Organizing Committee and Symposium Organizers
Javad Gholipour Baradari, National Research Council Canada

Clodualdo Aranas, University of New Brunswick
Deep Decarbonization Pathways for Pyrometallurgical Processes: Opportunities
& Challenges
Chair: Elmira Moosavi, École de technologie supérieure (ÉTS)
Sina Mostaghel, SNC-Lavalin
Jafar Safarian, Norwegian University of Science and Technology
In-Ho Jung, SNU, South Korea
Isabelle Nolet, Hatch, South Africa
Electrochemical Degradation of Multi-Component Materials
Chair: Penghao Xiao, Dalhousie University
James Noël, Western University
Chongyin Yang, Dalhousie University
Michael Metzger, Dalhousie University
Energy and Environmental Materias
Chair: Jinhyuk Lee, McGill University
Yu Zou, University of Toronto
Light Metals for the Transportation Industry
Chair: Mihaiela Isac, McGill University
Julie Lévesque, Québec Metallurgy Centre
Dimitry Sediako, University of British Columbia
Abdallah Elsayed, University of Guelph
Jessica Hiscocks, Kingston Process Metallurgy
Mohsen Mohammadi, University of New Brunswick
Roderick Guthrie, McGill University
Carsten Siemers, Institute for Materials Science, Technische Universität Braunschweig
Sumanth Shankar, McMaster University
Processing of Critical Materials
Co-Chair: Niels Verbaan, SGS Natural Resources, Metallurgy and Consulting
Co-Chair: Mike Johnson, SGS Natural Resources
Gisele Azimi, University of Toronto
John Goode, J.R. Goode and Associates
Ian London, C2M2A
Towards Sustainable Circularity: Mining to Materials
Co-Chair: Brittany MacKinnon, Hatch
Co-Chair: Christopher Pickles, Queen’s University
Charlotte Gibson, Queen’s University
Qian Zhang, Queen’s University
Karen Chovan, Enviro Integration Strategies Inc.
Margarete Kalin, Boojum Research Limited
Gillian Holcroft, Canada Mining Innovation Council
Douglas Kao, Artisanal Gold Council
Laurie Reemeyer, Resourceful Pathways
Lorena Tere, Apple
Peer Reviewers
We gratefully acknowledge the time and expertise generously provided by our volun-
teer peer reviewers.
Advances in Materials Manufacturing VI – Existing and Emerging Materials
Peng Peng, University of Waterloo
Deep Decarbonization Pathways for Pyrometallurgical Processes: Opportunities
& Challenges
Elmira Moosavi, École de technologie supérieure (ÉTS)
Sina Mostaghel, SNC-Lavalin
Isabelle Nolet, Hatch
Jafar Safarian, Norwegian University of Science and Technology
Electrochemical Degradation of Multi-Component Materials
Penghao Xiao, Dalhousie University
Energy and Environmental Materials
Jinhyuk Lee, McGill University
Light Metals for the Transportation Industry
Abdallah Elsayed, University of Guelph
Roderick Guthrie, McGill University
Jessica Hiscocks, Kingston Process Metallurgy Inc.
Mihaiela Isac, McGill University
Julie Lévesque, Québec Metallurgy Centre
Mohsen Mohammadi, University of New Brunswick
Carsten Siemers, Institute for Materials Science, Technische Universität Braunschweig
Processing of Critical Materials
Gisele Azimi, University of Toronto
Jean-François Boulanger, Université du Québec en Abitibi-Témiscamingue
Alex Cao, Euro Manganese
Mike Dry, Arithmetek
Alex Forstner, Cyclic Materials
x Peer Reviewers
John Goode, J.R. Goode and Associates

Sunil Jayasekera, SJ Mets
Mike Johnson, SGS Natural Resources
Wenying Liu, University of British Columbia
William Mercer, Bill Mercer Consulting
Georgiana Moldoveanu, University of Toronto
Ron Molnar, MetNetH2O Inc.
Yemi Oyediran, NOK Associates Limited
Niels Verbaan, SGS Natural Resources
Ben Yu, National Research Council Canada
Towards Sustainable Circularity: Mining to Materials
Charlotte Gibson, Queen’s University
Bryn Harris, Independent Consultant
Gillian Holcroft, Canada Mining Innovation Council
Margarete Kalin, Boojum Research Limited
Britt MacKinnon, Hatch
Chris Pickles, Queen’s University
Samira Sokhanvaran, Hatch
Lorena Tere, Apple
Qian Zhang, Queen’s University
Contents
Part I Advances in Materials Manufacturing VI – Existing and Emerging

Materials
Al Alloys and Manufacturing Processes for Lightweight Applications

in Electric Vehicles �� 3
Anita Hu, Henry Hu, Yuxian Li, Sufeng Liu, and Wutian Shen
Laser Powder Bed Fusion of M789 Steel on Wrought N709

Steel Substrate�� 9
Kudakwashe Nyamuchiwa, Yuan Tian, Kanwal Chadha, and Clodualdo Aranas Jr

Monotonic and Cyclic Deformation Behavior of a Silafont®-36
Cast Aluminum Alloy in an Overaged Condition �� 15
S. S. Dash, D. J. Li, X. Q. Zeng, D. Y. Li, and D. L. Chen
Fabrication of Thin Rotary Tools Through a Combination

of Cross-Rolling and Austempering-Tempering�� 21
H. M. Shirazi, R. Georges, B. Ugulino, R. Hernández, and C. Blais
Heat Treatment of Multi-Material Additively Manufactured

Maraging Steel and Stellite Alloy �� 25
Jubert Pasco, Yuan Tian, Kanwal Chadha, and Clodualdo Aranas Jr.
Microstructural Characterization of Pack-Borided H13 Tool Steel

Processed Using Directed Energy Deposition�� 29
Owen Craig, Riley Roache, and Kevin Plucknett

Influence of Layer Thickness Upon the Dimensional Accuracy and Surface
Roughness of AISI D2 Tool Steel Manufactured Using Directed Energy
Deposition �� 39
S. M. T. Omar and K. P. Plucknett
xii Contents

Laser Ablation of Thin Films for Flexible Sensor Manufacturing�� 43
Ningyue Mao, Pablo D. Enrique, Y. Norman Zhou, and Peng Peng

Transient Softening at the Fusion Boundary of Spot Welds
in Advanced High-Strength Steels: Dark Horse
of the Heat-Affected Zone �� 47
Oleksii Sherepenko, Dileep Ramachandran, Alireza Mohamadizadeh,
and Elliot Biro
Synthesis of Fast-Weathering Nutrient-Doped Silicate Minerals

and Their Prospect as a Negative Emissions Technology �� 53
Asif Ali, Yi Wai Chiang, and Rafael M. Santos
Direct Laser Writing of Copper/Copper Oxide Patterns for Emerging

Roles in Advanced Electronics�� 59
Joshua Jones and Peng Peng
Open Pore Effect on Structural Adhesive Joining Using

an Epoxy Adhesive �� 65
M. M. Tiwari, S. Noormohammed, X. -G. Chen, and D. K. Sarkar
Tribological Performance of PH 13-8Mo Maraging Stainless Steel

Fabricated via Wire Arc Additive Manufacturing in As-Printed
and Heat-Treated Conditions�� 73
Elham Afshari, Mahya Ghaffari, Alireza Vahedi Nemani, and Ali Nasiri
High-Speed Nanoindentation Mapping of Additive Manufactured

Titanium Alloys for Aerospace Application �� 77
Zhiying Liu, Haoxiu Chen, and Yu Zou

Picosecond Laser Surface/Deep Patterning of Alumina Ceramics�� 83
Hamidreza Yazdani Sarvestani, Warren Amsellem, Zachary Katz, Ilhan Esmail,
Clement Beausoleil, Javad Gholipour Baradari, and Behnam Ashrafi

Hybrid Manufacturing of IN718�� 91
Sila Atabay, Priti Wanjara, Javad Gholipour Baradari, Josh Soost,
and Mathieu Brochu
Part II Deep Decarbonization Pathways for Pyrometallurgical Processes:

Opportunities & Challenges
Electrically Heated Fluid Bed as a Low-Carbon Option

for Pyrometallurgical Processes�� 101
Kamal Adham, S. Francey, and C. McIntyre
Contents xiii
Circored Fine Ore Direct Reduction Plus DRI Smelting – One Route
Towards Green Steel�� 109
Sebastian Lang, Max Köpf, and Timo Haimi
High-Temperature Hydrogen Reduction of Bauxite Residue

for Iron Recovery �� 119
Arman Hoseinpur, Per Ivar Friborg, Casper Van Der Eijk, and Jafar Safarian
Hydrogen the Silver Bullet for Decarbonization?�� 129

Is
Mansoor Barati
Electrical Resistivity of Charge Materials in the Si Process

and Its Effect on Energy and CO2 Emissions�� 133
Haley Hoover, Meggi Głowacki, Gudrun Sævarsdottir, and Merete Tangstad

Fossil-Free Carburization of Hydrogen-Based DRI �� 145
T. K. Sandeep Kumar, Hesham Ahmed, Bo Björkman, and Johanna Alatalo
Utilization of Carbon Fibers (CF) as an Alternative Reduction Agent

for Slag Valorization Processes �� 151
Ludwig Blenau, Manuela Wexler, Jonathan Mahl, Werner Baumann,
Dieter Stapf, and Alexandros Charitos
Deep Decarbonization and Pathways to Net Zero Emissions:

A Steel Industry Perspective �� 157
Ian Cameron

Utilization of Pretreated Mn-Ore in a Pilot-Scale Ferromanganese Furnace:
Effect of Ore Pretreatment on Carbon and Energy Consumption �� 161
Tichaona Mukono, Jonas E. Gjøvik, Heiko Gaertner, Trine Asklund Larssen,
Maria Wallin, Eli Ringdalen, and Merete Tangstad
Technology Developments and the Metals Industry

in a Low Carbon Economy�� 171
P. J. Mackey
Predictive Furnace Refractory Maintenance Procedures

to Extend Campaign Life and Reduce Waste�� 175
Afshin Sadri, Matthew Cramer, and Wai Li Winnie Ying

Green Hydrogen for Steel Production: A Case Study�� 193
Gareth Richardson, Peter Appleby, Maria De Campos, and Sina Mostaghel
Stabilizing Chromium in Ferrochromium Slag for Utilization

as Aggregate Material�� 203
Tahmeed Bin Tasnim and Leili Tafaghodi
xiv Contents
Iron Segregation Roasting as a Potential Decarbonization Pathway

for Ironmaking �� 207
P. Kerr, Q. Liu, and T. H. Etsell

Using Hydrogen as a Reductant in Fire Refining at Aurubis
Hamburg’s “Down-town” Smelter �� 211
Torben Edens and Johann Steindor
Part III Electrochemical Degradation of Multi-Component Materials

Corrosion of Mg Alloy AM60B Micro-Alloyed with As�� 231
Joey Kish, Dong Fu, Beth McNally, Darren Feenstra, Amjad Javaid,
and Kumar Sadayappan
Studying Stress Corrosion Cracking Mechanisms Using Novel

Nanoscale Characterization�� 237
S. Y. Persaud
On-Line ICP-MS in Electrocatalysis Research: Platinum

Dissolution Studies �� 245
Valentín Briega-Martos and Serhiy Cherevko
Part IV Energy and Environmental Materials
Adsorption of Naphthenic Acids from Oil Sand Process-Affected

Water (OSPW) Using Commercially Viable Petcoke-Sourced
Activated Carbon �� 251
Elmira Nazari, Tyler M. Roy, Oliver K. L. Strong, and Andrew J. Vreugdenhil
Crystallization of Nanoscale Garnet-Type Li-Ion Conductors

for All-Solid-State Batteries (ASSBs) �� 255
Senhao Wang, Hsien-chieh Chiu, and George P. Demopoulos

Resin Technologies to Improve Battery Metal Recovery�� 259
Jeffrey Sehn and Mike Weatherill
Study of Adsorption Thermodynamics of Model Naphthenic Acids

for Determining Adsorption Mechanism Using Commercially Viable
Petroleum Coke-Activated Carbon�� 265
Tyler M. Roy and Andrew J. Vreugdenhil
Enhanced Strength and Electrical Conductivity of the Al-Mg-Si

Conductor Alloys by Adding Ag and Applying the Preaging�� 269
Siamak Nikzad Khangholi, Mousa Javidani, Alexandre Maltais, and X. Grant Chen
Contents xv

Economic Analysis of Electrolytic Treatment of Waste Cyanide�� 279
Jacob Schmidt and Eduard Guerra
Part V Light Metals for the Transportation Industry
Titanium: From Ore Extraction and Processing to Its Applications

in the Transportation Industry �� 293
Carsten Siemers and Fabian Haase
HSBC Back-Meniscus Stability for a Double-Impingement Liquid

Metal Feeding System�� 305
D. R. Gonzalez-Morales, M. M. Isac, and R. I. L. Guthrie
Development of Fe-Based Metallic Glass Coating on Magnesium

Alloys by Cold Spray �� 315
G. Asala, O. T. Ola, and O. A. Ojo
pplications of Advanced Thermal Analysis for Aluminum Foundries �� 329

A
François Audet and Yohan Tremblay
Microstructure and Texture in Additively Manufactured Maraging

Steel Lattice Structures �� 335
Farzad Khodabakhshi and Mohsen Mohammadi

Si-Containing Titanium Alloys for Laser Powder Bed Fusion (PBF-L) �� 343
Fabian Haase, Carsten Siemers, Maximilian Goldapp, and Joachim Rösler
Aluminum Auxetic Structures Produced Using Combined

Stereolithography and Investment Casting �� 355
N. Alfano, H. Simha, and A. Elsayed

High-Strength, Low-Modulus Nanostructured: Ti 13Nb 13Zr Alloy�� 359
L. Klinge, C. Siemers, and C. Jahnke
A Novel Experimental Set-up for Generating Microbubbles

for Removal of Inclusions�� 369
A. Panicker, D. Lu, R. Tiwari, L. Calzado, M. M. Isac, and R. I. L. Guthrie

Asymmetrical Rolling of Aluminum Alloy 6061�� 383
Julie Lévesque and Alexandra Béland
Castability of Al-Fe-Ni Alloys�� 391

S. Kotiadis and A. Elsayed
xvi Contents
Maintenance Strategy for Reliable Instruments in Gas Treatment

Center (GTC) of Aluminum Smelter�� 395
Stanly Jeyaraj
Experimental and Numerical Study of 3003-H14 Aluminum

Alloy-Welded Joints with Different Clamping Conditions�� 405
M. Hernández, R. R. Ambriz, C. García, C. Gómora, and D. Jaramillo
Out-of-Plane Bending Assessment of Heat-Treated Aluminum

Alloys Subjected to Free-Fall Impact Loads �� 409
J. A. Frutos, E. R. Barragán, R. R. Ambriz, C. García, C. Gómora, and D. Jaramillo
A Study on Aluminum-Based Lightweight Entropic Alloys

with High Strength at Elevated Temperature�� 413
Liying Cui, Zhan Zhang, Dilip K. Sarkar, Duygu Kocaefe, and X. -Grant Chen
The Development of the HSBC (Horizontal Single-Belt Casting)

Process From Inception to Present Day �� 425
D. R. Gonzalez-Morales, M. M. Isac, and R. I. L. Guthrie
Scanning Electron Microscopy Analyses of Horizontal Single Belt

Casting Thin Strips of AA6111 Alloy �� 437
C. Riviere, D. Lu, M. M. Isac, and R. I. L. Guthrie

The Effect of Hydrogen Embrittlement on Aluminum Alloys Used
for Hydrogen-Fueled Internal Combustion Engine�� 447
Rashiga Walallawita, Joshua Stroh, and Dimitry Sediako
Development of LiMCA (Liquid Metal Cleanliness Analyzer)

Since Its Invention to Date�� 451
G. D. Di Silvestro, R. Tiwari, M. M. Isac, and R. I. L. Guthrie

Cr-Oxide: Why Should You Have It in Your Refractory?�� 463
Dean Gregurek, Philip W. Schantl, Jürgen Schmidl, and Alfred Spanring
Microstructural Study of Containerless Solidification

of Al-20wt%Ce Alloys�� 471
Hani Henein, Akankshya Sahoo, and Jonas Valloton
Processing, Properties, and Applications of Lightweight Al

and Mg Alloys�� 475
P. Emadi, B. Andilab, A. Belcastro, and C. Ravindran

Characterization of the Microstructure of Al, Mg and Ti Alloys
with State-of-the-Art Electron Microscopy �� 489
Raynald Gauvin, Nicolas Brodusch, and Stéphanie Bessette
Contents xvii
Self-Piercing Riveting of High-Ductility Al–Zn–Mg Casting Alloy

(Nemalloy HE700) in F Temper�� 493
Yunsong Guo, Sumanth Shankar, Mukesh K. Jain, Laura Baenen,
Anthony Lombardi, and Glenn Byczynski
Natural Ageing of Al–Fe(Zn, Mg) Alloy Castings for Structural

Automotive Application �� 503
Chimaobi Orji, Anthony Lombardi, Glenn Byczynski, Mohamed Hamed,
and Sumanth Shankar
Part VI Processing of Critical Materials
Integrating Carbon Capture in Mining Through Metallurgy. Part 1:

Leaching and Reclamation of Asbestos Tailings: Thetford Mines
Carbon Capture and Remediation Project�� 515
Mike Kelland, Greg Rau, Brock Battochio, Jason Vallis, Mykolas Gladkovas,
Sridevi Thomas, Kevin Bradley, Clive Brereton, Srinath Garg, and Alex Mezei
Developing a Commercial Heavy Rare Earth Processing Facility

at Browns Range�� 529
Louis de Klerk and Robin Jones
Integrating Carbon Capture in Mining Through Metallurgy. Part 2:

Integration into the Piaui Nickel–Cobalt Laterite Operations�� 549
Anne Oxley, Brock Battochio, Mike Oxley, Mike Kelland, Mykolas Gladkovas,
Sridevi Thomas, Kevin Bradley, Srinath Garg, Greg Rau, and Alex Mezei
Deep Eutectic Solvents for the Recovery of Metals: Electrodissolution

of Magnets for Rare Earth Recycling�� 561
Guillaume Zante, Shannon Stodd, Rodolfo Marin Rivera, Jennifer Hartley,
and Andrew P. Abbott

Heavy Rare Earth Ore Sorting at Browns Range�� 565
Louis de Klerk, Bill Rayson, Paul Beukes, and Angus Tod

Catalytic Dissolution of Metals Using Deep Eutectic Solvents�� 583
Rodolfo Marin Rivera, Phil Hunt, Guillaume Zante, Jennifer M. Hartley,
Karl S. Ryder, and Andrew P. Abbott
Neodymium(III) and Lanthanum(III) Separation by a Flat-Sheet

Supported Liquid Membrane System�� 591
Lin Li, Krystal Davis, Aaron King, Mauro Dal-Cin, Andrzej Nicalek,
and Ben Yu
xviii Contents
Process Development for Selective Recovery of Lithium from Black

Mass of Spent LiFePO4 Batteries�� 601
Tianyu Zhao, Harshit Mahandra, Rajashekhar Marthi, Michael Traversy,
Yeonuk Choi, and Ahmad Ghahreman

Dynamic Simulation of Uranium Ion Exchange Processes�� 607
Kevin Heppner

Enhanced Elution Efficiency of Scandium from Ion Exchange Resins�� 621
Mehdi Mostajeran and Rory Cameron
Processing Black Mass into a Mixed Nickel/Cobalt Hydroxide

Precipitate and Industrial-Grade Lithium Carbonate:
A Techno-Economic Analysis Perspective�� 627
Christopher P. Baxter, Andriy Plugatyr, and Ben Yu
Detailing of Mineral Processing Operations Within Strategic

Mine Planning Algorithm: A Quantitative Cornerstone
for Enabling Process Innovation�� 637
Aldo Quelopana, Javier Ordenes, and Alessandro Navarra
Potential Processes for Producing High-Purity Lithium Hydroxide:

A Critical Review �� 645
M. A. Halim, V. I. Lakshmanan, J. Chen, S. Kumar, S. DasGupta, and M. Dehn

The Revival of the French Rare Earth Industry �� 655
Alain Rollat
Cobalt–Nickel Separations Using Supported Liquid Membranes

for End-of-Life Lithium-Ion Battery Recycling�� 661
Mauro Dal-Cin, Krystal Davis, Aaron King, Lin Li, Andrzej Nicalek,
Gilles Robertson, and Ben Yu
On the Solvometallurgical Extraction of Lithium and Cobalt

from Secondary Resources�� 675
Georgios Kolliopoulos and Halimeh Askari Sabzkoohi
The Beneficiation and Metallurgical Process for the Production

of Mixed REE Product from Nechalacho Deposit�� 681
Kevin Bradley, Jing Liu, Ray Anguelov, and Tony Hadley
The Beneficiation and Hydrometallurgical Process Development

for the Lofdal HREE Project�� 699
Jing Liu, Niels Verbaan, Micheal Archer, Tassos Grammatikopoulos,
Barbara Mulcahy, and Rainer Ellmies
Contents xix
Hydrometallurgical Flowsheet Options for Treatment

of Wicheeda Lake Flotation Concentrate: Acid Bake Versus
Caustic Crack �� 721
Niels Verbaan, Tassos Grammatikopoulos, Mike Johnson, Jing Liu,
John Goode, Kris Raffle, and Craig Taylor

Toward Carbon-Negative Cement and Battery Materials �� 737
Jeremy Ley, David Dreisinger, Mike Johnson, Niels Verbaan,
and Sridevi Thomas

The Demonstration of the Cuprion Process for Recovery of Nickel, Cobalt,
Copper, and Manganese from Seabed Nodules�� 753
David Dreisinger, Niels Verbaan, and Marlon Canizares
Magnet Rare Earth Production Must Double in Ten Years.

Where Will Production Come From?�� 773
J. R. Goode
Extraction of Nickel, Cobalt, and Scandium from Low-Grade

Philippine Laterite Ores Using Atmospheric Leaching
with Reduction Pre-Treatment �� 777
Edith Andrea D. Lucy, Geleni Shalaine P. Bello, Chris G. Arnejo,
Christine P. Naira, Ven Jordan Diputado, and Juancho Pablo S. Calvez
Innovative Solvent Extraction to Produce High-Purity Cobalt

and Nickel Sulphates from Sulphide Concentrate�� 787
M. A. Halim, S. Kumar, J. Chen, V. I. Lakshmanan, S. DasGupta,
and J. F. NdoutouMve
The Extraction of Nickel and Cobalt from Laterite Ores

with Concurrent Carbon Sequestration�� 797
Fei Wang and David Dreisinger

II-VI’s Capabilities in Scandium Recovery and Purification�� 811
Gomer Abrenica, Ghazaleh Nazari, and Shailesh Patkar
Part VII Towards Sustainable Circularity: Mining to Materials
Selective Depression of Talc in Cuprite Sulfidation Flotation

by New Depressant Astragaloside�� 823
Qiuyue Sheng, Wanzhong Yin, Yahui Zhang, and Kelly Hawboldt
Alternative Processing for Silver Extraction from High

Manganese-Silver Ores Using Leaching with SO2 Pretreatment�� 837
Ximena Nevárez, Daniel Manqueros, Flor Salazar, Fisher Wang,
and Persio Rosario
xx Contents

Imaging the Remaining Refractory Lining in Operating Furnaces�� 847
Rachel Santini, Afshin Sadri, and Wai Lai (Winnie) Ying
Dynamic Renewable Energy-Driven Framework Development

for Mineral Processing Circuits�� 859
Ryan Wilson, Andy Reynolds, and Alessandro Navarra

The Role of Cold Bonding in Sustainable Steelmaking�� 869
R. Elliott, B. MacKinnon, and I. Cameron

Canada as the Trailblazers in Net-Zero Carbon Mining �� 875
Susanna Zhang, Hakan Mustafa Tunc, and Ben Steyn
Development of a Smart Database for Wastes from the Canadian

Mining and Metallurgical Industry�� 879
Elhocine Derardja, Arash Rafiei, and Elmira Moosavi-Khoonsari
Decanter Centrifuges and Dewatering Technology Selection:

A Case Study�� 887
Thomas Boundy, Juan Alberto Juarez Garcia, Moises Uriel Raya Sotelo,
and Kathy Adams

Slag Resource Recovery Via Air Granulation �� 897
Santiago Faucher, Daniel Park, and Amit Shrivastava
Using Bio-Based Surfactants as Frothers in Froth Flotation

to Improve Renewable Carbon Index�� 907
Laura Benavides

Microwave Impacts on More Sustainable Kimberlite Processing�� 911
Muhammad A. Rasyid, Azlan Aslam, Arash Rafiei, Agus P. Sasmito,
and Ferri Hassani

Decarbonization of Mining and Metals Industry. A Critical Overview�� 917
Ashok D. Dalvi
Decarbonization at the Anglo-American Barro Alto Smelter

Through Implementation of the Ecombustible Technology �� 939
Daniel Brosig, Paykan Safe, Matt Russell, Jorge Arevalo, Carlos Kiyan,
Mario Rodrigues, and Marcelo Miranda
Ni/Co–Cu Sulphide Ores, Key Contributors to Sustainability:

A Review of Applications and Processing Challenges�� 949
S. Kelebek, S. Arslangil, and C. E. Gibson
Contents xxi
Water Recycling and Seasonal Water Quality Effects

in Mineral Processing�� 967
Patrick W. Rankin, Sadan Kelebek, Antonio Di Feo, and Jennifer R. Taylor
Production of Spodumene Concentrate from the North Carolina

Piedmont Lithium Project�� 979
Brian Kawenski Cook, Massoud Aghamirian, Hao Li, Nima Nabiri,
Jim Nottingham, Jarrett Quinn, Chris Gunning, Charlotte E. Gibson,
and Patrick Brindle
Using Polymetallic Nodules to Innovate Manganese Markets:

A New Manganese Feed for SiMn Alloy Production�� 993
M. N. Boulby, V. Canaguier, and J. R. Donald
Index�� 1007
Part I Advances in Materials Manufacturing
VI – Existing and Emerging Materials
Al Alloys and Manufacturing Processes
for Lightweight Applications in Electric Vehicles
Anita Hu1, Henry Hu1 (*), Yuxian Li1, Sufeng Liu1, and Wutian Shen1
1
Department of Mechanical, Automotive and Materials Engineering, University of Windsor,
Windsor, ON, Canada
huh@uwindsor.ca
Abstract. Due to increasing environmental concerns, battery-powered electric vehi-

cles (BEV) have gained popularity in the automotive for the past few years. An induc-
tion motor is an essential component of the propulsion system in integrated BEVs
working on different operating conditions [1]. Since a rotor of the induction motor is
configured with an electrical sheet, a rotor bar, and end-ring (Fig. 1), which are usu-
ally manufactured by high-pressure die casting (HPDC) processes. Figure 2a pres-
ents an illustration of a cold-chamber high-pressure die casting machine. The process
starts with ladling the liquid metal into the shot sleeve for each cycle. Die cavity and
plunger tip normally are sprayed with an oil or lubricant to provide protection from
severe degradation. This increases the die material’s life and reduces the adhesion of
the solidified component. The cycle using the cold-chamber high-pressure die casting
process as a model is shown in Fig. 2b. In a cycle, liquid metal is ladled into an injec-
tion system (i), which is then immediately pushed (ii) through a sleeve and runner
system (iii) into a die cavity at a very high speed (iv), and a high pressure is applied
for intensification. High pressures are maintained on the alloy during solidification.
After complete solidification, the die opens (v) and finally the component is ejected
(vi) [2]. To obtain the high efficiency of an induction motor, a heavy transition metal,
copper, having a high electrical conductivity and low resistivity is often used [3]. For
weight reduction of the BEVs with extended range, however, aluminum with a den-
sity of 2700 kg/m3 and a melting temperature of 660 °C as a lightweight metal is
strongly advocated by the automotive industry to replace heavy Cu. This is because
Al is over two times lighter than copper. Its low melting point compared with that of
Cu makes Al and its alloys easy to be manufactured for components in the induction
motor via casting processes such as high-pressure die casting [3].
Keywords: Al alloys · Manufacturing processes · Electrical conductivity · Electric

vehicle · Tensile properties · Strengthening mechanisms.
Due to increasing environmental concerns, battery-powered electric vehicles (BEV)

have gained popularity in the automotive for the past few years. An induction motor is
an essential component of the propulsion system in integrated BEVs working on differ-
ent operating conditions [1]. Since a rotor of the induction motor is configured with an
electrical sheet, a rotor bar, and end-ring (Fig. 1), which are usually manufactured by
high-pressure die casting (HPDC) processes. Figure 2a presents an illustration of a
cold-chamber high-pressure die casting machine. The process starts with ladling the
liquid metal into the shot sleeve for each cycle. Die cavity and plunger tip normally are
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 3

Proceedings of the 61st Conference of Metallurgists, COM 2022,
https://doi.org/10.1007/978-3-031-17425-4_1
4 A. Hu et al.
Fig. 1. Components of electrical sheets, rotor bars, and an end-ring in an induction motor [1].
Stationary Platen
Ejector Platen
Ladle Die Cavity (i) (ii)
(iii) (iv)
Hydraulic Shot Sleeve

Cylinder Ejector
Plunger
Ejector Box (v) (vi)
Cover Die
Die
(a) (b)
Fig. 2. Illustrations of (a) a cold-chamber high-pressure die casting machine and (b) casting cycle for
cold chamber die casting [2].
sprayed with an oil or lubricant to provide protection from severe degradation. This
increases the die material’s life and reduces the adhesion of the solidified component.
The cycle using the cold-chamber high-pressure die casting process as a model is shown
in Fig. 2b. In a cycle, liquid metal is ladled into an injection system (i), which is then
immediately pushed (ii) through a sleeve and runner system (iii) into a die cavity at a
very high speed (iv), and a high pressure is applied for intensification. High pressures
are maintained on the alloy during solidification. After complete solidification, the die
opens (v) and finally the component is ejected (vi) [2]. To obtain the high efficiency of
an induction motor, a heavy transition metal, copper, having a high electrical conductiv-
ity and low resistivity is often used [3]. For weight reduction of the BEVs with extended
range, however, aluminum with a density of 2700 kg/m3 and a melting temperature of
Al Alloys and Manufacturing Processes for Lightweight Applications in Electric Vehicles 5
660 °C as a lightweight metal is strongly advocated by the automotive industry to

replace heavy Cu. This is because Al is over two times lighter than copper. Its low melt-
ing point compared with that of Cu makes Al and its alloys easy to be manufactured for
components in the induction motor via casting processes such as high-pressure die cast-
ing [3].
Pure aluminum having high electrical conductivity possesses very low mechanical
strengths, and its yield strength is about 10–20 MPa depending on purity. In electrical
vehicles, materials for the motor require both high electrical conductivity and high
mechanical strength. In general, pure metals can be strengthened by the introduction of
obstacles to the motion of dislocations. Contributions from different strengthening
mechanisms are typically taken to be additive, assuming that they act independently,
and the total strength of alloys might be estimated by Eq. 1 [4, 5].
S So Sgs Sss Sdis Sp , (1)
Where So is Peierls stress, Sgs is grain boundary strengthening, Sss is solid solution
strengthening, Sdis is dislocation strengthening, and Sp represents precipitation strength-
ening. But, all these strengthening contributions lead to a degradation of the electrical
conductivity, as conductive electrons are scattered by solute atoms and crystal defects.
Each scattering event causes an electron to lose kinetic energy and to change its direc-
tion, thus increasing electrical resistivity, or decreasing electrical conductivities. The
properties of coarse-grained pure Al and commercial wrought Al-Mg-Si wires used in
overhead power lines are summarized in Fig. 3. It is clearly seen that, in all cases, alloys
exhibit a higher mechanical strength compared to pure metals but at the expense of the
electrical conductivity.
Fig. 3. Room temperature electrical conductivity (in % of IACS) versus ultimate tensile strength for
coarse-grained (CG) pure Al and commercial wrought Al-Mg-Si alloys as well as their nanostructured
counterparts obtained via severe plastic deformation (SPD) processes including high-pressure torsion
(HPT) and equal-channel angular pressing (ECAP) [6].
6 A. Hu et al.
Although mechanical deformation processing techniques for wrought Al alloys can be

used to produce nanostructured Al alloys with high electrical conductivities, they are only
suitable for the fabrication of components with simple geometry and high lost. Since
electrical motor components have complicated shapes and even undercut casting pro-
cesses including high-pressure die casting and sand casting should be considered the
most cost-effective route for the massive production of parts in the BEVs [7, 8]. To ensure
the availability of the casting processes for manufacturing electrical motor components,
castable Al alloys with high electrical conductivities plus strengths are necessary.
Recently, Sivanesh et al. [9] invented an aluminum alloy exhibiting high yield
strengths and high conductivities, which was claimed to be high-pressure die castable
for the production of electric car parts. This novel alloy, named Tesla alloy, contains
about 5 wt% Ni, 0.35 wt% Fe, 0.03 wt% Ti, and the remaining wt% being Al and inci-
dental impurities. The high Ni content increases the cost of Al alloys significantly. The
developed HPDC alloy had a relatively low yield strength of about 90 MPa and electri-
cal conductivity of about 48% IACS. Figure 4 compares the electrical conductivities
and yield strengths between various cast and wrought Al alloys under the as-cast and
heat treatment conditions. Based on the yield strength versus conductivity plot, an alloy
Alloy 100.1 (99% AI) [Lit.]

Alloy 130.1-F (99.3% AI) [Lit.]
Alloy 150.1-F (99.5% AI) [Lit.]
Alloy 170.1-F (99.7% AI) [Ramcast]
A319-F [Nemak]
A319-T7 (WQ) [Nemak]
356-T6 (WQ) [Nemak]
A356-T6 (AQ) [Nemak]
A356-T6 (WQ) [Nemak]
A356 + 0.5 Cu - T6 (AQ) [Nemak]
A356 + 0.5 Cu - T6 (WQ) [Nemak]
Castasil 21-F [Rheinfelden]
Castasil 21-T5 [Rheinfelden]
300 Cu 10100-0
6101-T63 [Wrought]
250
Yield Strength (MPa)
Traditional
200 Cast Alloys
150
Alloy Design Space
100 48-55% IACS
90 MPa min. Yield
50 High Conductivity
Cast Alloys
0
20 30 40 50 60 70 80 90 100 110
% IACS
Fig. 4. A yield strength versus electrical conductivity plot for cast aluminum alloys [9].
Al Alloys and Manufacturing Processes for Lightweight Applications in Electric Vehicles 7
design space was proposed by Sivanesh et al. [9]. Cast aluminum alloys were divided
into two groups. One group had high yield strengths but low electrical conductivities,
and the other had high electrical conductivities but low yield strengths. The alloys
within the alloy design space as shown in Fig. 4 possessed an electrical conductivity of
about 48% to 55% IACS, and the range of yield strengths between 90 MPa and 130 MPa.
Aluminum alloy A319, as a representative traditional cast alloy, under the as-cast con-
dition had only a low yield strength of 120 MPa and a low electrical conductivity of
26% IACS. However, both the yield strengths and electrical conductivities of the tradi-
tional cast Al alloys were improved by the proper procedure of heat treatment. T7 treat-
ment increased the yield strength to about 270 MPa, but little in electrical conductivity.
Figure 5 presents the results of the castability study on the Tesla alloys in comparison
with the traditional cast Al alloys. The Tesla alloy had enhanced fluidity and hot tearing
susceptibility, which demonstrated good castability.
Figure 6 illustrates the shrinkages of cast Al alloys. It is desirable to design an alloy
with as little shrinkage from liquidus to solidus as possible. As can be observed, the
1000 24
800 20
Flow Length (mm)
Hot Tearing Index
16
600
12
400
8
200 4
0 0
(a) Pure AI AI-1Si 365-3 Magsimal59 (b) Pure AI AI-1Si 365-3 Magsimal59
AI-3.5Si AI-5.3 Ni Castaduct AI-3.5Si AI-5.3 Ni Castaduct
Fig. 5. Castability of cast pure aluminum and Al alloys, (a) fluidity (b) hot-tearing susceptibility [9].
9
8.18
8 7.74
6.93 7.08
7
6 5.54
Shrinkage (%)
4.89
5
0
Pure AI AI-3.5Si-0.5Mg AI-5.4Ni
AI-1Si-0.5Mg A390 A206
Fig. 6 Shrinkages of cast pure aluminum and Al alloys [9].

8 A. Hu et al.
Tesla alloy with 5.4 wt% expensive nickel exhibited only 5.54% shrinkage, which out-
performed most of the tested alloys. But, the Tesla alloy had a shrinkage of 5.54%
higher than that (4.89%) of the common HPDC Al alloy A390. Minimizing shrinkage
is important to keep parts within tolerances during production. It is impossible to make
a comparison of castabilities among the listed cast Al alloys since there is inadequate
information about their castabilities in the open literature. To massively implement
lightweight materials in the BEVs, low-cost castable Al alloys with both high strengths
and electrical conductivities need to be further developed.
Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research
Council of Canada, and the University of Windsor for supporting this work.
References
1. Boukhnifer M, Raisemche A. Fault tolerant control for induction motor in electrical vehicle. In:
2012 IEEE international conference on control applications. IEEE; 2012, October. p. 136–41.
2. Vinarcik EJ. High integrity die casting process. USA: Wiley; 2003.
3. Popovich A, Sufiiarov V, Polozov I, Borisov E, Masaylo D, Orlov A. Microstructure and mechani-
cal properties of additive manufactured copper alloy. Mater Lett. 2016;179:38–41.
4. Kamikawa N, Huang X, Tsuji N, Hansen N. Strengthening mechanisms in nanostructured high-
purity aluminium deformed to high strain and annealed. Acta Mater. 2009;57:4198–208.
5. Asgharzadeh H, Simchi A, Kim HS. Microstructural features, texture and strengthening mech-
anisms of nanostructured AA6063 alloy processed by powder metallurgy. Mater Sci Eng
A. 2011;528:3981–9.
6. Murashkin MY, Sabirov I, Sauvage X, Valiev RZ. Nanostructured Al and Cu alloys with superior
strength and electrical conductivity. J Mater Sci. 2016;51(1):33–49.
7. Ishikawa H, Takashima Y, Okada Y. Squirrel-Cage motor Rotor and Squirrel-Cage Motor.
U.S. Patent No. 9,935,533, 4 March 2018.
8. Xia S, Sanjiu Y, Xiao B, Sirui L. Effects of Si, Cu and Mg addition on electrical conductivity and
mechanical properties of cast aluminum alloys. Spec Cast Nonferrous Alloys. 2010;12:12–20.
9. Sivanesh P, Charlie K, Robert SJ, Ethan F, Paul E. Aluminum alloys for die casting. U.S. Patent No.
WO 2020/028730 A1, 6 Feb 2020.
Laser Powder Bed Fusion of M789 Steel on Wrought
N709 Steel Substrate
Kudakwashe Nyamuchiwa1 (*) , Yuan Tian2, Kanwal Chadha1 ,

and Clodualdo Aranas Jr1
1
Department of Mechanical Engineering, University of New Brunswick, Fredericton, NB,
Canada
kuda.nyamuchiwa@unb.ca
2
voestalpine Additive Manufacturing Centre Ltd, Mississauga, ON, Canada
Abstract. A newly developed maraging steel, M789, was additively manufactured on

top of a wrought precipitation hardening steel, N709, using laser powder bed fusion
(LPBF). A fully dense M789 was achieved. The EBSD confirmed a completely martens-
itic structure with less than 1% retained austenite. The results show that the M789 steel
had a lower hardness of 550 HV compared to 680 HV for N709. Moreover, no significant
diffusion of alloying elements from the M789 to the N709 material was detected. Based
on the initial assessment, a post-processing heat treatment is necessary to achieve homog-
enous mechanical properties between dissimilar alloys.
Keywords: Laser Powder Bed Fusion · Interface · Aging
1 Introduction
The use of metal additive manufacturing has changed the approach to designing metal-
lic components. With the growing demand for more efficient and green technologies,
the usefulness of additively manufactured parts is gradually increasing in the aerospace
and energy industries [1]. Out of the seven currently existing additive manufacturing
(AM) techniques, the laser powder bed fusion (LPBF) has emerged as a head candidate
for manufacturing specialized components such as cabin brackets and impeller internal
lattices because of its ability to attain densities close to unity, final smooth finishes and
complex geometries [1–3]. This technique utilizes metal powders thinly spread on a
substrate; then, a laser beam selectively melts the powder layer based on a CAD model.
Subsequent layers are added, and the process is repeated until the final product is com-
plete. Despite its merits, LPBF processes in comparison to other metal additive tech-
niques are relatively expensive and are best suited for high-end use components such as
those used in the aerospace industry. Furthermore, the build size of LPBF components
is currently limited to between 300 and 400 mm [1]. System productivity rates of con-
ventional processes are estimated to exceed 100 kg/h, whilst that for LPBF is lower
than 70 g/h [4]. As such, it cannot be incorporated into mass-production operations
because of the inevitable overhead costs. In addition to the slow-building rates, the

https://doi.org/10.1007/978-3-031-17425-4_2
10 K. Nyamuchiwa et al.
incorporation of different metals in LPBF printing equipment spurs the total expendi-
ture for single operations because printing parameters are fixed and optimized for sin-
gle alloys. Printing parameters usually vary for each alloy system, and it usually takes
multiple trial and error operations to develop optimized parameters. This makes this
technology fail to meet the tooling and aerospace industries’ needs that often require
the use of different metals in combination to meet specific functional requirements.
These factors have slowed the adoption of this technology combined with currently
limited means to expand its usefulness and full potential. The usefulness of LPBF can
be expanded by combining traditional manufacturing techniques and additive manufac-
turing, known as hybrid additive manufacturing [5]. Contrary to monolithic alloy pro-
duction, hybrid additive manufacturing offers cost reduction options by combining the
low cost of the traditional cast and wrought alloys with additive manufacturing to attain
components with enhanced functionality [6]. However, differences in properties
between the dissimilar alloys can cause heterogeneous structures and anisotropic
mechanical properties [7]. In this work, M789 steel was printed on N709 wrought alloy,
and the phase and mechanical properties were analysed to assess the feasibility of this
alloy combination to form a hybrid material.
2 Materials and Methods
The M789 steel was deposited on a wrought precipitation hardening N709 alloy using
an EOS M290 LPBF machine at the voestalpine Additive Manufacturing Centre in
Mississauga, Ontario, Canada. The powder size diameters used were in the range of 15
to 60 μm with a flowability of 15.84 s/50 g as measured by the Hall Flowmeter. The
printing was conducted in an inert atmosphere of argon using a volumetric laser energy
density between 75 and 85 J/mm3. The laser power used was in the range of 250 to
270 W with a scanning speed held between 1050 and 1200 mm/s. A hatch distance was
kept between 0.10 and 0.11 mm with a powder layer thickness of approximately 40 μm.
A fully dense M789 material was achieved. Table 1 displays the composition of the
alloys.
An electron backscattering diffraction (EBSD) analysis was done on selected and
polished as-printed samples to determine the phase distribution. Then, electron diffrac-
tion spectroscopy (EDS) was performed across the interface using a JSM – 6010LA
analytical SEM. Moreover, the hardness values of the as-printed M789-wrought N709
hybrid material were obtained using Nanomechanics iMicro nanoindentation machine.
Table 1. The composition of M789 powder and N709 steel (in wt%).
C Cr Ni Mo Al Ti Fe
M789 <0.02 12.20 10.00 1.00 0.60 1.00 Balance
N709 0.03 12.70 8.10 2.20 1.10 – Balance
Laser Powder Bed Fusion of M789 Steel on Wrought N709 Steel Substrate 11
Fig. 1. Phase map for M789-N709 alloy.
3 Results and Discussion
3.1 EBSD Phase Map

A representative microstructure of the hybrid material (see Fig. 1) describes the phase
distribution after manufacturing. From left to right of the image, the EBSD phase map
shows the M789, M789-N709 interface and the N709 steel sections. The structure is
approximately 99.1% body-centred tetragonal (BCT), as denoted by red area, with a
sparsely distributed retained austenite (RA) amounting to about 0.9%. For maraging
steels, it is well known that the predominant structure is martensite, and the amount of
retained austenite (RA) is dependent on the alloy composition and processing condi-
tions [8]. The N709 has equiaxed grains with sizes of approximately 20 microns.
Towards the interface, the microstructure appears more refined, which can be attributed
to the fast cooling rates (106–108 K/s) experienced during LPBF [9]. In addition, the
initial layers experience a faster cooling rate than the succeeding layers, as evidenced
by a change in the grain morphology from fine to elongated grains [10]. The interface
spans almost 50 microns, after which epitaxial growth of elongated grains is formed. As
expected, the grain orientation is along the direction of the highest heat gradient.
3.2 Microhardness
In the as-printed state, the hardness values of the M789 and N709 are 540 HV and
680 HV, respectively (see Fig. 2). These high hardness values can be attributed to the
high residual stresses and dislocation density associated with additive manufacturing
processes [9]. It was deduced that the N709 steel section is 24% harder than the M789
section because of the hardening pre-treatment of the N709 substrate before printing.
This difference in properties can be minimized by performing heat treatment to achieve
homogeneity. Based on the previous work [11, 12], the hardness of M789 steel can be
increased to more than 540 HV by employing solution treatment at 1000 °C for 1 hour,
followed by aging at 500 °C for 2 hours (see Fig. 3). At these conditions, it was discov-
ered that LPBF-M789 precipitates an ETA-Ni3(Ti,Al), also known as η-phase, with a
volume fraction of approximately 5% [11–13]. In other works [7, 14] solution treat-
ment has been reported to increase the amount of retained austenite (temperatures
above 900) thus reducing the mechanical properties. In addition, large martensite lathes
have been reported with reduced mechanical properties [15]. Alternatively, the solution
treatment step can be skipped and the samples can be directly aged [6]. However, there
are possibilities of alloy element microsegregation especially for elements such as Mo
750
700
650
Vicker hardness (HVN)

600
550
500
450
400 N709 M789
350
300
250
200
150
100
50
0
Fig. 2. Microhardness at the M789 and N709 sections, respectively.
Fig. 3. Heat treatment schedule for precipitation strengthening of M789 [11.]
and Ti because of the low temperatures that restrict their diffusion into grain interiors
[16].
3.3 EDS Point and Map Analysis

An EDS point analysis was carried out across the interface, as shown in Fig. 4. The
presence of Ti was detected after point 003 (see N709 section). The chemical composi-
tions of M789 and N709 differ in the titanium composition, with the latter having a total
of 1.00 wt.% Ti (Table 1). A map analysis was carried out (see Fig. 5) on the sample and
a concentration profile for titanium was established. A concentration change from
0 wt.% Ti in the N709 and 0.98 wt.% in the M789 section was observed. The diffusion
distance was estimated to be less than 200 μm. Furthermore, a distinct interface was
observed and from these observations, titanium seems to have not diffused further into
the N709 side. However, this is not conclusive, and a more detailed characterization
analysis is required.
Laser Powder Bed Fusion of M789 Steel on Wrought N709 Steel Substrate 13
Fig. 4. EDS point mapping across the M789–N709 interface.
Fig. 5. EDS mapping across the M789–N709 interface for Ti.
4 Conclusion
M789 steel was successfully printed on top of wrought precipitation hardening N709
alloy, and the phases had less than 1% retained austenite. The M789 section had a low
hardness of 540 HV compared to 680 HV for N709 before any heat treatment. It was
observed that no significant diffusion of titanium from M789 to N709 was detected.
Heat treatment of the specimen can potentially improve the homogeneity of the
mechanical properties based on the previous work [11, 12].
Acknowledgements The authors acknowledge funding received from the Natural Sciences and
Engineering Research Council of Canada, Canada Foundation for Innovation, New Brunswick
Innovation Foundation, and the McCain Foundation.
References
1. Azizi H, Ghiaasiaan R, Prager R, Ghoncheh MH, Samk KA, Lausic A, Byleveld W, Phillion
AB. Metallurgical and mechanical assessment of hybrid additively-manufactured maraging
tool steels via selective laser melting. Addit Manuf. 2019;27:389–97. https://doi.org/10.1016/j.
addma.2019.03.025.
2. Bai Y, Wang D, Yang Y, Wang H. Effect of heat treatment on the microstructure and mechanical
properties of maraging steel by selective laser melting. Mater Sci Eng A. 2019;760:105–17. https://
doi.org/10.1016/j.msea.2019.05.115.
3. Blakey-Milner B. Metal additive manufacturing in aerospace: a review. 2021;33. https://doi.
org/10.1016/j.matdes.2021.110008.
4. Conde FF, Escobar JD, Oliveira JP, Béreš M, Jardini AL, Bose WW, Avila JA. Effect of ther-
mal cycling and aging stages on the microstructure and bending strength of a selective laser
melted 300-grade maraging steel. Mater Sci Eng A. 2019;758:192–201. https://doi.org/10.1016/j.
msea.2019.03.129.
5. DebRoy T, Wei HL, Zuback JS, Mukherjee T, Elmer JW, Milewski JO, Beese AM, Wilson-Heid A,
De A, Zhang W. Additive manufacturing of metallic components – process, structure and proper-
ties. Prog Mater Sci. 2018;92:112–224. https://doi.org/10.1016/j.pmatsci.2017.10.001.
6. Gutowski TG, Branham MS, Dahmus JB, Jones AJ, Thiriez A, Sekulic DP. Thermodynamic
analysis of resources used in manufacturing processes. Environ Sci Technol. 2009;43(5):1584–90.
https://doi.org/10.1021/es8016655.
7. LeBrun T, Nakamoto T, Horikawa K, Kobayashi H. Effect of retained austenite on subsequent
thermal processing and resultant mechanical properties of selective laser melted 17–4 PH stainless
steel. Mater Des. 2015;81:44–53. https://doi.org/10.1016/j.matdes.2015.05.026.
8. Murthy CVS, Krishna AG, Reddy GM. Dissimilar welding of maraging steel (250) and
13-8 Mo stainless steel by GTCAW, LBW and EBW processes. Trans Indian Inst Metals.
2019;72(9):2433–41. https://doi.org/10.1007/s12666-019-01695-z.
9. Oliveira JP, LaLonde AD, Ma J. Processing parameters in laser powder bed fusion metal additive
manufacturing. Mater Des. 2020;193:108762. https://doi.org/10.1016/j.matdes.2020.108762.
10. Palad R, Tian Y, Chadha K, Rodrigues S, Aranas C. Microstructural features of novel corrosion-
resistant maraging steel manufactured by laser powder bed fusion. Mater Lett. 2020;275:128026.
https://doi.org/10.1016/j.matlet.2020.128026.
11. Rafiee M, Farahani RD, Therriault D. Multi-material 3D and 4D printing: a survey. Adv Sci.
2020;7(12):1902307. https://doi.org/10.1002/advs.201902307.
12. Shakerin S, Hadadzadeh A, Amirkhiz BS, Shamsdini S, Li J, Mohammadi M. Additive manu-
facturing of maraging steel-H13 bimetals using laser powder bed fusion technique. Addit Manuf.
2019;29:100797. https://doi.org/10.1016/j.addma.2019.100797.
13. Shakerin S, Sanjari M, Amirkhiz BS, Mohammadi M. Interface engineering of additively manu-
factured maraging steel-H13 bimetallic structures. Mater Charact. 2020;170:110728. https://doi.
org/10.1016/j.matchar.2020.110728.
14. Tian Y, Palad R, Aranas C. Microstructural evolution and mechanical properties of a newly
designed steel fabricated by laser powder bed fusion. Addit Manuf. 2020;36:101495. https://doi.
org/10.1016/j.addma.2020.101495.
15. Tian Y, Palad R, Jiang L, Dorin T, Chadha K, Aranas C. The effect of heat treatments on
mechanical properties of M789 steel fabricated by laser powder bed fusion. J Alloys Compd.
2021;885:161033. https://doi.org/10.1016/j.jallcom.2021.161033.
16. Turnier Trottier W, Kreitcberg A, Brailovski V. Structure and mechanical properties of laser pow-
der bed-fused and wrought PH13-8Mo-type precipitation hardening stainless steels: comparative
study. J Manufact Mater Process. 2021;5(3):67. https://doi.org/10.3390/jmmp5030067.
Monotonic and Cyclic Deformation Behavior
of a Silafont®-36 Cast Aluminum Alloy
in an Overaged Condition
S. S. Dash1(*) , D. J. Li2, X. Q. Zeng2, D. Y. Li3, and D. L. Chen1

1
Department of Mechanical and Industrial Engineering, Toronto Metropolitan University
(formerly Ryerson University), Toronto, ON, Canada
soumya.sobhan.dash@ryerson.ca
2
State Key Laboratory of Metal Matrix Composites, School of Materials Science and
Engineering, Shanghai Jiao Tong University, Shanghai, China
3
Department of Chemical and Materials Engineering, University of Alberta,
Edmonton, AB, Canada
Abstract. The aim of this study is to explore and understand the deformation mecha-
nism in an overaged Al-Si high-pressure die cast alloy. Uniaxial tensile and strain-con-
trolled low-cycle fatigue properties were correlated with the microstructure obtained after
solutionizing and overaging. Nodularization of coralloid, eutectic Si particles occurred
during solutionizing, which were uniformly distributed in the primary Al matrix. As a
result, the alloy in the T7 overaging state showed higher ductility at the expense of the
yield and ultimate tensile strength. The alloy also exhibited largely stable cyclic deforma-
tion characteristics without an obvious cyclic hardening/softening effect.
Keywords: Al-Si alloy · Overaging · Tensile behavior · Cyclic deformation.
1 Introduction
Lightweight Silafont®-36 alloy containing a low amount of Fe has been developed for
automotive applications due to its relatively high tensile strength, long fatigue life, and
superior ductility in the as-cast condition. The mechanical properties of this alloy are
expected to vary with its tempering condition. T7 heat treatment involving a longer
aging time after solution heat treatment (overaging) is one of the commonly used tem-
pering conditions to achieve stabilized aluminum alloys. This treatment helps decrease
the susceptibility to stress corrosion cracking and increase the exfoliation resistance
and thermal conductivity of some Al-Si alloys, despite the accompanied moderate
strength [1–3]. The load-bearing structural application of such alloys inevitably involves
strain-controlled cyclic deformation behavior. Although the information on some ten-
sile properties of the alloy experiencing T7 treatment has been reported, no studies on
its low-cycle fatigue behavior could be found in the open literature.

https://doi.org/10.1007/978-3-031-17425-4_3
16 S. S. Dash et al.
The objective of this study is to evaluate the cyclic deformation behavior of a high-
pressure die-cast (HPDC) Silafont®-36 alloy in the T7 state in relation to its microstruc-
tures and clarify underlying deformation mechanisms.
2 Experimental Methods
In this work, Silafont®-36 alloy ingots authorized by Rheinfelden GmbH were melted
and cast using vacuum-assisted HPDC system. The alloy after casting had a chemical
composition (in wt.%) of 10.34 Si, 0.62 Mn, 0.32 Mg, 0.091 Fe, determined via plasma
mass spectrometry. The cast ASTM E606 standard specimens were subjected to heat
treatment under the T7 temper condition (solutionizing at 500 °C for 2 hours, water
quenching, and then aging at 225 °C for 8 hours). Prior to the solution and aging heat
treatment, both furnaces were calibrated via a bundle of four K-type thermocouples. The
microstructural features were examined under SEM with assisted EBSD detectors.
The standard specimens with square cross-section of 6 mm × 6 mm and a parallel
length of 32 mm were subjected to uniaxial tensile tests in a United testing system at room
temperature. The tests were conducted at two different strain rates, 1 × 10−4 and 1 × 10−2 s−1.
Strain-controlled fatigue tests were run on an Instron 8801 servo-hydraulic testing system
at room temperature, with fully reversed cycling and with triangular waveform at a con-
stant strain rate of 1 × 10−2 s−1. The recorded stress–strain results were analyzed.
3 Results
3.1 Microstructure
After T7 heat treatment, the microstructural features of Silafont®-36 alloy varied sig-
nificantly from its as-cast counterpart [4], which showed nodularized Si particles dis-
tributed over the primary Al matrix (Fig. 1a). An EBSD map of the alloy surface was
collected to analyze the phase distribution for Al and Si (Fig. 1b). A very minimal frac-
tion of Mg2Si precipitates was detected by the EBSD mapping.
Fig. 1. (a) SEM micrograph of Silafont®-36 alloy in the T7 heat treatment condition at a magnification
of 3000× and (b) EBSD phase distribution map of the alloy at 1500×.
Monotonic and Cyclic Deformation Behavior of a Silafont®-36 Cast Aluminum… 17
3.2 Tensile Properties
The tensile properties of the alloy in the T7 condition are listed in Table 1, along with
several other cast Al-Si alloys from the literature. At least three samples were tested at
two different strain rates in the present study. Since no strain-rate sensitivity of the ten-
sile properties was observed, the average value of all test samples was considered here.
It can be seen that the alloy in the present work has a lower average yield strength
(YS = 130 ± 9 MPa) and ultimate tensile strength (UTS = 174 ± 21 MPa) than other
Al-Si alloys, such as A356 + 0.5%Cu and AA6060 alloys.
The percent elongation (%EL), however, showed improvement for the alloy in the T7
condition (7 ± 3%), compared with that in the as-cast condition (6 ± 2%). This elonga-
tion was associated with the overall softening of the alloy occurring during the solution
heat treatment and further overaging at a relatively high aging temperature of 225 °C for
a relatively long time of 8 h. Indeed, similar YS and elongation were obtained for a
HPDC-Aural 2 alloy aged at the same temperature of 225 °C for 2 h, where the higher
UTS was probably related to the fourfold shorter aging time (i.e., with a lesser extent of
overaging) [2]. While the aging time was further longer (11 h) for an AA6060 alloy [5],
its aging temperature was lower, also leading to a lower degree of overaging and higher
strength along with the effect of other alloying elements. A strong strengthening effect
by adding 0.5%Cu in an A356 alloy was observed [6]. It should be noted that an A319
alloy in the T7 condition was reported to have relatively high strength, but the ductility
was fairly low (i.e., 1.6% only) [3]. Based on these comparisons and analyses, it is
expected that a higher strength for the Silafont®-36 alloy in the T7 condition could be
achieved at a slightly lower aging temperature and/or somewhat shorter aging time.
3.3 Cyclic Deformation Characteristics
Typical stress–strain hysteresis loops at the mid-life cycles and different total strain
amplitudes are shown in Fig. 2a. The minimum points are translated to the origin in
order to check if Masing behavior would be present in the alloy. The non-Masing phe-
nomenon was observed because of the increased dislocation–precipitate interactions
which promoted such behavior [7]. The loop size or area increased with increasing
total strain amplitude applied during cyclic deformation. The variation of stress ampli-
tude with the number of cycles (N) is plotted in Fig. 2b for different strain amplitudes.
It is seen that the stress amplitudes remained almost constant without significant
Table 1. Tensile properties of similar cast Al alloys in the T7 overaged condition.

Alloy Condition YS, MPa UTS, MPa %EL E, GPa
Present work T7 (Aging: 225 °C for 8 h) 130 ± 9 174 ± 21 7±3 72 ± 3
A356 + 0.5%Cu [6] T7 (Aging: 200–240 °C for 5 h) 210 272 4.7 72
HPDC-Aural 2 [2] T7 (Aging: 225 °C for 2 h) 133 ± 9 201 ± 6 8±4 –
AA6060 [5] T7 (Aging: 203 °C for 11 h) 160 ± 7 190 ± 12 10 ± 4 –
A319 [3] T7 (Aging: 250 °C for 4.25 h) 170 175 1.6 –
18 S. S. Dash et al.
350 170
(a) 0.8% (b)
300 150 0.6%
Stress amplitude, MPa

0.4%
250 130
Stress, MPa
0.3%
200
0.1% 0.4% 0.6% 0.8% 110
0.2%
150 0.3%
90
100 0.2% 0.1%
50 70
0 50
0 0.5 1 1.5 2 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Strain, % Number of cycles, N
Fig. 2. (a) Cyclic stress–strain loops of Silafont®-36-T7 alloy at mid-life cycles and varying total
strain amplitudes, with the minimum points translated to the origin and (b) changes of stress amplitude
with the number of cycles at different total strain amplitudes.
hardening or softening, reflecting a pretty stable state of cyclic deformation in the T7

condition. At the higher strain amplitudes of 0.6% and 0.8%, an initial hardening
occurred up to the third cycle, followed by some extent of cyclic softening until the
specimen failure.
4 Discussion
The solutionizing step during T7 heat treatment leads to larger Al grains and the
spheroidization of coralloid, eutectic Si particles (Fig. 1a). A lower amount of phase
boundaries due to the spheroidized Si particles reduces the work hardening rate, and the
flow stress develops at a slower rate compared with the as-cast condition [2]. This
results in a significant loss in the strength of the T7 heat-treated alloy, but the slow
build-up of the internal stress leads to improved ductility. Further overaging could lead
to the presence of β”-Mg2Si precipitates (Fig. 1b) that impart the minimum required
secondary strengthening in the alloys containing adequate magnesium. However, due to
the low content of magnesium in this alloy, the effect was not obvious.
Stabilization effect of the heat treatment process on the alloy is attained from Fig. 2b
as the stress amplitude values remain almost constant throughout the cyclic loading,
especially at both low and intermediate strain amplitudes applied. Like other cast alu-
minum alloys, the fatigue life of this alloy is significantly affected by the presence of
casting and heat treatment defects such as gas pores, surface blisters, and undissolved
AlFeMnSi inclusions [2, 6].
5 Conclusions
This article reports a study on the tensile and cyclic deformation behaviors of
Silafont®-36 alloy in the T7 overaging condition. As the yield strength and ultimate
tensile strength became lower, the alloy exhibited increased ductility in comparison
Monotonic and Cyclic Deformation Behavior of a Silafont®-36 Cast Aluminum… 19
with its as-cast state. The changes in the mechanical properties were discussed based on
the distinct changes in the microstructure due to solutionizing and overaging. The
cyclic stabilization feature of the alloy was basically observed with no hardening/soft-
ening effect until final failure. The lower fatigue life of the alloy was attributed to the
presence of pores, surface blisters, and undissolved inclusions.
Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research
Council of Canada (NSERC), Alberta Innovates, and Major Science and Technology Projects of
Yunnan Science and Technology Department (No. 202102AB080009) for the financial support. D.J.L.
and X.Q.Z. would like to thank the financial support of Yunnan Zeng Xiaoqin Expert Workstation (No.
202005AF150059). The authors would also like to thank Messrs. S.M.A.K. Mohammed, A. Machin,
and Q. Li for their assistance and easy access to the laboratory facilities of Toronto Metropolitan
University (formerly Ryerson University) and their assistance in the experiments.
References
1. Davis JR. Aluminum and aluminum alloys introduction. ASM Int. 2001:351–416. https://doi.
org/10.1361/authb2001
2. Liu R, Zheng J, Godlewski L, Zindel J, Li M, Li W, et al. Influence of pore characteristics and
eutectic particles on the tensile properties of Al–Si–Mn–Mg high pressure die casting alloy. Mater
Sci Eng A. 2020:783. https://doi.org/10.1016/j.msea.2020.139280.
3. Vandersluis E, Lombardi A, Ravindran C, Bois-Brochu A, Chiesa F, MacKay R. Factors influenc-
ing thermal conductivity and mechanical properties in 319 Al alloy cylinder heads. Mater Sci Eng
A. 2015;648:401–11. https://doi.org/10.1016/j.msea.2015.09.091.
4. Dash SS, Li DJ, Zeng XQ, Chen DL. Heterogeneous microstructure and deformation behavior of
an automotive grade aluminum alloy. J Alloys Compd. 2021;870:159413. https://doi.org/10.1016/j.
jallcom.2021.159413.
5. Zhang Y, Ji S, Scamans G, Fan Z. Interfacial characterisation of overcasting a cast Al-Si-Mg
(A356) alloy on a wrought Al-Mg-Si (AA6060) alloy. J Mater Process Technol. 2017;243:197–204.
https://doi.org/10.1016/j.jmatprotec.2016.12.022.
6. Natesan E, Eriksson S, Ahlström J, Persson C. Effect of temperature on deformation and fatigue
behaviour of A356–T7 cast aluminium alloys used in high specific power IC engine cylinder heads.
Materials (Basel). 2020;13:1202. https://doi.org/10.3390/ma13051202.
7. Dash SS, Li DJ, Zeng XQ, Li DY, Chen DL. Cyclic deformation behavior and fatigue life predic-
tion of an automotive cast aluminum alloy: a new method of determining intrinsic fatigue tough-
ness. Fatigue Fract Eng Mater Struct. 2022;45:725–38. https://doi.org/10.1111/ffe.13629.
Fabrication of Thin Rotary Tools Through
a Combination of Cross-Rolling
and Austempering-Tempering
H. M. Shirazi1 (*), R. Georges2, B. Ugulino2, R. Hernández3, and C. Blais1
1
Department of Mining, Metallurgical and Materials Engineering, Université Laval,
Quebec City, QC, Canada
hamidreza.mirzakouchakshirazi.1@ulaval.ca
2
FPInnovations, Quebec City, QC, Canada
3
Department of Wood and Forest Sciences, Université Laval, Quebec City, QC, Canada
Abstract. Production of thin rotary cutting tools involves several challenges related to
the maximization of uniform tensile and bending properties and levelness. This research
project investigates the feasibility of manufacturing these tools through cross-rolling and
austempering-tempering of a high-carbon low-alloy steel. In this regard, the combination
of the new deformation process and austempering-tempering was compared to a com-
mercial circular saw blade made by conventional hot rolling followed by quenching and
tempering. The results reveal that the cross-rolled austempered materials show higher
resistance to tempering. They also show that the combination of cross-rolling and
austempering-tempering results in higher: hardness (9%), yield stress (YS) (19%), UTS
(%19), and bending strength (BS) (22%), while maintaining similar ductility when com-
pared to the quenched and tempered material. The reason for such an improvement is
related to the presence of homogeneous α and ε fiber textures in combination with a
microstructure made of lower bainite. Tempering of bainite close to the bainitic transfor-
mation temperature range permitted to adjust strength/ductility and maximize the level-
ness of the final product. The novel manufacturing approach proposed enables the
production of thinner rotary tools that minimize waste during the cutting operation.
Keywords: Cross-rolling · Austempering-tempering · Mechanical properties ·

Crystallographic texture
1 Introduction
Cross-rolling reduces the asymmetry of the mechanical properties along rolling (RD)
and transverse direction (TD) by changing the slip systems and the Schmid factor [1].
Narayanswamy et al. [2] observed that cross-rolling strengthens the {100}<110> tex-
ture of α fiber in the deformed sheets. Austempering (AUST) near the Ms temperature
is reported to generate fine lower bainite and retained austenite (RA) having ultra-high
strength (>2 GPa) and considerable toughness [3, 4]. The previous studies revealed that
the original microstructure of the low-temperature bainite is retained during tempering
[5, 6]. This phenomenon is more distinct for high-carbon steels [6]. The rotary tool-like

https://doi.org/10.1007/978-3-031-17425-4_4
22 H. M. Shirazi et al.
circular saws used for guided sawing applications should have excellent isotropic
mechanical properties and perfect levelness [7]. The present study investigates the fea-
sibility of manufacturing high-performance circular saws through cross-rolling and
austempering-tempering (AUST-T).
High-carbon low-alloy steel plates (Fe- 0.70C- 0.60Si- 0.54Mn- 0.40Cr- 0.73Ni-
0.069Mo- 0.28Al (wt.%)) were cold-rolled to reduce their thickness from 5.3 to 2 mm
(62% reduction). For one series of specimens, unidirectional-rolling (UR) was per-
formed, while for the second one, cross-rolling (CR) was achieved by rotating the spec-
imens by 90° after each pass. The Ms and Bs temperatures were calculated at 222 and
397 °C, respectively, using the MUCG83 software [8]. The sheets were austenitized at
860 °C for 30 minutes, followed by austempering at 250 °C for 1 hour. The austem-
pered materials were clamp tempered at 400 °C for 2 hours. The microstructure and
texture evolution were characterized through OM, FESEM, XRD, EDS, and EBSD. The
hardness, tensile, and flexural properties were also measured and compared to a com-
mercial saw manufactured from the same material through the quench and tempering
(Q-T) process.
Figure 1 shows the microstructure of the CR AUST and AUST-T materials. AUST
develops lath-like bainite and filmy RA (Fig. 1a). Tempering causes the coarsening of
the microstructure and decomposition of RA to bainitic ferrite and carbide. The remain-
ing RA is characterized by blocky and filmy morphology (Fig. 1b). Transformation-
induced plasticity (TRIP) after AUST results in an ultra-high strength in the materials
(Table 1). The CR samples showed higher resistance to softening than unidirectionally
UR sheets during tempering (Table 1).
UR developed the <110>//RD and <111>//ND fibers and CR induced distinct
{100}<110> texture of α and ε fibers (Fig. 2). The {100}<110> component of CR
samples retained after AUST-T (Fig. 2). These components locate on the high-density
(a) (b)
RA
RA
10 µm 10 µm
Fig. 1. FESEM micrographs of the CR AUST (a) and AUST-T (b) sheets etched with Nital.
Fabrication of Thin Rotary Tools Through a Combination of Cross-Rolling… 23
Table 1. Mechanical properties of the AUST, AUST-T, and Q-T specimens.

Series dentification YS (MPa) UTS (MPa) BS (MPa) Hardness (HRC)
UR- AUST 1822 ± 22 2308 ± 16 4434 ± 31 58.0 ± 0.4
CR- AUST 1836 ± 15 2304 ± 12 4511 ± 76 57.5 ± 0.5
UR- AUST-T 1488 ± 20 1626 ± 23 3076 ± 45 48.0 ± 0.5
CR- AUST-T 1586 ± 25 1740 ± 19 3386 ± 19 49.0 ± 0.6
Q-T 1331 ± 11 1477 ± 2 2777 ± 28 45.0 ± 0.3
{100} <110>
UR UR-AUST-T
{111} <110>
{111}<112> CR-AUST-T
CR
Fig. 2. φ2 = 45° ODFs of the rolled and AUST-T sheets.
110 planes resulting in uniform tensile behavior along perpendicular directions.

However, UR sheets only have ductile behavior along RD. The intense texture of CR
samples showed higher resistance to softening during tempering (Table 1).
4 Conclusions
Based on the above research, it can be concluded that: (1) AUST-T results in a bainitic
microstructure with an excellent strength/ductility balance. (2) CR generates distinct
uniform {100}<110> texture, increasing material resistance to softening. (3) Production
of thinner rotary tools with the same material is possible with CR and AUST-T.
Acknowledgments The authors acknowledge the FPInnovations and the Natural Sciences and
Engineering Research Council of Canada (NSERC) for providing financial support (grant #CRDPJ
518967-17).
References
1. Amininejad A, et al. Improvement of strength-ductility balance of SAE 304 stainless steel

by asymmetric cross rolling. Mater Chem Phys. 2020;256:123668. https://doi.org/10.1016/j.
matchemphys.2020.123668.
2. Narayanswamy S, et al. Cross-rolling mediated microstructure and texture evolution in severely
cold-rolled and annealed ultrafine pearlite. Mater Charact. 2021;171:110751. https://doi.
org/10.1016/j.matchar.2020.110751.
3. Zhao L, et al. The combining effects of ausforming and below-Ms or above-Ms austempering on
the transformation kinetics, microstructure and mechanical properties of low-carbon bainitic steel.
Mater Des. 2019;183:108123. https://doi.org/10.1016/j.matdes.2019.108123.
24 H. M. Shirazi et al.
4. Jo M, et al. Enhancement of ballistic performance enabled bytransformation-induced plasticity

in high-strength bainitic steel. J Mater Sci Technol. 2021;84:219–29. https://doi.org/10.1016/j.
jmst.2020.12.059.
5. Chen K, et al. Enhanced mechanical properties by retained austenite in medium–carbon Si-rich
microalloyed steel treated by quenching–tempering, austempering and austempering–tempering
processes. Mater Sci Eng A. 2020;790:139742. https://doi.org/10.1016/j.msea.2020.139742.
6. Wang F, et al. Tempering response and improved mechanical properties in secondary hardened steel
by introducing an optimized austempering process. Mater Sci Eng A. 2021;807:140895. https://doi.
org/10.1016/j.msea.2021.140895.
7. Khorasany RMH, et al. Vibration characteristics of guided circular saws: experimental and numeri-
cal analyses. J Vib Acoust. 2012;134(6):061004. https://doi.org/10.1115/1.4006650.
8. Peet M, Bhadeshia HKDH. University of Cambridge, Cambridge, UK. https://www.phase-trans.
msm.cam.ac.uk/map/steel/programs/mucg83.html. Last accessed 2022/4/4.
Heat Treatment of Multi-Material Additively
Manufactured Maraging Steel and Stellite Alloy
Jubert Pasco1 (*) , Yuan Tian2, Kanwal Chadha1, 3 , and Clodualdo Aranas Jr.1
1
University of New Brunswick, Fredericton, NB, Canada
jubert.pasco@unb.ca
2
Voestalpine Additive Manufacturing Centre Ltd, Mississauga, ON, Canada
3
Planetary and Space Science Centre, University of New Brunswick, Fredericton, NB, Canada
Abstract. The ability to simplify the fabrication of a multi-material component with

complex architecture to a single part with sectional variation of properties has attracted
research interest in multi-material Additive Manufacturing (AM). For instance, mechani-
cal components that encounter demanding contact conditions such as bearings, cooling
channels, and fuel nozzles can take advantage of a multi-material framework with a tough
core and a wear-resistant surface coating [1]. Accordingly, the present work investigates
the possibility of combining the fabrication of a high-strength MS1 (iron-based) alloy and
a high wear resistance MP1 (cobalt-based) alloy which can be explored for such applica-
tions. MS1 (maraging steel) is a low-carbon age-hardenable steel with high mechanical
strength and low thermal deformation which makes it suitable for aerospace, defense,
biomedical, and tooling industry applications [2–4]. On the other hand, MP1 alloy has
high wear resistance and decent corrosion resistance which is desirable in applications
with multi-material conjunctions such as in automative, cladding, power generation, and
medical implants [5, 6]. Texture analysis of a dual-metal MP1-MS1 structure printed
using LPBF has been investigated in a previous work [1]. Results show that the necessary
mechanical properties of both materials cannot be achieved after printing and will there-
fore demand post-fabrication processing. Accordingly, the present work will also attempt
to establish a single heat treatment strategy appropriate for the MP1-MS1 dual-metal
structure. The resulting heat treatment effects on the microstructure, compositional varia-
tion in the interface, and mechanical properties will be investigated.
Keywords: Multi-material additive manufacturing · Laser powder bed fusion ·

Maraging steel · Stellite alloy
1 Introduction
The ability to simplify the fabrication of a multi-material component with complex

architecture to a single part with sectional variation of properties has attracted research
interest in multi-material Additive Manufacturing (AM). For instance, mechanical
components that encounter demanding contact conditions such as bearings, cooling
channels, and fuel nozzles can take advantage of a multi-material framework with a
tough core and a wear-resistant surface coating [1]. Accordingly, the present work

https://doi.org/10.1007/978-3-031-17425-4_5
26 J. Pasco et al.
investigates the possibility of combining the fabrication of a high-strength MS1 (iron-

based) alloy and a high wear resistance MP1 (cobalt-based) alloy which can be explored
for such applications. MS1 (maraging steel) is a low-carbon age-hardenable steel with
high mechanical strength and low thermal deformation which makes it suitable for
aerospace, defense, biomedical, and tooling industry applications [2–4]. On the other
hand, MP1 alloy has high wear resistance and decent corrosion resistance which is
desirable in applications with multi-material conjunctions such as in automative, clad-
ding, power generation, and medical implants [5, 6]. Texture analysis of a dual-metal
MP1-MS1 structure printed using LPBF has been investigated in a previous work [1].
Results show that the necessary mechanical properties of both materials cannot be
achieved after printing and will therefore demand post-fabrication processing.
Accordingly, the present work will also attempt to establish a single heat treatment
strategy appropriate for the MP1-MS1 dual-metal structure. The resulting heat treat-
ment effects on the microstructure, compositional variation in the interface, and
mechanical properties will be investigated.
2 Methodology
A dual-metal component produced using LPBF was fabricated by voestalpine Additive

Manufacturing Center Ltd. (vAMC) in Mississauga, Ontario, Canada, using MS1 and
MP1 gas-atomized powders. The LPBF process utilized an EOS M290 machine to
continuously print layers of MP1 powder on top of layers of MS1 powder at equal vol-
umes. The velocity of laser scanning was set to 1 cm/s, and the layer thickness was kept
to 50 μm for both powders using a 300 W laser system. The schematic diagram of the
printing process and the composition of the powders are reported in Reference [2].
Afterward, heat treatment was performed on representative samples of the as-printed
dual-metal component. A solutionizing heat treatment was first applied at 1148 °C for
6 hours followed by an aging heat treatment which soaked the samples to 490 °C for
6 hours under an argon atmosphere.
Fig. 1 Microstructure of the L-PBF MS1-MP1 dual-metal: (a) after printing and (b) after heat
treatment
Heat Treatment of Multi-Material Additively Manufactured Maraging Steel and Stellite… 27
A high relative density and crack-free surface was consistently observed in both alloys
after dual-metal LPBF printing, and the final product is shown in the left of Fig. 1. After
printing, the MP1 alloy region displays a heterogenous cellular microstructure having
diameters of 0.3–0.8 μm mixed with occasional columnar cells surface, while the MS1
region shows non-recrystallized martensite grains that contain a mixed fine cellular and
elongated columnar microstructure, as shown in Fig. 1a. No visible macrocracks, key-
hole defects and lack-of-fusion defects were detected in the interface. Furthermore,
columnar cells highlighted by red arrows show epitaxial growth across the interface and
the melt pool tracks, which signify good bonding and improved mechanical properties.
On the other hand, traces of previous melt tracks and the associated cellular microstruc-
ture were not detected in both the MP1 and MS1 regions after heat treatment, as shown
in Fig. 1b. The MP1 region contains larger, equiaxed γ-phase grains, while the MS1
region contains blocks of martensitic microstructure. Multiple bundles of parallel,
densely dislocated MS1 laths extend to the MP1 region following the <100>//BD and
<110>//BD directions in the interface. The elongated columnar structures after printing
also conform in this pattern due to the thermal gradient, which suggests that epitaxial
growth is maintained even if homogenization eliminated the cellular microstructure.
EBSD analysis, nano-indentation, and mechanical testing was then performed to
compare the properties of the as-printed and heat-treated samples, and part of the results
are presented in Fig. 2. The average hardness in the MS1 region substantially increased
from 501 ± 8 HV to 771 ± 9 HV after heat treatment, signifying the occurrence of pre-
cipitation strengthening due to formation of fine particles such as Ni3Ti, Ni3Mo and
Fe2Mo [7]. The IPF map shown in Fig. 2c also shows that the MS1 region revealed
Fig. 2 Selected test results of MS1–MP1 interface investigation (a) Microstructure of the nanoinden-
tation test points, (b) Vickers hardness profile along the interface of as-printed and heat-treated sam-
ples, (c) IPF grain orientation map along the interface of the heat-treated sample
28 J. Pasco et al.
massive martensite blocks that are almost twice the size of the microstructure after
printing [1]. The average grain size increased to 17 μm from 9 μm and the consistency
of the grain sizes is apparent. The extremely high solutionizing temperature granted
sufficient stress relief and homogenization, forming equiaxed lath martensite after
nitrogen cooling. On the other hand, the average hardness in the MP1 region slightly
decreased from 451 ± 6 HV to 355 ± 7 HV after heat treatment, which reflects success-
ful homogenization and slight increase in ductility [8]. The IPF map shown in Fig. 2c
also shows that the heavily dislocated fiber structure after printing [1] was replaced
with a massive equiaxed γ-phase solution after heat treatment. Extensive stress relief
can be observed as the average grain size increased from 98 μm to 238 μm and the
aspect ratio decreased to 2.8 from 6.9. This suggests that recrystallization occurred
using the driving force of residual stresses present in the fibers. Lastly, the interface
region had an average hardness of 470 ± 7 HV and 513 ± 9 HV before and after heat
treatment, respectively. Since both hardness ranges lie between the values of the base
material at both conditions, this can be indirect evidence of the absence of intermetal-
lics with higher/lower hardness values that can form from the base alloys.
References
1. Chadha K, Tian Y, Pasco J, Aranas C. Dual-metal laser powder bed fusion of iron- and cobalt-based
alloys. Mater Charact. 2021;178:111285. https://doi.org/10.1016/j.matchar.2021.111285.
2. Bai Y, Zhao C, Zhang Y, Wang H. Microstructure and mechanical properties of additively
manufactured multi-material component with maraging steel on CrMn steel. Mater Sci Eng
A. 2021;802:140630. https://doi.org/10.1016/j.msea.2020.140630.
3. Tian Y, Palad R, Aranas C. Microstructural evolution and mechanical properties of a newly
designed steel fabricated by laser powder bed fusion. Addit Manuf. 2020;36:101495. https://doi.
org/10.1016/j.addma.2020.101495.
4. Chadha K, Tian Y, Bocher P, Spray JG, Aranas C Jr. Microstructure evolution, mechanical proper-
ties and deformation behavior of an additively manufactured maraging steel. Materials. 2020;13
https://doi.org/10.3390/ma13102380.
5. Ganesh P, Moitra A, Tiwari P, Sathyanarayanan S, Kumar H, Rai SK, Kaul R, Paul CP, Prasad RC,
Kukreja LM. Fracture behavior of laser-clad joint of Stellite 21 on AISI 316L stainless steel. Mater
Sci Eng A. 2010;527:3748–56. https://doi.org/10.1016/j.msea.2010.03.017.
6. Vamsi Krishna B, Xue W, Bose S, Bandyopadhyay A. Functionally graded Co–Cr–Mo coat-
ing on Ti–6Al–4V alloy structures. Acta Biomater. 2008;4:697–706. https://doi.org/10.1016/j.
actbio.2007.10.005.
7. Bai Y, Yang Y, Wang D, Zhang M. Influence mechanism of parameters process and mechanical
properties evolution mechanism of maraging steel 300 by selective laser melting. Mater Sci Eng
8. Kajima Y, Takaichi A, Kittikundecha N, Nakamoto T, Kimura T, Nomura N, Kawasaki A, Hanawa
T, Takahashi H, Wakabayashi N. Effect of heat-treatment temperature on microstructures and
mechanical properties of Co–Cr–Mo alloys fabricated by selective laser melting. Mater Sci Eng
Microstructural Characterization of Pack-Borided
H13 Tool Steel Processed Using Directed Energy
Deposition
Owen Craig1 (*), Riley Roache1, and Kevin Plucknett1

1
Mechanical Engineering, Dalhousie University, Halifax, NS, Canada
Ow728801@dal.ca; Kevin.Plucknett@dal.ca
Abstract. H13 tool steel samples were prepared using directed energy deposition (DED)
and were then pack-borided at temperatures of 800 °C, 850 °C, 900 °C, or 950 °C, with a
hold time of 4 hours for each temperature. As a baseline, wrought H13 tool steel samples
were also pack-borided under identical conditions. The microstructures and crystalline
phases for the borided samples were characterized using X-ray diffraction (XRD), scan-
ning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS),
wavelength-dispersive X-ray spectroscopy (WDS), hardness, and confocal laser scanning
microscope (CLSM). After applying the thermochemical heat treatment to the tool steel
samples, a characteristic “saw-tooth” appearance of the boride layer is observed in the
cross-section of all samples, which is slightly flattened due to the high alloying in the tool
steel. This boride layer consists of a FeB layer (tensile) on the outer surface, with a Fe2B
layer (compressive) below. With increasing boriding temperature, the borided layer
increased in thickness. WDS confirmed that the elements Si, C, and Cr were redistributed
due to the formation of the boride layer. It was observed that the boriding process signifi-
cantly improved the hardness of the samples. There was no clear trend in terms of the
surface roughness after boriding at the various temperatures, with samples processed at
900 °C having the highest surface roughness. The surface of the as-borided samples shows
a porous appearance with cracking due to the residual tensile stresses of the FeB layer.
Keywords: Directed energy deposition (DED) · H13 tool steel · Pack Boriding
Hardness.
1 Introduction
Boriding is a thermochemical surface-hardening technique that not only is commonly

applied to ferrous materials but also has applications with non-ferrous materials (e.g.,
titanium alloys). Boronizing is achieved by having the sample in contact with a boron-
containing solid powder, paste, liquid, or gas forms at temperatures ranging from 700 to
1200 °C for diffusion to occur [1, 2]. At these elevated temperatures, the boron atoms
diffuse into the metal lattice, forming borides with the substrate and alloying elements,
which develops a borided surface layer. The time and temperature of this process will
influence the thickness of the boride layer, with higher temperatures and times promot-
ing faster diffusion times and thicker layers [1, 3, 4]. Typical layer thicknesses range

https://doi.org/10.1007/978-3-031-17425-4_6
30 O. Craig et al.
from 12.5 to 50 μm [3]. The purpose of this process is to enhance the surface hardness
and abrasive wear resistance of the sample, where the hardness can reach hardness val-
ues of 2000 HV for steels, while maintaining a low coefficient of friction and with the
hardness retained at higher temperatures compared to other surface hardening tech-
niques [1, 2, 4]. It also combats other wear mechanisms such as adhesion, tribo-
oxidation, and surface fatigue [3]. In steels, FeB and Fe2B phases are developed with
the morphology of the boride layer dependent on the composition of the substrate. The
formation of the boride layer can be single phase or dual phase, forming a saw-tooth
appearance. The single phase is comprised of Fe2B and the dual phase is comprised of
an exterior FeB layer and interior Fe2B layer [1–3]. The FeB phase is undesired because
it is brittle and is under tensile stresses due to the coefficient of expansion of 23•10−6 K−1
compared to that of iron with a coefficient of expansion of 11.9•10−6 K−1, whereas Fe2B
is less brittle and is under compressive stresses (7.85•10−6 K−1) [1, 2, 5–7]. The FeB
layer is only detrimental if the layer is continuous and can lead to cracking between the
FeB/Fe2B interface [3]. Silicon (Si) and carbon are insoluble in FeB-Fe2B and are driven
ahead of the layer [1, 3]. A major disadvantage of boriding is the layer is easily removed,
or fractures from grinding, and care must be taken to obtain close tolerances [3]. Another
disadvantage is the inflexibility of the process, time, and labor required which is not as
cost effective as other thermochemical surface-hardening techniques [3]. Several stud-
ies have investigated boriding tool steel [5–9]. Pack boriding is the most common
method due to the ease of handling, safety, and possibility to change the boron powder
chemistry [4]. Powder is packed around the sample and placed in a heat-resistant con-
tainer. For this method, it is recommended that the process be performed in a controlled
atmosphere because oxygen-bearing compounds adversely affect the boriding process
[3]. The focus of this study was examining the influence of pack boriding laser-directed
energy deposition (DED) processed H13 tool steel under varying process temperatures
and comparing additive manufactured samples to a wrought equivalent.
2 Experimental Procedure
2.1 Raw Materials and Sample Geometries
In this investigation, gas-atomized, pre-alloyed H13 powder was used, with a starting
particle size of D10 = 58.3 μm, D50 = 78.7 μm, and D90 = 107 μm. The apparent density
of the powder was 4.260 g/cm3. Cylindrical samples were printed on wrought-annealed
H13 tool steel substrates (101.6 mm × 101.6 mm × 6.35 mm) using varying printing
parameters resulting in four samples. The samples have a diameter of 20 mm and are 15
layers thick. Printing parameters are listed in Table 1. Wrought AISI H13 hot work tool
steel was also purchased from Hudson Tool Steel Corporation (Rockford, IL, USA) for
baseline evaluation. The as-received (AR) wrought material was supplied as bar stock,
with a diameter of 19.05 mm, with four samples sectioned into a thickness of 5 mm; the
bar was also received in the annealed condition. The composition of the materials used
in the study are listed in Table 2.
Microstructural Characterization of Pack-Borided H13 Tool Steel Processed Using… 31
Table 1. DED system printing parameters.
Parameter Value Units

Laser power 400 W
Scan speed 500 mm/minute
Feed rate 4.5 g/minute
Hatch spacing 0.381 mm
Layer thickness 0.251 mm
Spot size 600 μm
Nozzle stand off 9.525 mm
Hatch orientation 0°, 45°, 90°, 135° Degrees
Table 2. Chemical compositions of the “as-received” H13 powder (prior to DED use), wrought H13
substrate. AISI standard specifications are provided vs. measured values (note that Fe comprises the
balance in each case).
Composition (wt.%)
Designation C Si Mn Cr Mo V
AISI H13 0.32–0.45 0.8–1.2 0.2–0.5 4.75–5.5 1.1–1.75 0.8–1.2
Wrought H13 0.4 0.69 0.34 4.64 1.17 0.76
As-received powder 0.372 1.02 0.42 5.74 1.46 1.05
H13 substrate 0.39 0.63 0.30 4.73 1.37 0.80
2.2 Baseline Heat Treatments
All samples were polished before heat treating. The heat treatments consisted of pack
boriding samples in stainless steel bags with a solid medium (Ekabor-II powder), con-
ducted in a vacuum furnace at a pressure of 1 × 10−6 kPa at temperatures of 800, 850,
900, and 950 °C for 4 hours. This was followed by a furnace cool to room temperature.
The wrought samples are identified by the prefix “W” followed by the heat treatment
temperature (i.e., W 800). The laser DED printed samples are identified in a similar
manner (i.e., DED 800).
The surface roughness increased with boriding temperature, as shown in Table 3. This
increase in roughness was observed in prior research [10–14]. Increasing roughness is
due to the formation of the boride crystals [15]. SEM imaging of the boride surface has
a porous appearance, as illustrated in Fig. 1. Since FeB has tensile residual stresses,
cracking can be seen in on the as-borided surface (Fig. 1).
Typical of the boriding process for steels of lower chromium (Cr) content, the boride
layer consists of a FeB on the outer surface with a “saw-tooth” morphology, appearing
darker in contrast to the lighter Fe2B inner layer, as seen in Fig. 2, and confirmed by
XRD (not shown). It is known that the FeB is brittle and is under tensile residual stresses
32 O. Craig et al.
Table 3. Surface roughness (Sa) of as-borided wrought and DED H13 tool steel.
Sample Sa (μm)
W 800 0.232 ± 0.007
DED 800 0.223 ± 0.002
W 850 0.202 ± 0.006
DED 850 0.220 ± 0.002
W 900 0.254 ± 0.027
DED 900 0.257 ± 0.005
W 950 0.305 ± 0.003
DED 950 0.233 ± 0.006
Fig. 1. As-borided surface of wrought H13 heat treated at 950 °C. The red arrow indicating surface
cracking.
with transverse cracking observed in this layer. Portions of the FeB layer were also
observed to be spalling off. The process temperature influenced the thickness of the
boride layer, with processing temperatures of 950 °C and 800 °C producing the thickest
and thinnest, respectively. The values of the FeB, Fe2B, and total layer thicknesses are
listed in Table 4. It is evident that the higher temperatures resulted in a thicker boride
layer due to accelerated diffusion at elevated temperatures. Alloying additions of Si, Cr,
V, and Mo will reduce the thickness of the overall boride layer [3, 16], but the chro-
mium content has a greater influence on the boride layer morphology. As the chromium
content increases, it forms chromium-boron phases and flattens the boride layer [2, 3].
If the silicon content exceeds 0.8 wt%, it promotes the formation of brittle structures
and soft ferrite situated under the layer, which has a low loading capacity [1, 3]. EDS
mapping of the cross-section suggests that Si and Cr are displaced by the boride layer.
Si has accumulated in the transition zone as illustrated in Fig. 3. Cr is also known to
accumulate in the transition region, but this redistribution may be due to Cr-C reactions
Fig. 2. Cross-sectional view of DED 800 (a) and DED 950 (b).
that are forming alloy carbides [17, 18]. Another study found that the Cr is depleted
from the matrix and enters the boride layer [19]. These results agree with this study, as
the EDS mapping of Cr suggests the transition zone is depleted. Cr also promotes the
formation of boron-rich formations and will reduce the thickness of the boride layer
[20]. Cr is a ferrite-stabilizing element; therefore, it raises the transformation tempera-
ture [21, 22]. The Si has been displaced ahead of the boride layer because it is insoluble
in this phase. An EDS map (Fig. 3) highlights the displacement of Si ahead of the
boride layer. This is not ideal because Si is a ferrite stabilizer which reduces the wear
34 O. Craig et al.
Table 4. FeB, Fe2B, and total boride layer thicknesses.

Sample FeB thickness (μm) Fe2B thickness (μm) Total thickness (μm)
W800 5.61 ± 1.14 6.67 ± 1.74 12.28 ± 2.08
DED 800 5.90 ± 1.53 7.28 ± 1.90 13.19 ± 2.44
W850 11.98 ± 1.22 9.89 ± 2.07 21.87 ± 2.40
DED 850 13.05 ± 2.26 10.76 ± 2.20 23.81 ± 3.16
W900 23.92 ± 3.29 19.39 ± 4.74 43.28 ± 5.77
DED 900 21.38 ± 2.29 20.28 ± 4.12 41.66 ± 4.72
W950 44.23 ± 4.42 33.01 ± 8.02 77.23 ± 9.16
DED 950 44.52 ± 4.85 34.25 ± 3.28 78.78 ± 5.86
Fig. 3. EDS map of Cr and Si distribution due to the formation of the boride layer in sample DED 900.
resistance, with failure caused by the “eggshell effect” [21]. A higher Si concentration
at the transition layer makes it difficult to fully austenitize this region, potentially result-
ing in issues when an additional heat treatment is required [22, 23]. Figure 4a, b are
WDS analyses, which indicate that boron distribution in the DED sample is similar to
the wrought equivalent. The boron content is initially higher at the surface and then
decreases until it reaches a negligible amount into the matrix. There may be some devi-
ations in the formation of the boride layer in the laser DED processed samples, as the
top surface encounters less thermal cycling compared to the lower layers which have
been reheated several times during the DED “build.” This cyclic heating prior to the
boriding treatment may have an impact on the formation of the boride layer but this
study only focused on the top layer (i.e., last printed layer) of the sample.
Fig. 4. Boron content (wt%) of samples processed at 800 °C and 850 °C (a) and for temperatures
900 °C and 950 °C (b).
4 Conclusions
From the present study, the following conclusions have been made about pack boriding
of wrought and DED processed H13 tool steel:
36 O. Craig et al.
• The surface of the borided samples appears porous with surface cracking seen in the
samples processed at 950 °C, due to significant FeB formation.
• The borided structure consists of two layers; the outer layer composed of FeB and
the inner layer comprised of Fe2B. This layer has a “saw-tooth” morphology.
• Higher heat treatment temperatures resulted in a thicker boride layer.
• Insoluble alloying additions such as Si are pushed ahead of the boride layer and
accumulate in the transition region. The addition of Cr is depleted from the matrix
and is distributed in the boride layer.
• There seems to be no significant differences in the formation of the dual-phase
boride layer between wrought H13 and DED processed H13 tool steel.
Acknowledgments The authors would like to thank the technicians, Pat with helping imaging and
Danny conducting the WDS analyses. This work was support through the NSERC Strategic Project
Network Grant Holistic Innovation in Additive Manufacturing (HI-AM) through Grant No. NETGP
494158-16.
References
1. Casteletti LC, Lombardi AN, Totten GE. Encyclopedia of tribology. Boston: Springer Boston;
2013. https://doi.org/10.1007/978-0-387-92897-5.
2. Hutchings I, Shipway P. Tribology: friction and wear of engineering materials. 2nd ed. Oxford:
Elsevier; 2017.
3. Davis JR. Surface hardening of steels, vol. 1, no. 1. Materials Park: ASM International; 2002.
4. ASM International. ASM handbook, Vol. 4. Heat-treating. ASM International; 1991.
5. Rile M. Reasons for the formation of cracks in boride coatings on steel. Met Sci Heat Treat.
1974;16(10):836–8. https://doi.org/10.1007/BF00664246.
6. Kulka M, Makuch N, Piasecki A. Nanomechanical characterization and fracture toughness
of FeB and Fe2B iron borides produced by gas boriding of Armco iron. Surf Coat Technol.
2017;325:515–32. https://doi.org/10.1016/j.surfcoat.2017.07.020.
7. Ebert HP, Braxmeier S, Neubert D. Intercomparison of thermophysical property measurements on
iron and steels. Int J Thermophys. 2019;40(11):1–18. https://doi.org/10.1007/s10765-019-2568-3.
8. Muhammad W. Boriding of high carbon high chromium cold work tool steel. IOP Conf Ser Mater
Sci Eng. 2014;60(1) https://doi.org/10.1088/1757-899X/60/1/012062.
9. Manthani N, Joshi A, Pawar V, Singh R. Improvement in wear resistance of AISI H13 steel by
Pack-Boronizing method. J Nanosci Nanotechnol Res. 2018;2(2):2–4. [Online]. Available: http://
www.imedpub.com/journal-nanoscience-nanotechnology-research/
10. Erdoğan A. Investigation of high temperature dry sliding behavior of borided H13 hot work tool
steel with nanoboron powder. Surf Coat Technol. 2019;357(September 2018):886–95. https://doi.
org/10.1016/j.surfcoat.2018.10.066.
11. Cárdenas EEV, et al. Characterization and wear performance of boride phases over tool steel sub-
strates. Adv Mech Eng. 2016;8(2):1–10. https://doi.org/10.1177/1687814016630257.
12. Günen A, et al. Properties and corrosion resistance of AISI H13 hot-work tool steel with Borided
B4C powders. Met Mater Int. 2019:1329–40. https://doi.org/10.1007/s12540-019-00421-0.
13. Krelling AP, Teixeira F, Da Costa CE, De Almeida EADS, Zappelino B, Milan JCG. Microabrasive
wear behavior of borided steel abraded by SiO 2 particles. J Mater Res Technol. 2019;8(1):766–76.
https://doi.org/10.1016/j.jmrt.2018.06.004.
14. Krelling AP, Milan JCG, De Costa CE. Tribological behaviour of borided H13 steel with different
boriding agents. Surf Eng. 2015;31(8):581–7. https://doi.org/10.1179/1743294414Y.0000000423.
15. Şahin S. Effects of boronizing process on the surface roughness and dimensions of AISI 1020, AISI
1040 and AISI 2714. J Mater Process Technol. 2009;209(4):1736–41. https://doi.org/10.1016/j.
jmatprotec.2008.04.040.
16. Carbucicchio M, Palombarini G. Effects of alloying elements on the growth of iron boride coat-
ings. J Mater Sci Lett. 1987;6(10):1147–9. https://doi.org/10.1007/BF01729165.
17. Palombarini G, Carbucicchio M. Influence of carbon on the chromium redistribution when borid-
ing iron alloys. J Mater Sci Lett. 1993;12(11):797–8. https://doi.org/10.1007/BF00277975.
18. Matiašovský K, Chrenková-Paučírová M, Fellner P, Makyta M. Electrochemical and ther-
mochemical boriding in molten salts. Surf Coat Technol. 1988;35(1–2):133–49. https://doi.
org/10.1016/0257-8972(88)90064-3.
19. Badini C, Gianoglio C, Pradelli G. Preferential distribution of chromium and nickel in the
borided layer obtained on synthetic Fe-Cr-Ni alloys. J Mater Sci. 1986;21(5):1721–9. https://doi.
org/10.1007/BF01114731.
20. Carbucicchio M, Sambogna G. Influence of chromium on boride coatings produced on iron alloys.
Thin Solid Films. 1985;126(3–4):299–305. https://doi.org/10.1016/0040-6090(85)90324-4.
21. Smith WF. Structure and properties of engineering alloys. 1st ed. New York: McGraw-Hill
Inc.; 1981.
22. Fiedler HC, Hayes WJ. The formation of a solf layer in borided hot work die steels. Metall Trans.
1970;1(4):1071–3. https://doi.org/10.1007/BF02811810.
23. Yang H, Wu X, Cao G, Yang Z. Enhanced boronizing kinetics and high temperature wear resis-
tance of H13 steel with boriding treatment assisted by air blast shot peening. Surf Coat Technol.
2016;307:506–16. https://doi.org/10.1016/j.surfcoat.2016.09.029.
Influence of Layer Thickness Upon the Dimensional
Accuracy and Surface Roughness of AISI D2 Tool
Steel Manufactured Using Directed Energy
Deposition
S. M. T. Omar1 and K. P. Plucknett1 (*)
1
Dalhousie University, Halifax, NS, Canada
kevin.plucknett@dal.ca
Abstract. This research focuses upon the influences on the dimensional accuracy and
surface roughness of directed energy deposition (DED) processed AISI D2 tool steel
through alteration of the layer thickness, while keeping the remaining system operating
parameters at constant values. The DED process involved laser deposition of AISI D2 tool
steel powder on an annealed AISI D2 substrate, in order to build multi-layered, rectangu-
lar samples. After deposition, the DED manufactured specimens were assessed without
any further machining processes. The dimensional accuracy (i.e., actual height of the
deposited specimen in comparison to the designed height) was evaluated using both a
manual approach, using calipers, and a computational method, using confocal laser scan-
ning microscopy (CLSM). It was observed that an increase in the layer thickness decreases
the degree of overbuilding. Moreover, CLSM was used to examine the top and side sur-
faces roughness of the DED processed specimens. It was found that the use of lower layer
thickness improves the top surface roughness. However, there was not a monotonic effect
for layer thickness upon the side surface roughness of the DED processed samples.
Keywords: Additive manufacturing (AM) · Direct energy deposition (DED) ·

Layer thickness · Surface roughness · Tool steel
1 Introduction
The popularity of laser-based directed energy deposition (DED) additive manufactur-

ing is rapidly being increased, thanks to their capability of generating complex parts,
which are hard to be accomplished using conventional manufacturing methods, as well
as being occupied for repairing corroded or cracked surfaces, which increase the life-
time high-value tools and dies [1].
Many researchers have exerted considerable efforts to characterize the influence of
layer thickness upon the geometrical properties of DED processed structure. Mazumder
et al. [2] found that the increase of layer thickness resulted in a rougher wall surface. In
addition, Craig et al. [3] demonstrated that overbuilding of the proposed height was

https://doi.org/10.1007/978-3-031-17425-4_7
40 S. M. T. Omar and K. P. Plucknett
observed using a finer layer thickness, while using coarser layer thickness can lead to
underbuilding.
2 Experimental
In this work, the laser DED technique was used to deposit AISI D2 tool steel powder
onto a substrate of the same material. A detailed description for the experimental setup
for the DED system, (i.e., process parameters and the dimensions of DED processed
parts) is illustrated in Table 1. Three different layer thicknesses, with values of 0.125,
0.25, and 0.5 mm, were used for samples D1, D2, and D4, respectively, while the other
process parameters were kept at constant values, as listed in Table 1.
The dimensional accuracy was evaluated using both a manual technique, using cali-
pers, and a computational method, using confocal laser scanning microscopy (CLSM).
CLSM was also used to examine the top and side surfaces roughness of the DED pro-
cessed specimens.
3 Results
3.1 Dimensional Accuracy
The relation between height error and layer thickness is presented in Fig. 1. It was
found that the use of lower layer thicknesses led to higher height error. This arose
because of the deposition of a higher number of layers when using lower thickness,
Table 1. Set of process parameters.
Parameter Value
Sample dimensions, L × W × H (mm) 10 × 10 × 10
Laser power (W) 400
Scanning speed (mm/min) 600
Powder feed rate (g/min) 6.5
Layer thickness (mm) 0.125, 0.25, 0.5
Hatch spacing (mm) 0.381
3
Height error
2
(mm)
1
0
D1 D2 D4
Fig. 1. Layer thickness effects upon measured height error.

Influence of Layer Thickness Upon the Dimensional Accuracy and Surface Roughness… 41
50
40
Sa (μm)
30
20
10
0
D1 D2 D4
Top roughness Side roughness
Fig. 2. The influence of the layer thickness upon the measured top and side surface roughness.
which ultimately leads to a greater degree of overbuilding (the as-printed height is

higher than the designed one).
3.2 Surface Roughness
As illustrated in Fig. 2, it was observed that the top surface roughness is significantly
improved by decreasing the layer thickness. This could be attributed to the higher heat
content at lower layer thickness, as a result of the deposition of a greater number of
layers, and hence an increased number of heating cycles. This, in turn, leads to better
solidification conditions which results in a reduced amount of partially adhered powder
particles on the surface. However, in contrast, there was no significant trend for the
impact of layer thickness upon the side surface roughness, with the minimum side sur-
face roughness being obtained at layer thickness of 0.25 mm.
4 Conclusions
The present work has focused upon the effects of layer thickness upon the deposition
height error and surface roughness of the DED processed samples. The following con-
clusions were obtained:
• In terms of dimensional accuracy, it was found that the lower the layer thickness, the
higher the height error. This arises because of the higher amount of deposited pow-
der at lower layer thickness, compared to thicker ones.
• The use of lower layer thickness does slightly reduce the surface roughness.
Acknowledgments The authors would like to thank the technical assistance of Mr. Randy Cooke and
Dr. Greg Sweet (Dalhousie University) with DED printing of the samples. This work was financially
supported through the NSERC Strategic Project Network Grant Holistic Innovation in Additive
Manufacturing (HI-AM) through [Grant No. NETGP 494158-16].
42 S. M. T. Omar and K. P. Plucknett
References
1. Toyserkani E, Khajepour A, Corbin S. Laser cladding. 2nd ed. Florida: CRC Press; 2005.
2. Mazumder J, Schifferer A, Choi J. Direct materials deposition: designed macro and microstructure.
Mat Res Innov. 1999;3:118–31.
3. Craig O, Bois-Brochu A, Plucknett K. Geometry and surface characteristics of H13 hot-work
tool steel manufactured using laser-directed energy deposition. Int J Adv Manuf Technol.
2021;116:699–718. https://doi.org/10.1007/s00170-021-07322-w.
Laser Ablation of Thin Films for Flexible Sensor
Manufacturing
Ningyue Mao1, Pablo D. Enrique1, Y. Norman Zhou1, and Peng Peng1 (*)
1
Centre of Advanced Materials Joining, Department of Mechanical and Mechatronics
Engineering, University of Waterloo, Waterloo, ON, Canada
peng.peng@uwaterloo.ca
Abstract. This study investigates a mask-free and inkless method using nanosecond
pulsed UV laser ablation for sensor manufacturing. This method greatly simplifies the
patterning process into a single step and allows for a short turnaround time for custom
designs. In general, the quality of the fabricated sensor depends on laser power, scanning
speed, pulse frequency, and scanning strategy. In this study, the influence of these laser
parameters on the ablation width and edge quality of NiCr thin films is investigated. The
laser ablation process is optimized to obtain high-quality thin-film sensors that have the
potential to offer custom sensor production at a low cost and with short lead times.
Keywords: Laser processing · Thin films · Flexible devices.
1 Introduction
Today, with the development of the Internet of Things (IoT), the demand for specialized
sensors that can measure physical properties is greatly increased. Conventional elec-
tronics manufacturing techniques such as photolithography are typically used for large-
scale production of sensors; however, this process offers little flexibility in terms of
customized designs for small production volumes, due to high fixed costs and long lead
times. A promising alternative manufacturing method is using laser ablation to pattern
flexible thin films, which has been implemented for applications such as antenna [1],
temperature [2], and pressure sensors [3].
In this work, NiCr thin films on flexible polyimide substrate were used to investigate
the effect of various laser parameters including average power, repetition frequency,
and scanning speed on the ablation width for flexible thin-film sensors manufacturing.

https://doi.org/10.1007/978-3-031-17425-4_8
44 N. Mao et al.
Fig. 1. (a) Schematic diagram of the experimental setup; (b) microscope image showing the ablated
trace with different scanning speed.
Table 1. Summary of laser parameters used in this study.

Factors Values
Power (W) 0.2; 0.4; 0.6; 0.7; 0.9; 1
Scanning speed (m/s) 0.1; 0.3; 0.5
Frequency (MHz) 0.03; 0.05; 0.075; 0.1
The samples used in this study were 5-um thick NiCr (80%Ni, 20%Cr) foils with
polyimide-laminated substrates. Figure 1a illustrates the UV laser system used for thin-
film patterning. A diode-pumped solid-state (DPSS) Nd: YVO4 UV pulsed laser
(Samurai Desktop 3500) with a wavelength of 355 nm assembled with a 103 mm f-theta
lens was used for the experiment, exhibiting a focused beam diameter of 10 μm. The
maximal output power of the laser system is 1 W and the repetition frequency could
range from 30 kHz to 100 kHz.
The detailed laser parameters used in this study to ablate a straight line with a length
of 500 μm on NiCr film are summarized in Table 1. The surface morphology of the
laser-ablated lines was evaluated using an Olympus BX51M optical microscope and the
width of the lines ablated using different parameters was measured and compared. To
quantify the effect of laser parameters on ablation width, a response surface methodol-
ogy was implemented for this analysis using Develve software.
Laser Ablation of Thin Films for Flexible Sensor Manufacturing 45
A high-accuracy model was obtained that summarized the relationship between abla-
tion width and process parameters as shown in Eq. 1 (R2 = 0.937). The actual and pre-
dicted values of the ablated width are distributed uniformly along the identity line
(Fig. 2).
Width 4.51 Power 127.79 Frequency 4.5 Scanning Speed 16.96 (1)
Figure 3 shows the comparison of the impact of the three parameters on the ablation
width. Each data point represents the average of a range of line widths created at that
parameter level. The laser power and repetition frequency were major factors in deter-
mining the amount of energy passed onto the thin film, hence, they have a more signifi-
cant effect on the ablation width. An increase in frequency decreases the energy per
pulse, resulting in narrower ablation widths, while an increase in power increases the
energy per pulse and widens the ablation width. The scanning speed has a less pro-
nounced effect, although the slight decrease in width observed with faster scanning
speeds can be attributed to less overlap between pulses and reduced heating of the foil.
The above results can be used to develop process maps that allow for the selection
of parameters based on required feature sizes in thin-film sensors. A lower bound to the
operating space is set by selecting parameters with sufficient overlap between adjacent
pulses, while parameters that form large heat-affected zones provide an upper bound.
Fig. 2. A comparison of actual and predicted ablation width using the response surface model.
46 N. Mao et al.
16 16 16
Ablation Width (μm)
14 14 14
12 12 12
10 10 10
8 8 8
6 6 6
4 4 4
0.2 0.4 0.6 0.8 1.0 100 200 300 400 500 20 30 40 50 60 70 80 90 100 110
Power (W) Scanning Speed (mm/s) Frequency (kHz)
A B C
Fig. 3. The effect of (a) power, (b) scanning speed, and (c) frequency on ablation width with error bars
representing standard deviations.
Acknowledgments This work was supported with funding from the Natural Sciences and Engineering
Research Council of Canada (NSERC) and the Canada Research Chairs (CRC) Program and was per-
formed at the University of Waterloo’s Centre for Advanced Materials Joining.
References
1. Esakkimuthu M, Balakrishnapillai Suseela S, Sankararajan R, Gupta A, Rana G. Laser patterning

of thin film copper and ITO on flexible substrates for terahertz antenna applications. JLMN-J Laser
Micro/Nanoeng. 2017;12(3) https://doi.org/10.2961/jlmn.2017.03.0023.
2. Dankoco MD, Bènevent E, Bergeret E, Gallais L, Bendahan M. Temperature sensor on flex-
ible substrate patterned by laser ablation. Conf. Proc. - 10th Int. Conf. Adv. Semicond.
Devices Microsystems, ASDAM 2014, pp. 9–12, Dec 2014, https://doi.org/10.1109/
ASDAM.2014.6998634.
3. Das PS, Chhetry A, Maharjan P, Rasel MS, Park JY. A laser ablated graphene-based flexible
self-powered pressure sensor for human gestures and finger pulse monitoring. Nano Res 128.
2019;12(8):1789–95. https://doi.org/10.1007/S12274-019-2433-5.
Transient Softening at the Fusion Boundary of Spot
Welds in Advanced High-Strength Steels: Dark Horse
of the Heat-Affected Zone
Oleksii Sherepenko1 (*), Dileep Ramachandran1, Alireza Mohamadizadeh1,

and Elliot Biro1
1
Department of Mechanical and Mechatronics Engineering, University of Waterloo,
Waterloo, ON, Canada
oleksii.sherepenko@uwaterloo.ca
Abstract. The use of Advanced High-Strength Steels (AHSS) in body-in-white (BiW)

components provide significant potential for weight reduction of a car body. Presence of
metallic surface coatings, used for corrosion protection, such as Zn and Al-Si, combined
with a high strength of AHSS cause a decrease in joinability, thus creating a challenge to
produce resistance spot welds meeting commercial quality requirements. Welding with
longer welding times is recommended for joining AHSS in several standards and is some-
times used to improve spot weld joinability during manufacturing. Welding with longer
times, however, has been connected to a localized softening at the fusion boundary of spot
welds, referred to as halo ring. Present work gives an overvirew of halo appearances in the
literature and sums-up the research effort of the recent years to study the mechanisms
reslponsible for the halo formation in spot welds of AHSS.
It has been shown that the halo is caused by local decarburization at the fusion
boundary if nugget growth saturates early in the welding process and the fusion boundary
remains stationary, allowing for enough time for carbon diffusion to occure. Welds with
severe softening at the fusion boundary tend to fracture along the softened region, some-
times leading to reduced of energy absorption in the fracture process. Promoting nugget
growth at the end of the welding process has been shown to mitigate decarburisatio and
reduce softening, giving a pathway to halo elimination.
Keywords: Resistance spot welding · HAZ softening · Metallurgy · Advanced

high strength steels · Decarburization · Fracture
The use of advanced high-strength steels (AHSS) in body-in-white (BiW) components

provides significant potential for weight reduction in car body design [1]. High strength
and stiffness of the components along with the presence of surface coatings (Zn and
Al-Si) used for corrosion protection lead to a reduction of joinability and may impose
a challenge to produce resistance spot welds meeting commercial quality requirements.
When joining these components with RSW, the use of prolonged welding times was
shown to be a viable strategy to increase the process window [2]. Welding time prolon-
gation is also one of the methods used to enhance joinability used in adaptive quality
control systems [3, 4].

https://doi.org/10.1007/978-3-031-17425-4_9
48 O. Sherepenko et al.
For welds with prolonged welding times, the presence of a softened region at the
fusion boundary (FB) with a hardness drop of up to 300 HV has been reported by
numerous authors [5–16]. When etched with Nital, the softened region at the FB appears
as a bright ring around the nugget in the micrographs (see Fig. 1) and thus was referred
to as “halo” [15]. From the literature, there is evidence supporting the presence of halo
in spot welds on hot-stamped and dual-phase AHSS. In some cases, its presence is
mentioned by the authors [7–12, 15, 17], in others, it is disregarded [18–20], although
it can be clearly seen in micrographs and sometimes in hardness profiles.
The earliest mentions of the softened region go back to 2004 when Sierlinger et al.
[21] described a softened region at the fusion boundary of DP steel welds and suggested
that it might be caused by the formation of ferrite close to the FB.
Quiao et al. [7] mentioned the presence of FB softening in alternating current (AC)
spot welds of Al-Si coated hot-stamped steel (welding time 20 cycles / 318 ms) without
discussing its possible causes. Raath et al. [5, 12] showed optical micrographs with a
clear presence of halo in welds on hot-stamped 22MnB5 steel made using both standard
spot welding, and RSW adaptive quality system using prolonged welding times of more
than 700 ms [5]. The reported hardness values at the FB were as low as 300 HV as
compared to the surrounding material (FZ and CGHAZ), with the hardness of ~550 HV
and ~490 HV, respectively. The authors suggested that a softer phase with a lower
tetragonality as compared to martensite is present might be present at the fusion bound-
ary. They claimed this phase to be tempered martensite, although no discussion on how
the tempering has occurred at the fusion boundary of the spot weld, heated way above
AC3, and then quenched. The same argument is made by Russo et al. in [6] for dissimilar
welds on Zn-coated transformation-induced plasticity (TRIP) and dual-phase (DP)
steels.
Fig. 1. Halo in a spot weld on a 1.2 mm hot-stamped steel with Al-Si coating (1 pulse, weld time
500 ms), etched with 2% Nital for 2 seconds followed by immediate etching by 12% aqueous sodium
metabisulfite solution for 35 seconds.
Transient Softening at the Fusion Boundary of Spot Welds in Advanced High-Strength… 49
Li et al. [16] showed the presence of halo in spot welds performed on DP steel, with
application of external magnetic field. They suggested delta ferrite formation at the FB
to be responsible for the softening, along with the decarburization of the solid metal due
to carbon diffusion toward the liquid. They concluded that a more severe halo is caused
by more pronounced decarburization of the solid material at the fusion boundary due to
the liquid flow in the external magnetic field. Delta ferrite formation at the fusion
boundary was also hypothesized in [13] and [11]; however, no experimental evidence is
provided to support this.
Previous investigations by the authors of this manuscript suggest the following for-
mation mechanism of the halo. When welding using a long weld time, heat generation,
from Joule heating, and heat extraction by the water-cooled copper electrodes, will
reach equilibrium, resulting in a halt to nugget growth. Otherwise, at long times, a con-
dition exists where a static molten nugget is contained within the unmelted sheets. For
a double-pulse weld on a 1.5 mm hot-stamped steel, it was shown that the nugget arrest
happened in the last pulse for approximately 170 ms [9]. The stable position of the
liquid/solid boundary leads to carbon diffusion from the solid metal toward the liquid.
These results were obtained by numerical simulation using a phase field model and
measurements of carbon content using secondary mass ion spectroscopy [9]. The delta
ferrite which exists at the fusion boundary during welding, transforms into austenite
and subsequently into low-carbon martensite when cooled at high rates during resis-
tance spot welding. The resulting formation of carbon-depleted martensite results in a
hardness differential between that of the fusion boundary, with a hardness of 430 HV,
and the fusion zone (FZ), with a hardness between 500 HV and 550 HV [9]. An even
more severe hardness drop was observed in the halo of the same material with a longer
weld time of 750 ms (360 HV in the halo as compared to 510 HV in the FZ). Recent
investigations of halo formation in 3G AHSS [10] have shown that segregation of Mn
and Si happens along with C redistribution at the fusion boundary. The effect of chemi-
cal composition in the redistribution of these alloying elements at the FB and the result-
ing impact on the mechanical properties of welds is not fully understood and will be
subject to further studies.
Understanding that the halt of nugget growth leads to halo formation opens a path-
way to its effective suppression. As has been shown in [8], halo does not form when
nugget is actively growing (very short weld time of190 ms). Although a softer and
wider halo was observed at longer welding times, the halo formation will begin at the
time point when the nugget growth saturates up to a halt. This point depends on the
material chemistry, its thermophysical properties, and surface condition. It can be sug-
gested that early melting of sheet–sheet interface (e.g., happening in materials with
high contact resistances and high-resistive inclusions or coatings at the surface [22]),
the halt in nugget growth, might promote halo formation at shorter weld times.
The investigations by Sherepenko et al. [23] have shown the influence of surface
conditions on halo formation in 22MnB5. Figure 2 shows the results of high-density
microhardness maps (HV, 50 gf, indent spacing 45 μm) in the first quadrant of cross-
sections of spot welds, performed on sheets hot stamped with different heat treatments
(HT) 700 °C, 800 °C, and 930 °C with a furnace time of 6 minutes. The welds were
performed with short weld time ST (1 pulse, 190 ms), standard welding time according
to SEP 1220-2 [24] SE (1 pulse, 380 ms), and a long weld time LT (2 pulses without
Fig. 2. Influence of heat treatment conditions of Al-Si-coated hot-stamped steel on halo growth in spot
welds [23].
pause in between 350 and 400 ms). In all cases, the welding current was adjusted to
receive an equal nugget size for all HT conditions. As discussed in [25], the Al-Si coat-
ing of hot-stamped parts becomes more resistive with higher furnace temperatures and
longer furnace times. This happens due to interdiffusion between the Al-Si coating and
steel substrate and the formation of highly resistive layers in the coating (resistivity up
to 600% higher than the steel substrate [25]). As can be clearly seen in Fig. 2, for welds
on sheets heat treated at 700 °C, thus having lower surface layer resistivity, no pro-
nounced halo can be seen even at long weld times. No pronounced halo can also be seen
in welds on sheets heat treated at 800 °C welded with ST and SE, halo emerges only for
the LT welds. In the 930 °C HT condition, halo emerges already at the SE weld time
(380 ms).
Such difference in halo development can be explained by the influence of resistive
coating on nugget growth. At lower HT temperatures, the surface coatings are less
resistive, leading to slow heat introduction in the welding process thus leading to a
slower and more continuous nugget growth over the welding time. For higher HT tem-
peratures with the presence of more resistive coating, the nugget starts to grow earlier
and comes to halt earlier in the welding process. The fusion boundary remains station-
ary over a longer time, resulting in a more severe decarburization, which causes a more
pronounced softening at the FB.
As has been shown in [2], reducing weld time might lead to a reduction of the pro-
cess window (welding current range). The choice of a short weld time to mitigate halo
and still have a sufficient process window imposes a challenge and makes it difficult to
apply adaptive spot weld quality systems relying on weld time prolongation. A mitiga-
tion strategy, proposed in [10], using a short and high welding pulse at the end of the
weld time has shown promising results in halo mitigation for 3G AHSS. An application
of such a pulse led to a rapid increase of nugget diameter before the end of the welding
Transient Softening at the Fusion Boundary of Spot Welds in Advanced High-Strength… 51
process. The reheating of the material close to the fusion boundary allowed to redistrib-
ute the carbon thus alleviating the halo.
The influence of the halo on joint mechanical properties in the literature is mixed.
Some studies have shown that the halo does not impact CTS in welds made of hot-
stamped steel [8, 11, 15]. However, a recent study [10] showed that the presence of the
halo reduced CTS strength for welds in 3G AHSS by up to 33%. In all cases, the pres-
ence of halo affects joint fracture mode and energy absorption of welds in cross-tension
test (CTT); a decrease in failure energy up to 110% in Q&P 980 steel has been reported
in [9]. In welds that have a halo, fracture proceeds along the softened region at the
fusion boundary; this has been seen in experimental CTT and has been replicated in
numerical simulations of fracture process with cross-tension loading [8]. Such failure
path is responsible for reduced energy absorption. As a comparison, a full interfacial
fracture and partial interfacial fracture through the weld nugget is observed in halo-free
welds. Mitigating halo using the short post-pulse technique [10] led to a significant
increase in the strength and energy absorption of spot welds.
It has been shown that softening at the fusion boundary of spot welds “halo” happens
in a wide variety of automotive steels. Its formation was linked to carbon diffusion from
the sloid metal into the liquid across the solid/liquid interface, leading to the formation
of low-carbon martensite at the fusion boundary of welds. Halo has been shown to
reduce the energy absorption of spot welds in hot-stamped steels and negatively impact
both energy absorption and strength in 3G AHSS. The proposed mitigation strategy by
applying a short and high post-pulse has shown a pathway to halo alleviation. However,
a better understanding of how thermophysical properties of steels and surface condi-
tions of welded parts influence the halt of nugget growth will help define the starting
point of halo formation for diverse welding tasks and thus will allow for better tailoring
of halo mitigation strategies to the requirements of an assembly line. Understanding the
connection of halo formation with the material chemistry, primarily Mn and Si diffu-
sion during welding and subsequent paint baking will help develop effective halo miti-
gation strategies for 3G AHSS. It will also contribute to developing steels with lower
susceptibility to FB softening for automotive applications.
References
1. Matlock DK, Speer JG. Processing opportunities for new advanced high-strength sheet steels.
Mater Manuf Process. 2010;25:7–13. https://doi.org/10.1080/10426910903158272.
2. Sherepenko O, Holtschke N, Jüttner S. Microstructural imperfections in the HAZ of resistance
spot welded ultra high strength steel 22MnB5 and their impact on joint fracture. In: 5th interna-
tional conference on steels in cars and trucks : proceedings. Amsterdam: Germany Steel Institute
VDEh; 2017.
3. Sawanishi C, Okita Y, Matsuda H, Ikeda R. Development of resistance spot welding technology
applying multi-stage adaptive control for narrow pitch spot welding. Weld Int. 2019;33:17–29.
https://doi.org/10.1080/09507116.2020.1860489.
4. Cho YJ, Chang IS, Lee HB. Advanced resistance spot welding technologies: new machine, adap-
tive control and FEM simulation. In: Advanced welding and micro joining / packaging for the 21st
century. Trans Tech Publications Ltd; 2008. p. 367–70.
5. Raath ND, Norman D, McGregor I, et al. Characterization of loading responses and failure loci of
a boron steel spot weld. Metall Mater Trans A Phys Metall Mater Sci. 2018;49:1536–51. https://
doi.org/10.1007/s11661-018-4502-x.
6. Russo Spena P, de Maddis M, D’Antonio G, Lombardi F. Weldability and monitoring of resistance
spot welding of Q&P and TRIP steels. Metals (Basel). 2016;6 https://doi.org/10.3390/met6110270.
7. Qiao Z, Li H, Li L, et al. Microstructure and properties of spot welded joints of hot-stamped
ultra-high strength steel used for automotive body structures. Metals (Basel). 2019;9 https://doi.
org/10.3390/met9030285.
8. Sherepenko O, Mohamadizadeh A, Zvorykina A, et al. Determination of resistance spot weld fail-
ure path in ultra-high-strength press-hardened steel by control of fusion boundary transient soften-
ing. J Mater Sci. 2021;56:14287–97. https://doi.org/10.1007/s10853-021-06165-w.
9. Sherepenko O, Kazemi O, Rosemann P, et al. Transient softening at the fusion boundary of resis-
tance spot welds: a phase field simulation and experimental investigations for al–si-coated 22mnb5.
Metals (Basel). 2020;10 https://doi.org/10.3390/met10010010.
10. Ramachandran DC, Figueredo B, Sherepenko O, et al. A study on improving the mechanical per-
formance by controlling the halo ring in the Q&P 980 steel resistance spot welds. J Manuf Process.
2022;75:320–30. https://doi.org/10.1016/j.jmapro.2022.01.019.
11. Sherepenko O, Jüttner S. Transient softening at the fusion boundary in resistance spot welded ultra-
high strengths steel 22MnB5 and its impact on fracture processes. Weld World. 2019;63:151–9.
https://doi.org/10.1007/s40194-018-0633-3.
12. Raath ND, Norman D, McGregor I, et al. Effect of weld schedule on the residual stress distribution
of boron steel spot welds. Metall Mater Trans A Phys Metall Mater Sci. 2017;48:2900–14. https://
doi.org/10.1007/s11661-017-4079-9.
13. Zhao Y, Zhang Y, Lai X. Analysis of fracture modes of resistance spot welded hot-stamped boron
steel. Metals (Basel). 2018;8:1–15. https://doi.org/10.3390/met8100764.
14. Lu Y, Peer A, Abke T, et al. Subcritical heat affected zone softening in hot-stamped boron
steel during resistance spot welding. Mater Des. 2018;155:170–84. https://doi.org/10.1016/j.
matdes.2018.05.067.
15. Mohamadizadeh A, Biro E, Worswick M. Shear band formation at the fusion boundary and failure
behaviour of resistance spot welds in ultra-high-strength hot-stamped steel. Sci Technol Weld Join.
2020;25:556–63. https://doi.org/10.1080/13621718.2020.1773057.
16. Li YB, Li DL, David SA, et al. Microstructures of magnetically assisted dual-phase steel resis-
tance spot welds. Sci Technol Weld Join. 2016;21:555–63. https://doi.org/10.1080/13621718.201
6.1141493.
17. Schuster L, Sommer S. Characterization and modelling of soft zones around spot welds in high
strength steels. In: 5th international conference on steels in cars and trucks : proceedings; 2017.
18. Huin T, Dancette S, Fabrègue D, Dupuy T. Investigation of the failure of advanced high strength
steels heterogeneous spot welds. Metals (Basel). 2016;6 https://doi.org/10.3390/met6050111.
19. Gaul H, Brauser S, Weber G, Rethmeier M. Methods to obtain weld discontinuities in spot-
welded joints made of advanced high-strength steels. Weld World. 2011;55:99–106. https://doi.
org/10.1007/BF03321547.
20. Wohner M, Mitzschke N, Jüttner S. Resistance spot welding with variable electrode force-devel-
opment and benefit of a force profile to extend the weldability of 22MnB5+AS150. https://doi.
org/10.1007/s40194-020-01001-2/Published.
21. Sierlinger R, Szinyur S, Ritsche S. Optimising properties of spot welded joints on multiphase steels
(in German). In: Treffpunkt Widerstandsschweißen. Vorträge der gleichnamigen Sondertagung in
Duisburg am 26. und 27. Mai 2004. DVS Media GmbH. Duisburg; 2004.
22. Han X, DiGiovanni C, McDermid J, et al. Effect of internal oxidation on the weldability of CMnSi
steels. Weld World. 2019;63:1633–9. https://doi.org/10.1007/s40194-019-00798-x.
23. Sherepenko O, Schreiber V, Schischin I, et al. Untersuchungen zum Einfluss des Oberflächen- und
Werkstoffzustandes auf die Widerstandspunktschweißbarkeit formgehärteter Bleche. 2019;127
24. SEP 1220-2:2011-08 Testing and documentation guideline for the joinability of thin sheet of steel -
part 2: resistance spot welding.
25. Sherepenko O, Schreiber V, Schischin I, et al. Influence of surface layers on resistance spot join-
ability of partially hardened steel 22MnB5 with aluminum-silicon and zinc coatings. Weld World.
2020;64:755–71. https://doi.org/10.1007/s40194-020-00864-9.
Synthesis of Fast-Weathering Nutrient-Doped Silicate
Minerals and Their Prospect as a Negative Emissions
Technology
Asif Ali1 (*), Yi Wai Chiang1, and Rafael M. Santos1
1
School of Engineering, University of Guelph, Guelph, ON, Canada
aali38@uoguelph.ca
Abstract. Enhanced weathering of silicate minerals is a negative emissions technology

that captures atmospheric CO2 via the formation of carbonates. When certain silicate
rocks are mixed with agricultural soils, improved agricultural yields are witnessed, imply-
ing that the benefits are twofold. Natural silicates, such as basalt and olivine, weather too
slowly in basic soils and regions with low rainfall rates and cold temperatures, opening
the opportunity for faster weathering synthetic silicates to be produced via carbon-neutral
pathways. The minerals we are synthesizing for this purpose include calcium silicates
(larnite (β-C2S), calcio-olivine (γ- C2S), cuspidine, and pseudowollastonite), and calcium-
magnesium silicates (åkermanite, bredigite, and merwinite). To boost their soil fertilizing
ability, thus increasing their agronomic value, these silicate minerals will be doped with
nutrients such as K, P, and Se. The manufacturing process of these minerals is being
studied to achieve even higher weathering rates by creating dislocations and preferential
reaction sites through milling process and by controlling the cooling time for regulating
polymorphicity and crystallinity. The successful synthesis of these silicate minerals at
laboratory scale and their performance in capturing atmospheric carbon at higher rate will
pave further ways for industrial production and large-scale implementation.
Keywords: Fast weathering · Mineral synthesis · Carbon capture and sequestra-

tion · Silicate minerals · Negative emissions technology.
1 Introduction
The globe is facing several challenges owing to the hazardous effects of greenhouse gas
emissions including but not limited to global warming, ozone layer depletion, etc. The
reduction in dangerous pollutant emissions would not alone suffice the efforts to achieve
UN climate agreement goals [1]. Consequently, there is a serious need to capture and
sequester CO2 from the atmosphere. The subsequent technologies are known as nega-
tive emissions technologies (NET), in which enhanced silicate weathering (ESW) is
demonstrating significant potential. Silicate minerals naturally neutralize CO2 in the
atmosphere, but their weathering rate is quite slow. ESW aims at artificially boosting
the weathering rate of minerals for capturing CO2 from the atmosphere at an effective
pace [2]. Some of the recent studies are reported below, which used alkaline silicate
minerals for accumulating inorganic carbon in soils.

https://doi.org/10.1007/978-3-031-17425-4_10
54 A. Ali et al.
Renforth et al. studied the role of artificial soils in carbon precipitation, for assessing
their behavior as atmospheric CO2 sink and found that the inorganic pathway is three
times more effective in carbonation than the organic one, witnessing the significant role
of silicate mineral synthesis in carbon capture [3]. Washbourne et al. investigated the
atmospheric CO2 removal through urban soils and found that the Ca- and Mg-based sili-
cate minerals resulting from concrete demolition weather rapidly [4]. Manning et al.
added finely grounded basalt and dolerite fines from basalt quarries blended with com-
post to the soil and found the SIC sequester rate of 36.7 t CO2 ha−1 for a depth of 1 meter
[5]. All these findings reinforced that the inorganic carbon sequestration pathway pos-
sesses considerable potential for battling against climate change. Silicates with fast
carbonation performance can be synthesized in net-negative pathway and mixed with
agricultural soils for gaining both environmental and agricultural benefits.
2 Method
The silicate minerals with high weathering rates are rare in nature. The minerals will be
synthesized using solid-state sintering technique in the laboratory. The synthesis meth-
odology and process parameters such as sintering duration, temperature, and cooling
trajectories have been selected based on the available literature [6, 7] with minor adap-
tations. The synthesis conditions for the target minerals are presented in Table 1.
The stoichiometric quantities of analytical grade of the constituent oxides (such as
MgO, SiO2, CaO, and Fe2O3) will be used for synthesis of respective minerals. CaF2
would also be used for the synthesis of cuspidine. The thermal decomposition of pure
CaCO3 at a temperature of 900 °C will produce CaO, and the possible pre-existing car-
bonates and hydroxides would be removed by refreshing MgO. For handling the insta-
bility of 𝛽-polymorph of C2S, the mixture would be added with B2O3 (0.4% by weight).
The minerals will be synthesized in batches of 50 g, and two batches of each mineral will
be produced, for obtaining sufficient material for characterization and experiments.
A CO2 incubator (Binder C170, Tuttlingen Germany) will be used for conducting
carbonation and accelerated weathering under static conditions. For dynamic condi-
tions, a stirred slurry process will be used for conducting carbonation and accelerated
weathering. For pressurized slurry carbonation, a stirred batch reactor with bolted clo-
sure pressure vessel (Parker Autoclave Engineers, Erie, PA, USA) will be used for
Table 1. The mineral synthesis conditions for experiments.

Mineral Temperature (°C)/Time (hours) Milling time (hours) Cooling method
β-C2S 1450/3 2 Rapid cooling in air flow
γ-C2S 1400/48 2 5 °C/min
Wollastonite 1200/4 2 5 °C/min
Cuspidine 1100/20 (Ar atmosphere) 2 5 °C/min
Akermanite 1300/24 2 5 °C/min
Bredigite 1200/1 10 5 °C/min
Merwinite 1500/20 2 1 °C/min
Srebrodolskite 800/1 + 1100/36 2 5 °C/min
Synthesis of Fast-Weathering Nutrient-Doped Silicate Minerals and Their Prospect… 55
attaining higher pressure and temperature. The mixture will be stirred at 1000 rpm for
4 hours and after completion, the vessel will be depressurized and cooled. For these
carbonation processes, the carbonated slurry will be filtered using filter paper (Whatman
2V 150 mm, Maidstone UK) and the recovered solids will be dried in an oven. Analytical
tests such as pH test (using a pH meter (Fisher Scientific (Waltham, MA, USA) Accumet
AE150)), furnace test, calcimeter test (using a calcimeter (Eijkelkamp 08.53, Giesbeek,
The Netherlands)), and XRD analysis (using XRD equipment (Malvern Panalytical,
(Almelo, The Netherlands) Empyrean XRD Diffractometer)) will be conducted for
characterization of the samples.
3 Findings
One of the potential solutions to carbon emissions problem is the use of basalt rock,
which is abundant on earth. It also contains plant-nutrient elements which can be used
to fertilize agricultural lands, as well as for capturing CO2 from environment. One
recent study investigated the effects of mineralogy, physical, and chemical properties of
basalts on their performance as carbon capture and crop-fertilizing reagents through
enhanced weathering [8]. They characterized the surface area and mineralogy of six
commercial basalts and used their characteristics in reactive transport modelling for
cropland soil. They considered basalt baseline application to the agricultural field as
50-ton ha−1 and the results forecasted a CO2 removal range of around 1.3- to 8.5-ton
CO2 ha−1 after 15 years [8].
Other studies indicated that weathering of basalt further slows down with cold tem-
peratures, basic soils, and dry regions. According to them, there is an uncertainty in the
rates of CO2 removal via enhanced weathering. The sequestration rate of basalt rocks
ranges from 200 to 800 kg CO2 sequestered per ton of rock, while it is greater than 1 ton
CO2 sequestered per ton rock of other Ca- and Mg-rich silicate minerals [9–12].
Basalt rocks can serve the purpose in regions having acidic soils and tropical cli-
mate. The role of silicate minerals via enhanced weathering cannot be ignored for their
carbon capture and sequestration potential application in other parts of the world.
Comparing the difference in CO2 sequestering rate, Haque et al. amended Canadian
agricultural soils with wollastonite and found a CO2 sequestering rate of around 24.3-
ton ha−1 month−1 via enhanced weathering [13]. This indicates improved weathering
rate of capturing CO2 from environment. Bodor et al. studied the susceptibility of min-
erals present in steel slags toward carbonation and found the prominent role of bredigite
conversion, as shown in Fig. 1 [6]. This provides an insight on the carbonation affinity
of various silicate minerals.
Chai et al. used in situ XRD for understanding the reaction pathways of synthetic
mineral carbonation [14]. It allowed quantification of the carbonation conversion rates
for individual silicate phases, as shown in Fig. 2. These studies have motivated the syn-
thesis and doping of various synthetic silicates with plant nutrients and testing their faster
rate of carbonation and agricultural capabilities. The novelty of this work lies in studying
the manufacturing process of synthetic silicates such, especially the role of milling and
cooling operations on the carbonation activity of minerals. The doping of these minerals
56 A. Ali et al.
IC.7d AC.15min AC.4hr Max,Ca content

100%
90% 95%
86% 87% 87%

80%
CO2 uptake (% maximal)
70% 74% 73% 73%

67% 66%
60%
59% 60%
50% 55%
50%49%
40% 45%
30% 33%
27%
20% 22%
19%
10% 12%
9%
0%
ite ite in
e 2S S ite ite
an ig id -C C2 in lsk
rm ed p a a- er
w
do
e Br us Bet m M ro
Ak C am eb
G Sr
Fig. 1. TGA analysis of synthetic carbonated minerals based on CO2 uptake [6]. (The re-use permis-
sion from the publisher is pending).
a) b)
14
250 250
12
200 200
10
Intensity (counts)
Time (hrs)
150 8 150
6
100
14 100
12 4
10
50
8 50 2
6
4 42 44
36 38 40 0 0
Time (hrs) 2 32 34
26 28 30 0 20 22 24 26 28 30 32 34 36 38 40 42 44
0 20 22 24 Deg 2-theta Deg 2-theta
Fig. 2. (a) 3D HXRD diffractogram of synthetic bredigite carbonated in wet CO2 at 6 bar and 70 °C
with 40 μL of water (b) A top-down 2D view of HXRD image [14]. (Re-used with permission from
Elsevier (5260430791940)).
with agricultural nutrients is another innovative approach of the present work, where suc-
cessful results can open ways for industrial production and large-scale implementation.
4 Conclusion
Mineral carbonation can serve dominant role in carbon capture and sequestration as a
negative emissions technology. There are several uncertainties about the carbonation of
minerals since weathering can take several years. Basalt is one of the abundant rocks on
Synthesis of Fast-Weathering Nutrient-Doped Silicate Minerals and Their Prospect… 57
earth and mixing its powder with agricultural soils can be a profitable solution to global
warming problem. However, basalt weathers very slowly especially in cold tempera-
tures, basic soils, and nontropical regions. The present research aims at synthesizing
and gaging the fast weathering of silicate minerals. The novelty of this work lies in
investigating the role of milling and cooling operations during the synthesis of miner-
als, on their carbonation rates. Also, another innovative idea is the doping of these
minerals with agricultural nutrients. This would result in double benefits, i.e., atmo-
spheric carbon sequestration and agricultural fertilizing. Upon successful evaluation,
this research can pave ways for large-scale production and implementation of synthetic
silicates.
References
1. Climate Change 2022: Impacts, Adaptability and Vulnerability. https://www.ipcc.ch/report/

ar6/wg2/.
2. Smith P, Davis SJ, Creutzig F, et al. Biophysical and economic limits to negative CO2 emissions.
Nature. Climate Change. 2015;6:42–50.
3. Schuiling RD, Tickell O. Enhanced weathering of olivine to capture CO2. Appl Geochem.
2010;12:510–19.
4. Washbourne CL, Lopez-Capel E, Renforth P, et al. Rapid removal of atmospheric CO2 by urban
soils. Environ Sci Technol. 2015;49(9):5434–40.
5. Manning DAC, Renforth P, Lopez-Capel E, et al. Carbonate precipitation in artificial soils pro-
duced from basaltic quarry fines and composts: an opportunity for passive carbon sequestration.
Int J Greenhouse Gas Control. 2013;17:309–17.
6. Bodor M, Santos RM, Kriskova L, et al. Susceptibility of mineral phases of steel slags towards car-
bonation: mineralogical, morphological and chemical assessment. Eur J Miner. 2013;25:533–49.
7. Hafezi-Ardakani M, Moztarzadeh F, Rabiee M, et al. Synthesis and characterization of nanocrys-
talline merwinite (Ca3Mg(SiO4)2) via sol-gel method. Ceram Int. 2011;37(1):175–80.
8. Lewis AL, Sarkar B, Wade P, et al. Effects of mineralogy, chemistry and physical properties of
basalts on carbon capture potential and plant-nutrient element release via enhanced weathering.
Appl Geochem. 2021;132:105023.
9. Kantola IB, Masters MD, Beerling DJ, et al. Potential of global croplands and bioenergy
crops for climate change mitigation through deployment for enhanced weathering. Biol Lett.
2017;13(4):20160714.
10. Schuiling RD, Krijgsman P. Enhanced weathering: an effective and cheap tool to sequester CO2.
Climate Change. 2006;74:349–54.
11. Renforth P. The potential of enhanced weathering in the UK. Int J Greenhouse Gas Control.
2012;10:229–43.
12. Moosdorf N, Renforth P, Hartmann J. Carbon dioxide efficiency of terrestrial enhanced weather-
ing. Environ Sci Technol. 2014;48(9):4809–16.
13. Haque F, Santos RM, Dutta A, et al. Co-benefits of wollastonite weathering in agriculture: CO2
sequestration and promoted plant growth. ACS. Omega. 2019;4:1425–33.
14. Chai EY, Miller QRS, Schaef HT, et al. Pressurized in situ X-ray diffraction insights into super/
subcritical carbonation reaction pathways of steelmaking slags and constituent silicate minerals. J
Supercrit Fluids. 2021;171:105191.
Direct Laser Writing of Copper/Copper Oxide
Patterns for Emerging Roles in Advanced Electronics
Joshua Jones1 (*) and Peng Peng1
1
Center for Advanced Materials Joining (CAMJ), Department of Mechanical and Mechatronics
Engineering, University of Waterloo, Waterloo, ON, Canada
jt2jones@uwaterloo.ca
Abstract. There has been increase in interest in flexible electronics for roles in applica-
tions such as wearable electronics, and solar technology. Current technologies for fabri-
cating flexible electronics tend to rely on costly nanomaterials as raw materials. This
paper presents an alternative processing method that circumvents the need for pre-
fabricated nanomaterials, while still achieving comparable electrical properties. Direct
Laser Writing (DLW) is a new manufacturing technique where an aqueous copper ion
film is applied to a flexible substrate and then selectively irradiated with focused laser
energy to produce a resulting structure of interconnected copper nanoparticles in a single
step. During DLW laser energy simultaneously reduces the copper ions present in the
solution to form nanoparticles, and subsequently sinters the nanoparticles into a continu-
ous structure. In addition to being low cost DLW also benefits from being suitable at room
temperature and not requiring or producing any environmentally harmful materials.
Through controlling laser parameters the chemical composition of the copper can be
controlled, resulting in a high degree of control over the electrical properties of the result-
ing patterns. We demonstrate in this paper the ability to achieve flexible patterns with
conductive and semi-conductive electrical properties. Finally, we intend to demon-
strate the unique capabilities of DLW by showcasing some of the flexible electronic
devices we have been able to fabricate using DLW. Namely a memresistive Copper /
Copper Oxide junction.
Keywords: Flexible electronics · Direct laser writing · Copper nanomaterials ·

Laser reduction · Nanofabrication · Resistive switching · Memristor
1 Introduction
There has been a growing interest in flexible electronics over the past several years [1].
The industries of medicine and healthcare [2], consumer electronics [3], and smart
textiles [4] are among the key industries leading the push for next-generation flexible
electronic devices. Currently, most of the fabrication processes capable of producing
high-quality flexible electronic components rely on costly raw materials such as pre-
cious metals or nanoparticles [5]. As well, most of the fabrication methods utilize com-
plex multi-step procedures that offer low throughput [6]. These two factors of prohibitive
cost and low production rate are currently two of the largest hurdles preventing the

https://doi.org/10.1007/978-3-031-17425-4_11
60 J. Jones and P. Peng
advancement of flexible electronics. As such, there is a desire for a fabrication method

that can produce high-quality components with inexpensive raw materials and high
production capabilities.
Direct Laser Writing (DLW) is a novel fabrication technique wherein a flexible sub-
strate is coated in a thin metal ion film and irradiated with focused laser energy. The
laser energy reduces the metal ions into metal nanoparticles, and then instantaneously
interconnects those particles to form continuous flexible conductive patterns. To date,
DLW has been demonstrated to be capable of fabricating high-quality flexible elec-
tronic components such as flexible electrodes [5] and sensors [7]. In this work, we will
be presenting a DLW procedure capable of rapidly producing flexible copper patterns.
Furthermore, we will demonstrate the ability to precisely control the electrical proper-
ties of the deposited copper patterns through controlling key laser parameters, of laser
speed, power, and defocus. The potential of DLW as a method of producing flexible
electronic components suitable for next-generation applications is then demonstrated
through the production of a novel flexible memristive interdigital electrode
component.
2 Experimental
A copper ion solution comprised of copper nitrate (Cu(NO3)2), polyethylene glycol

(PEG), polyvinyl pyrrolidone (PVP), and deionized water was drop casted onto a flex-
ible polycarbonate sheet and dried to remove excess moisture. The copper ion solution
was then irradiated with focused laser energy to deposit copper patterns onto the PC
substrate. The untreated regions of the copper ion film were then rinsed using deionized
water to leave only the copper patterns on the substrate. The conductivity of the printed
patterns was measured in order to investigate the effects of changing laser parameters
on electrical properties. The chemistry and microstructure of the deposited patterns
were also investigated in order to better relate the electrical properties of the patterns to
key chemical and microstructural changes.
Patterns at varying laser speeds and powers were deposited, and the resistivity of the
patterns was measured. Immediately it is clear that the chemical composition of the
deposited patterns varies with increasing laser power density, as shown in Fig. 1a. At
low laser power, the patterns exhibit a dark grey color and at increasing laser powers,
the patterns shift to a red-brown color. The red-brown is associated with metallic copper
and the grey color is indicative of copper oxide. The change in chemical composition of
the pattern with increasing laser power is a result of the reduction mechanism that
enables the reduction of the copper ion film to metallic copper, through the photodeg-
radation of PVP [5]. During laser irradiation, the lactam side ring group of PVP mole-
cule is degraded through the assistance of the hydroxyl group contributed by the end
Direct Laser Writing of Copper/Copper Oxide Patterns for Emerging Roles in Advanced… 61
Fig. 1. (a) Copper patterns deposited at varying laser energies. (b) Reduction mechanism of copper
ions through the photodegradation of PEG and PVP. (c) Current–Voltage plots showing the resistivity
change in patterns deposited at varying laser energies.
groups of the PEG molecules. The lactam ring degrades into methylamine, amorphous
carbon, methylene, and formic acid. The formic acid is then able to react with the cop-
per ions in solution and reduce them from the +2 state to the +1 state, and then subse-
quently to the metallic state [7]. Figure 1b shows the reduction of copper ions through
the formation of formic acid from PVP. At high laser power treatment parameters, there
is enough photodegradation of the PVP molecules to generate an abundance of formic
acid. As a result, the majority of the copper ions are fully reduced from the 2+ state, to
the metallic state. At decreasing laser energies, there is less degradation of the PVP,
which results in less formic acid. Subsequently, there is not enough formic acid present
to fully reduce most of the copper ions, as such many copper ions are only partially
reduced to the 1+ state, and form Copper(I)Oxide. This trend of increasing Copper(I)
Oxide at decreasing laser energies is reflected in the conductivity of the patterns depos-
ited at lower laser energies. Figure 1c shows the resistivity of patterns printed with
varying laser power densities. Since metallic copper is a much better electrical conduc-
tor than Copper(I)Oxide, increasing amounts of Copper(I)Oxide result in dramatic
reductions in the conductivity of the deposited patterns. XRD analysis of the deposited
patterns revealed that at a laser energy of 1 watt roughly half of the copper in the pattern
was deposited as Copper(I)Oxide, at 1.5 watts Copper(I)Oxide content decreased to
about 15% and above 2 watts the Copper(I)Oxide content is below 10%.
The need for highly conductive flexible patterns for flexible electronics is self-
evident [1]. However, there is also growing interest into utilizing the properties of cop-
per oxides. Copper oxides have been shown to have outstanding catalytic properties [6],
which have made them ideal choices for use in sensing applications. As well, there is
growing interest in using copper oxide as a semiconductor. Copper(I)Oxide is a natural
p-type semiconductor, wherein oxygen vacancies act as electron receptors [8]. Copper(I)
Oxide has demonstrated resistive switching effect where it can be set to either a high or
low resistivity state through the application of a voltage bias. The resistive switching
observed in Copper(I)Oxide is a result of the reversible formation of a conductive fila-
ment through the migration of oxygen vacancies [9]. The resistive switching capabili-
ties of Copper(I)Oxide are the basis of a novel new electrical component known as a
memristor. Memristors show promise in many roles in advanced electronics such as
62 J. Jones and P. Peng
Fig. 2. (a) Flexible memristor on polycarbonate substrate, (b) schematic of the memristive junction,
and (c) I-V hysteresis loop of memristive junction.
memory units and logic gates [10]. The ability for DLW to deposit both copper and
Copper(I)Oxide patterns makes it an ideal process for the fabrication of memristive
components.
A novel memristive interdigital electrode component has been fabricated in a single
step using DLW, shown in Fig. 2a. By tuning laser energy parameters throughout the
laser writing procedure, different regions of the pattern exhibit different properties.
Copper-rich electrodes were printed with a laser setting of 4 W at 10 mm/s and copper(I)
oxide-rich interdigital regions were printed with a laser setting of 1.55 W at 5 mm/s. A
voltage bias was applied from 0 - +1 V, +1 V - −1 V, −1 V – 0. The current flowing
through the component was measured. Figure 2c shows the current–voltage plot of the
fabricated component. Upon applying +1 V, the resistance of the junction dropped to a
low resistance state (LRS). This indicates the formation of the conductive filament
linked one copper region to another. Applying −1 V returned the junction to the high
resistance state (HRS), indicating that the formation of the conductive filament has
been reversed.
4 Conclusions
In summary, a new novel procedure that can produce high-quality flexible electronic
components has been presented. Highly conductive patterns can be produced without
the need for expensive raw materials, and in a single-step process. These aspects of
DLW make it very relevant to the growing need for next-generation flexible electronics.
It was shown that the reduction of copper ions through the photodegradation of PVP
allows for a high degree of control over the chemical composition of the fabricated cop-
per structures. By leveraging this control over chemical composition, a novel memris-
tive junction was fabricated, and the current–voltage loop characteristics were
presented.
Acknowledgments We would like to acknowledge the support of NSERC research grants, for sup-
porting the work.
Direct Laser Writing of Copper/Copper Oxide Patterns for Emerging Roles in Advanced… 63
References
1. Logothetidis S. Handbook of flexible organic electronics: materials, manufacturing and applica-

tions. Waltham: Woodhead Pub; 2014.
2. Someya T, Sekitani T, Iba S, Kato Y, Kawaguchi H, Sakurai T. A large-area, flexible pressure sen-
sor matrix with organic field-effect transistors for artificial skin applications. Proc Natl Acad Sci.
2004;101(27):9966–70. https://doi.org/10.1073/pnas.0401918101.
3. Loher T, Seckel M, Pahl B, Bottcher L, Ostmann A, Reichl H. Highly integrated flexible electronic
circuits and modules. In: 2008 3rd international microsystems, packaging, assembly circuits tech-
nology conference; 2008, Oct. p. 86–9. https://doi.org/10.1109/IMPACT.2008.4783814.
4. Kim J, Lee T-I, Shin J, Kim T-S, Paik K. Bending properties of anisotropic conductive films assem-
bled chip-in-flex packages for wearable electronics applications. IEEE Trans Compon Packag
Manuf Technol. 2016:208–15. https://doi.org/10.1109/TCPMT.2015.2513062.
5. Peng P, et al. One-step selective laser patterning of copper/graphene flexible electrodes.
Nanotechnology. 2019;30(18):185301. https://doi.org/10.1088/1361-6528/aafe4c.
6. Zhou X, Guo W, Yao Y, Peng R, Peng P. Flexible nonenzymatic glucose sensing with one-step
laser-fabricated Cu2O/Cu porous structure. Adv Eng Mater. 2021;23(6):2100192. https://doi.
org/10.1002/adem.202100192.
7. Zhou X, Guo W, Fu J, Zhu Y, Huang Y, Peng P. Laser writing of Cu/Cu O integrated structure on
flexible substrate for humidity sensing. Appl Surf Sci. 2019;494:684–90. https://doi.org/10.1016/j.
apsusc.2019.07.159.
8. Sawa A. Resistive switching in transition metal oxides. Mater Today. 2008;11:28–36. https://doi.
org/10.1016/S1369-7021(08)70119-6.
9. Rehman S, Hur J-H, Kim D. Resistive switching in solution-processed copper oxide (Cu x O)
by stoichiometry tuning. J Phys Chem C. 2018;122(20):11076–85. https://doi.org/10.1021/acs.
jpcc.8b00432.
10. Liu H. Laser assisted ink-printing of copper oxide nanoplates for memory device. Mater Lett.
2020;261:4.
Open Pore Effect on Structural Adhesive Joining
Using an Epoxy Adhesive
M. M. Tiwari, S. Noormohammed, X. -G. Chen, and D. K. Sarkar1 (*)
1
Department of Applied Science, University of Québec at Chicoutimi, Aluminum Research
Center REGAL, Chicoutimi, QC, Canada
dsarkar@uqac.ca
Abstract. Considering the ecological challenges, a very mild concentrated salt instead
of acids in anodizing AA6061-T6 aluminium alloy surfaces was used to improve adhesive
bond strengths. The aluminium surfaces anodized in 0.1 molar disodium hydrogen phos-
phate (Na2HPO4) salt at a controlled temperature of 5 °C with an optimized current den-
sity of 30 mA·cm−2 for 60 minutes provided a higher surface roughness of 1.04 μm
compared to 0.33 μm on the as-received surface. Scanning electron microscopy (SEM)
images revealed the effective growth of porous oxide layers with wide pores (⁓0.60 μm).
Mechanical testing performed on single-lap shear specimens prepared by bonding the
aluminium surfaces using a 2-C epoxy adhesive showed an increased joint shear strength
of 21 MPa on salt anodized aluminium compared to only 10 MPa on the as-received
counterpart. Emergence of pores, when anodized the AA6061-T6 surface for 60 minutes,
increased the true bonding area and promoted the mechanical interlocking of aluminium/
epoxy bonding at room temperature. The salt anodization technique produced a uniform
porous surface to anchor the epoxy in the aluminium surface, proving as more eco-
friendly than the acid anodization technique in improving bond performance.
Keywords: Salt anodization · Structural adhesive · Aluminium · SEM · Epoxy

adhesive · Mechanical interlocking.
1 Introduction
Climate change policies focus more on energy efficiency as a central strategy for green-
house gas emissions reduction [1, 2]. To mitigate the impact of the materials on green-
house gases emissions is to improve material efficiency (light-weighting); reducing
yield losses in the manufacturing processes; minimizing the by-products; and using
products more intensely or at a higher capacity. Structural adhesive joining of light-
weight materials, namely aluminium (Al), has a considerable role in industrial applica-
tions, particularly in transportation industries, on CO2 emission reduction due to
reduced fuel consumption associated with lightweighting [3]. Structural adhesive bond-
ing is well known in aircraft industry since they provide greater structural efficiency
and performance due to low structural weight, large area bonding with uniform stress
distribution (compared to localized stresses in rivets), excellent bondability of similar

https://doi.org/10.1007/978-3-031-17425-4_12
66 M. M. Tiwari et al.
and dissimilar materials of different shapes and dimensions [4, 5]. Although huge
research, dedicated to the advancement in structural adhesive bonding, has been con-
ducted and documented in the literature [6, 7], limitations such as high cost, long cure
times, fatigue and durability, health and safety issues, availability and so on still pose
challenges in their wide usage in many application areas [8–10]. The strength and sta-
bility of the bond between the adhesive and the adherent surface are generally con-
trolled by surface preparation, which governs the chemical and geometrical features of
the surface favourable to the adhesion of adhesives [11, 12]. Controlling the surface
features of Al, therefore, becomes primordial in obtaining robust adhesive joints with
high mechanical strengths. Several methods have been used to modify the surface prop-
erties in terms of morphology and chemistry [13, 14]. Acid anodizing, among many, is
an electrochemical process that promotes the formation of a strong porous oxide layer
(anodic layer) on the surface of the metal, which not only helps protect the material
from atmospheric agents or their environment in general [15] but also adds various fac-
tors with regard to degradation of electrolyte. Although acid anodization is a bench-
mark treatment for Al in adhesive bonding, the use of acids as anodizing electrolyte
poses health and environmental risks due to the release of hazardous gases in addition
to requiring more prudence in manipulating acids [6, 16, 17].
Considering the ecological and safety challenges, we have, in this study, used an
environmentally friendly salt solution as anodizing electrolyte instead of standard acids
to anodize the Al surfaces prior to bonding with an epoxy adhesive. The effect of the
pore diameters of the anodized porous oxide layer on the adhesion strength of single lap
shear specimens (SLS) was investigated. The anodization-induced surface roughness
and the pore size on the resulting surfaces, characterized by SEM/EDX and roughness
measurements, presented a strong correlation with the mechanical strength of the
bonded joints.
2 Methods
AA6061-T6 Al sheets of 4″ × 1″ × 0.12″ were cleaned ultrasonically in a dilute soap

solution and dried at ambient conditions. The anodization process was carried out using
a voltage-controlled DC power supply at a constant current mode, by utilizing an opti-
mized current density (j) of 30 mA·cm−2. Clean and dry AA6061-T6 substrates con-
nected to the positive output of the power supply served as anodes. The process was
carried out in a 60-ml electrolyte prepared by mixing ⁓14 g of solid Na2HPO4 salt
diluted into 1 L of deionized water, at a controlled electrolytic bath temperature of
5 ± 1 °C for a duration of 15, 30, 45 and 60 minutes. The anodized samples were further
washed with deionized water and dried on a hot plate at 150 °C for 2 hours prior to
bonding with a two-component (2-C) epoxy adhesive. The prepared surfaces were
characterized for surface morphology and chemistry using scanning electron micros-
copy (SEM, JEOL JSM-6480), equipped with energy-dispersive X-ray spectroscopy
(EDX). Chemical analysis was also performed using attenuated total reflection-Fourier
transform infrared (ATR, Agilent Technologies, Cary 630 FTIR). The root mean square
Open Pore Effect on Structural Adhesive Joining Using an Epoxy Adhesive 67
(rms) surface roughness was measured and analysed using a MicroXAM-100 HR 3D

optical surface profilometer.
Single lap shear (SLS) specimens were prepared with the epoxy adhesive according
to the ASTM D1002 standard and allowed to cure according to the manufacturer’s rec-
ommendation prior to mechanical testing. The joint strengths of the SLS were deter-
mined using an INSTRON 8801 at a traction speed of 1.0 mm/minute. The rupture
modes of the failed specimen were visually analyzed.
The salt anodization process results in the formation of a porous oxide layer on the
aluminium surface, leading to a rough microstructure. Figure 1 shows the rms rough-
ness obtained on the salt anodized Al surfaces at an optimized current density (j) of
30 mA·cm−2 as a function of anodization time. The untreated Al surface presented an
rms value of as low as 0.33 μm. However, it is observed that the surface, once anodized,
presented an increase in their rms values. It is clear from Fig. 1 that the roughness
increased with the increase in anodization times and reached the maximum value of
1.04 μm after 60 minutes of anodization. This increase in roughness may be attributed
to the increasing growth of porous oxide layer with the pores of varied dimensions
contributing to the micro rough structure. The surface with anodized pores having such
a high roughness value can be expected to promote the mechanical anchoring of the
adhesive into the rough pores, thereby enhancing the interfacial bonds between the
adhesive and the surface.
The morphological and chemical analysis by SEM/EDX (Fig. 2a-c) corroborates
with the surface roughness measurements. Figure 2a, b shows the SEM images of the
1.2
1
rms roughness (µm)
0.8
0.6
0.4
0.2
0
0 15 30 45 60
Anodization time (min)
Fig. 1. Variation in rms roughness of anodized AA6061-T6 Al surfaces as a function of anodization

time.
(a) As-received (b) t = 60 min
5 µm 5 µm
0.4
Al t = 60 min t = 60 min
(d)
Intensity (counts)
0.3
Intensity (counts)
Al-O
0.2
0.1
O
(c)
0
2 4 6 8 10 2000 1500 1000 500
Energy (keV) Wavenumber (cm-1)
Fig. 2. SEM images of (a) as-received Al; (b) 60-minute salt-anodized Al, (c) EDX spectrum of (b)
and (d) ATR-FTIR spectra of (b).
surfaces of the as-received clean Al and the 60-minute anodized counterpart, respec-
tively. The anodized surface revealed the emergence of porous microstructures with
wide pores of diameter of ⁓0.60 μm as confirmed by Fig. 2b. A comparison of the SEM
images of all the surfaces studied (Figures not shown) showed the evolution and emer-
gence of microporous structures contributing to the enhancement of rms surface rough-
ness with increasing anodization time. The morphology of the surfaces, however,
presented a porous texture in all cases.
The chemical nature of the treated surfaces was analyzed by both EDX and ATR-IR
spectral analyses. The EDX spectrum (Fig. 2c) confirmed the presence of O, which can
be associated with the formation of the anodic aluminium oxide layer. The ratio of the
atomic percentage of O and Al on the anodized surface was found to be 1.37 ± 0.20,
close to the theoretical value of 1.5. Figure 2d depicts the corresponding ATR-FTIR
spectrum of the surface shown in Fig. 2b. A broad band at 490–1000 cm−1 is attributable
to the Al-O vibrations confirming the presence of an oxide of aluminium on the surface,
as well as corroborating with the EDX results shown in Fig. 2c [18]. These observations
indicate that the 60-minute salt-anodized aluminium surfaces have potential of provid-
ing high joint strength between the these surfaces and the adhesive, when bonded
together.
Mechanical testing was carried out on the SLS specimens prepared by bonding the
60-minute salt-anodized aluminium coupons with a 2-C epoxy adhesive and their fail-
ure modes were further examined. Figure 3a shows the SLS test results obtained on the
as-received and treated specimens. It is evident that the lap shear strengths of the treated
specimens were higher than that of the as-received Al (10 ± 1 MPa). A maximum lap
shear strength of 21 ± 1 MPa was obtained with specimens prepared with surfaces
anodized for 60 minutes. The increasing trend in the lap shear strengths was similar to
the values of rms surface roughness. Therefore, a linear fit of the SLS values with the
rms roughness was established, as shown in Fig. 3b. This behaviour can be attributable
to the microstructural evolution and the chemical changes occurring on the treated Al
surfaces with varied anodization times. Furthermore, the rupture modes of the failed
specimens, shown in Fig. 3c, presented 100% cohesive failure on specimens prepared
with surfaces treated for 60 minutes, indicating that the adhesion at the adhesive/surface
interface was strong enough to make the failure within the adhesive. All other speci-
mens showed 100% adhesive failure at the interface between the adhesive and the sur-
face, indicating that the porous texture and dimensions with lower surface roughness
values on the surfaces anodized for less than 60 minutes were not sufficient to obtain a
strong interfacial bond between the adhesive and the surface and thereby failing at the
interface.
Therefore, 60 minutes of anodization time was required to achieve strong mechani-
cal interlocking of the adhesive within the anodized pores. The strong oxide as evi-
denced by EDX measurements, and IR spectral analyses may be expected to play an
additional role of chemical enrichment of the surface with high surface energy, allow-
ing the adhesive thoroughly wet the surface.
30 30
A. F. = Adhesive failure
25 (a) C. F. = Cohesive failure 25
(b)
Shear lap strength (MPa)
Shear lap strength (MPa)
20 20
15 15
10 10
5 A. F. A. F. A. F. A. F. C. F. 5
0 0
0 15 30 45 60 0.2 0.4 0.6 0.8 1 1.2
Anodization time (min) rms roughness (µm)
(c)
Fig. 3. (a) Single lap shear (SLS) strengths vs. anodization time of adhesively bonded SLS specimens
and (b) SLS strengths as a function of rms roughness (c) images of ruptured specimens (A.F. Adhesive
Failure, C.F. Cohesive Failure).
4 Conclusions
An environmentally friendly salt anodization technique using a Na2HPO4 salt electro-

lyte has revealed the formation of porous anodic oxide layers on the aluminium sur-
faces with varied surface features. The morphological analysis presented porous
structure on the treated surfaces, while the EDX and the FTIR analyses confirmed the
growth of strong oxides upon salt anodization. A correlation between the anodization
time and the surface roughness was established indicating an increase in rms roughness
values as a function of anodization time (0.33 μm on as-received to 1.4 μm on 60-minute
anodized surface). The same trend has further been evidenced with the mechanical
strengths of the adhesively bonded joints showing a systematic increase in the shear lap
strengths with anodization time. Maximum shear strength of ⁓21 MPa was obtained at
60 minutes of anodization, while the as-received SLS specimens reached only ⁓10 MPa,
proving the effectiveness of the salt anodization. The Al surfaces anodized for 60 min-
utes presented 107% enhancement in adhesion strengths compared to the as-received
counterpart.
Acknowledgments We acknowledge the financial support provided by the Natural Science and
Engineering Research Council of Canada (NSERC) and Aluminum Research Center (REGAL).
References
1. Adams RD. Structural adhesive joints in engineering. Netherlands: Springer; 2012. Repéré à
https://books.google.ca/books?id=RmR-BgAAQBAJ
2. Alexander BH, Checkoway H, Wechsler L, Heyer NJ, Muhm JM, O'Keeffe TP. Lung cancer in
chromate-exposed aerospace workers. J Occup Environ Med. 1996;38:1253–8. Repéré à https://
journals.lww.com/joem/Fulltext/1996/12000/Lung_Cancer_in_Chromate_Exposed_Aerospace_
Workers.11.aspx
3. Arikan E, Holtmannspötter J, Zimmer F, Hofmann T, Gudladt HJ. The role of chemical sur-
face modification for structural adhesive bonding on polymers - Washability of chemical func-
tionalization without reducing adhesion. Int J Adhes Adhes. 2019;95 https://doi.org/10.1016/j.
ijadhadh.2019.102409.
4. Blyberg L, Lang M, Lundstedt K, Schander M, Serrano E, Silfverhielm M, Stålhandske C. Glass,
timber and adhesive joints – innovative load bearing building components. Constr Build Mater.
2014;55:470–8. https://doi.org/10.1016/j.conbuildmat.2014.01.045.
5. da Silva LFM, Öchsner A, Adams RD. Handbook of adhesion technology. Berlin Heidelberg:
Springer; 2011. Repéré à https://books.google.ca/books?id=CsKoeKzXvl8C
6. Katnam KB, Da Silva LFM, Young TM. Bonded repair of composite aircraft structures: a review of
scientific challenges and opportunities. Prog Aerosp Sci. 2013;61:26–42. https://doi.org/10.1016/j.
paerosci.2013.03.003.
7. Kwakernaak A, Hofstede J, Poulis J, Benedictus R. Improvements in bonding metals (steel, alu-
minium). In: Dans advances in structural adhesive bonding; 2010. p. 185–236. https://doi.org/1
0.1533/9781845698058.2.185.
8. Lai M, Brun R. Latest developments in sheet metal forming technologies and materials for auto-
motive application: the use of ultra high strength steels at fiat to reach weight reduction at sustain-
able costs. Key Eng Mater. 2007;344:1–8. https://doi.org/10.4028/www.scientific.net/KEM.344.1.
9. Lehmann S. Low carbon construction systems using prefabricated engineered solid wood panels
for urban infill to significantly reduce greenhouse gas emissions. Sustain Cities Soc. 2013;6:57–67.
https://doi.org/10.1016/j.scs.2012.08.004.
10. Meschut G, Janzen V, Olfermann T. Innovative and highly productive joining technologies for
multi-material lightweight car Body Structures. J Mater Eng Perform. 2014;23:1515–23. https://
doi.org/10.1007/s11665-014-0962-3.
11. Meschut G, Matzke M, Hoerhold R, Olfermann T. Hybrid Technologies for Joining Ultra-
high-strength Boron Steels with Aluminum alloys for lightweight car body structures. Proc
CIRP. 2014;23:19–23. https://doi.org/10.1016/j.procir.2014.10.089.
12. Messler RW. Joining of materials and structures: from pragmatic process to enabling technology.
Elsevier; 2004. Science Repéré à https://books.google.ca/books?id=CRt97aa7cnMC
13. Murray R. Health of workers in chromate producing industry. Br J Ind Med. 1957;14:140.
14. Ngo TD, Kashani A, Imbalzano G, Nguyen KTQ, Hui D. Additive manufacturing (3D printing):
a review of materials, methods, applications and challenges. Compos Part B. 2018;143:172–96.
https://doi.org/10.1016/j.compositesb.2018.02.012.
15. Omairey S, Jayasree N, Kazilas M. Defects and uncertainties of adhesively bonded composite
joints. SN Appl Sci. 2021;3 https://doi.org/10.1007/s42452-021-04753-8.
16. Vallero D. Fundamentals of air pollution. Academic press; 2014.
17. Vrublevsky IA, Jagminas AI, Chernyakova KV. Re-anodizing technique as a method of investiga-
tion of thermally activated defects in anodic alumina films. J Electrochem Soc. 2013;160:C285–90.
https://doi.org/10.1149/2.124306jes.
18. Zain NM, Ahmad SH, Ali ES. Effect of surface treatments on the durability of green polyurethane
adhesive bonded aluminium alloy. Int J Adhes Adhes. 2014;55:43–55. https://doi.org/10.1016/j.
ijadhadh.2014.07.007.
Tribological Performance of PH 13-8Mo Maraging
Stainless Steel Fabricated via Wire Arc Additive
Manufacturing in As-Printed and Heat-Treated
Conditions
Elham Afshari1 (*), Mahya Ghaffari1, Alireza Vahedi Nemani1, and Ali Nasiri1
1
Department of Mechanical Engineering, Dalhousie University, Halifax, NS, Canada
elham.afshari@dal.ca
Abstract: Wire arc additive manufacturing (WAAM) is a metal additive manufacturing

technology with high material deposition rate. In this study, tribological properties of
precipitation hardened (PH) 13–8Mo martensitic stainless-steel parts produced by
WAAM have been investigated. The influence of anisotropic microstructures created dur-
ing WAAM process on the wear behavior of the fabricated part was investigated by per-
forming scratch testing. In addition, to assess the effect of post-printing heat treatment on
wear performance, the as-printed samples were subjected to solution treatment followed
by aging, after which the wear properties in different directions were evaluated. It was
found that in the as-printed condition, there is a considerable difference in scratch resis-
tance between different directions with a higher scratch resistance along the deposition
direction. Applying the aging heat treatment was found to improve the homogeneity of
the microstructure and resulted in a more uniform scratch resistance throughout the
WAAM fabricated part. Besides, the wear performance of heat-treated samples was supe-
rior to that of the as-printed ones. The role of the microstructural features on the wear
mechanism of the samples is thoroughly elucidated in this article.
Keywords: Wire arc additive manufacturing · PH 13-8Mo · Scratch test · Wear
1 Introduction
Wire arc additive manufacturing (WAAM) is a metal additive manufacturing technol-

ogy with high material deposition rate, suitable for fabricating and repairing large-scale
engineering components [1, 2]. However, the sequential heating and cooling cycles
associated with the deposition process as well as the high-temperature gradient during
WAAM can result in the formation of a heterogeneous microstructure and consequently
anisotropic mechanical behavior in the as-printed part [2]. One strategy to modify the
anisotropic behavior is through applying an appropriate post-printing heat treatment
cycle. Precipitation hardened (PH) 13-8Mo maraging stainless steel, in particular,
which can be hardened by the precipitation of β-NiAl intermetallic phase with a B2
superlattice structure has been recently investigated using additive manufacturing (AM)

https://doi.org/10.1007/978-3-031-17425-4_13
74 E. Afshari et al.
processing [2, 3]. Nevertheless, previous studies in this area are mainly focused on
evaluating the mechanical properties of AM parts, and very limited knowledge is avail-
able on the tribological performance of AM manufactured parts [4]. Targeting this
knowledge gap, this study aims to investigate the tribological performance of WAAM
PH 13-8Mo in the as-printed state and after applying a post-printing heat treatment.
A commercial grade PH 13-8Mo feedstock wire with a diameter of 1.143 mm was used
to fabricate a wall structure through WAAM. The details of the fabrication setup and
processing parameters are presented in one of the authors’ previous studies [2]. To
assess the effect of post-printing heat treatment on the tribological performance, the
as-printed samples were subjected to solution treatment at 1050 °C for 1 h followed by
still-air cooling and aging at 500 °C for 4 h. The microstructures of the as-printed and
heat-treated specimens were analyzed by performing scanning electron microscopy
(SEM) and transmission electron microscopy (TEM). The tribological behavior of the
fabricated part was investigated by performing scratch testing along the building direc-
tion and transverse direction. All samples were polished prior to the scratch testing to
ensure that the surface roughness of the samples is less than 0.15 μm. Scratch tests were
performed by employing a Universal Mikro-Tribometer under three constant loads of 5,
10, and 15 N at a constant speed of 10 mm/min and a 5-mm scratch length using a coni-
cal diamond indenter with a tip radius of 200 μm and a cone angle of 120°. Coefficient
of Friction (COF) values were automatically recorded by the instrument.

3.1 Microstructural Characterizations
In the case of the as-printed sample, as shown in Fig. 1a, as a result of the high-temperature
gradient along the building direction, columnar grains have grown aligned with the build-
ing direction. Besides, the microstructure of the as-printed component shows residual
δ-ferrite (elongated dark phase in Fig. 1a) distributed in a matrix of low-carbon martens-
ite, mainly elongated along the building direction. In fact, in equilibrium solidification, it
is expected that upon cooling, δ-ferrite transforms to austenite and then to martensite;
however, as a result of rapid cooling during the WAAM process, a fraction of δ-ferrite
remains at room temperature [2]. In the case of the heat-treated sample, as shown in
Fig. 1b, the columnar grains and residual δ-ferrite were disappeared and the microstruc-
ture was characterized by a fully martensitic microstructure. The solution treatment for
1 h at 1050 °C provides enough time for residual δ-ferrite to transform into austenite,
which then transforms into martensite during subsequent air cooling. In addition, TEM
results (Fig. 1c) revealed that the solution-aging treatment resulted in the formation of a
uniform distribution of nano-scale β-NiAl precipitates within the martensitic matrix.
Tribological Performance of PH 13-8Mo Maraging Stainless Steel Fabricated via Wire… 75
Fig. 1. (a) SEM micrograph of the as-printed sample, (b) SEM micrograph of the heat-treated sample,
(c) TEM bright field image of the heat-treated sample (arrowhead indicating β-NiAl precipitates), and
(d) scratch width vs. load curves obtained from scratch tests performed along the building direction
(BD) and transverse direction (TD) of the as-printed (AP) and heat-treated (HT) samples.
3.2 Tribological Performance
The scratch width of the as-printed (AP) and heat-treated (HT) samples for scratch
tests performed along the building direction (BD) and transverse direction (TD) are
compared in Fig. 1d, showing an anisotropic scratch resistance in the as-printed condi-
tion characterized by lower scratch resistance along the BD, while the scratch resis-
tance was almost identical in two perpendicular directions in the heat-treated sample.
The anisotropic scratch resistance in the as-printed sample can be ascribed to the
columnar grain structure in the as-printed condition, which provides a higher density
of grain boundaries along the TD and consequently, a larger required force for slipping
and accordingly higher scratch resistance [5]. However, in the heat-treated condition,
as a result of the elimination of heterogeneous microstructure as well as soft δ-ferrite
phase, and also precipitation of hard nano-scale NiAl precipitates, isotropic and higher
scratch resistance were observed. Additionally, after the solution-aging treatment, the
coefficient of friction was decreased compared to the as-built state. The average COF
of the heat-treated sample (~0.13) was 23% lower than the COF of the as-printed
sample (~0.17).
76 E. Afshari et al.
4 Conclusion
In this study, the tribological properties of PH 13-8Mo maraging stainless-steel wall

structure produced by WAAM have been investigated. It was found that in the as-printed
condition, there is a considerable difference in scratch resistance between different
directions of the fabricated part with the higher scratch resistance along the transverse
direction as compared to the building direction. Applying the solution-aging heat treat-
ment was found to improve the homogeneity of the microstructure and resulted in a
more uniform scratch resistance throughout the WAAM fabricated part. Besides, the
tribological performance of heat-treated samples was superior to that of the as-printed
ones.
References
1. W u B, Pan Z, Ding D, Cuiuri D, Li H, Xu J, Norrish J. A review of the wire arc additive manufac-
turing of metals: properties, defects and quality improvement. J Manuf Process. 2018;35:127–39.
2. Ghaffari M, Vahedi Nemani A, Nasiri A. Microstructure and mechanical behavior of PH
13–8Mo martensitic stainless steel fabricated by wire arc additive manufacturing. Addit Manuf.
2022;49:102374.
3. Hadadzadeh A, Shahriari A, Shalchi Amirkhiz B, Li J, Mohammadi M. Additive manufacturing of
an Fe–Cr–Ni–Al maraging stainless steel: microstructure evolution, heat treatment, and strengthen-
ing mechanisms. Mater Sci Eng A. 2020;787:139470.
4. Chang C, Yan X, Deng Z, Lu B, Bolot R, Gardan J, Deng S, Chemkh M, Liu M, Liao H. Heat
treatment induced microstructural evolution, oxidation behavior and tribological properties of
Fe-12Cr-9Ni-2Al steel (CX steel) prepared using selective laser melting. Surface Coat Technol.
2022;429:127982.
5. Yanga Y, Zhua Y, Khonsaric MM, Yang H. Wear anisotropy of selective laser melted 316L stainless
steel. Wear. 2019;428–429:376–86.
High-Speed Nanoindentation Mapping of Additive
Manufactured Titanium Alloys for Aerospace
Application
Zhiying Liu1, Haoxiu Chen1, and Yu Zou1 (*)
1
Department of Materials Science & Engineering, Faculty of Applied Science & Engineering,
University of Toronto, Toronto, ON, Canada
mse.zou@utoronto.ca
Abstract. High-speed nanoindentation mapping was carried out using a nanoindenter

(iMicro, KLA Tencor, USA) using the method of Nanoblitz3D. This method enables
generating high-resolution spatial H and E maps over large areas (up to hundreds of
microns) by performing up to thousands of indents. Each indent test took up to 1 second,
including the surface approaching, loading, unloading, retracting, and positioning the
sample for the next test. A diamond Berkovich tip (Synton-MDP, Switzerland) was used
for the nanoindentation tests. Several trial-and-error nanoindentation experiments were
conducted to identify an optimal indentation depth and indent spacing. In our study, we
performed the indentation mapping to capture the intrinsic H and E of the α and β
phases in a Ti-6Al-2Zr-V-Mo alloy. The ratio of indent spacing to indentation depth is
larger than 10 to avoid any overlapping effect between indentations. A maximum load of
1 mN, which results in a penetration depth ranging between 50 and 100 nm for α and β
grains, and 500 nm spacing between the indents were chosen for all the tests. The H and
E of α and β phases in the as-deposited and heat-treated samples are obtained from the
same load, which are valid for comparison. The nanoindentation mapping with ~10,000
indents (50 × 50 μm2 areas) and ~20,000 indents (50 × 100 μm2 area) were carried out on
the as-deposited and heat-treated samples, respectively. The indented areas of both sam-
ples contain more than 10 α and β grains to obtain microstructure-property correlations.
The hardness (H) and reduced elastic modulus (Er) at each test location can be calculated
using the standard Oliver and Pharr method. The Er includes the elastic contributions of
both the indenter and the indented materials.
Keywords: Additive manufacturing · Titanium alloys · Nanomechanics
1 High-Speed Nanoindentation Mapping
High-speed nanoindentation mapping was carried out using a nanoindenter (iMicro,

KLA Tencor, USA) using the method of Nanoblitz3D. This method enables generating
high-resolution spatial H and E maps over large areas (up to hundreds of microns) by
performing up to thousands of indents. Each indent test took up to 1 second, including
the surface approaching, loading, unloading, retracting, and positioning the sample for
the next test. A diamond Berkovich tip (Synton-MDP, Switzerland) was used for the

https://doi.org/10.1007/978-3-031-17425-4_14
78 Z. Liu et al.
nanoindentation tests. Several trial-and-error nanoindentation experiments were con-

ducted to identify an optimal indentation depth and indent spacing. In our study, the
indentation depth should be more than 50 nm to capture the intrinsic H and E of the α
and β phases, while the ratio of indent spacing to indentation depth should be larger
than 10 to avoid any overlapping effect between indentations [1]. A maximum load of
1 mN, which results in a penetration depth ranging between 50 and 100 nm for α and
β grains, and 500 nm spacing between the indents were chosen for all the tests. The H
and E of α and β phases in the as-deposited and heat-treated samples are obtained from
the same load, which are valid for comparison. The nanoindentation mapping with
~10,000 indents (50 × 50 μm2 areas) and ~20,000 indents (50 × 100 μm2 area) were
carried out on the as-deposited and heat-treated Ti-6Al-2Zr-V-Mo (TA15) samples [2],
respectively. The indented areas of both samples contain more than 10 α and β grains
to obtain microstructure-property correlations. The selected microstructures and
resulted micromechanical properties of both samples are representative. The hardness
(H) and reduced elastic modulus (Er) at each test location can be calculated using the
standard Oliver and Pharr method [3]. The Er includes the elastic contributions of both
the indenter and the indented materials, as expressed by the following equation:

2
1 1 v2 1 v
(1)
Er E E

where E′ (1141 GPa) and v′ (0.07) are the elastic modulus and Poisson’s ratio of the
diamond Berkovich indenter, respectively. The Poisson’s ratio v of the TA15 alloy is
~0.39 [2]. Because the calculated E is about 98% of the Er in this work, the Er is used
in this work to approximately represent the E of the TA15 alloys.
2 Microstructure Characterization
The microstructures and nanoindentation impressions of both as-deposited and heat-

treated samples were imaged using a scanning electron microscope (SEM, Hitachi
3500, Japan) equipped with a backscattered electron (BSE) detector. Phases in the as-
deposited and heat-treated samples were determined by X-ray diffraction (XRD,
Rigaku MiniFlex 600) over a 2θ range between 30° and 90° at a scanning rate of 0.025°
s−1. The quantitative chemical compositions of α and β phases and the element maps
were acquired using an electron probe microanalyzer (EMPA, JEOL JXA8230).
Figure 1a shows a schematic of the Laser Melting Deposition (LMD) process, and
Fig. 1b, c present the high-speed nanoindentation mapping process and nanoindenta-
tion impressions, respectively. Prior to the nanoindentation tests, we characterized the
High-Speed Nanoindentation Mapping of Additive Manufactured Titanium Alloys ... 79
Fig. 1. (a) A schematic of the LMD system for additive manufacturing of the titanium alloy (TA15).
(b) The SEM image of the nanoindenter with the Berkovich tip on the titanium sample surface. (c) The
BSE image revealing the nanoindentation impressions on the α and β phases of the TA15 sample after
the high-speed nanoindentation mapping.
pre-indent regions using Electron Probe Microanalyzer (EPMA) to elucidate the contri-
bution of chemical composition to the resulting H and E values of α and β grains.
Figure 2 shows the microstructure, the corresponding H and E maps of the
as-deposited TA15 sample [4]. The majority of the α grains, far larger than the β grains,
can be captured and correlated with the spatial H and E maps (Figs 2a, b and d). For
example, the three α grains, intersected by the line AB in Fig. 2a, involve tens of indents,
showing the profile in the H and E maps. We can accurately extract the H and E values
along with the line AB by point-to-point analysis, as shown in Fig. 2c. The α grains
show distinct H and E values. High-resolution H and E maps also virtualize the mor-
phology of the α and β grains, including a clear profile of the α/β boundaries. It is
believed α/β boundaries are critical to determine the local deformation mechanism.
Here, the high-speed nanoindentation mapping enables to measure the H and E distri-
bution in the vicinity of the α/β boundary. It should be noted the effect of the α/β bound-
ary on the H and E characterization is neglectable according to the absence of H or E
change when approaching α/β boundaries (Fig. 2). Figure 3 presents the microstructure,
corresponding H and E maps, and Al content of the heat-treated TA15 sample. The
excellent correlation between the microstructure and mechanical property maps (H and
E) is attributed to the large size of α and β grains. The contrasting H and E in the α + β
microstructure are clearly observed. The α grains exhibit distinct H and E, which is
consistent with the observation in the as-deposited sample. The homogeneous Al distri-
bution in the α grains excludes the contribution of the solid solution strengthening to the
contrasting H and E of the α grains. The H and E contrasts in the α grains should stem
from the anisotropy of the hcp crystal structure. In addition to the titanium alloy, high
speed nanoindentation mapping tests can be also used to capture grain-sclae mechani-
cal properties in other multiple phase alloys, e.g., graded γ-TiAl/Ti2AlNb alloys [5] and
steel [6].
80 Z. Liu et al.
Fig.2. (a) BSE image of the as-deposited TA15 alloy. (b) The hardness (H) and (d) elastic modulus
(E) of the rectangle region in (a). (c) The H and E values along with the line AB in 3(a), which involves
three α grains.
4 Conclusions
In this study, we employ the high-speed nanoindentation technique to characterize the

local mechanical properties (hardness and Young’s modulus) of a near-α titanium alloy,
Ti-6Al-2Zr-Mo-V or TA15, made by laser melting deposition. The following conclu-
sions are drawn:
1. The high-speed nanoindentation mapping is an effective approach to reveal mechan-
ical contrasts (H and E) of multiphase alloys with complex microstructures, such as
3D printed titanium alloys. The speed of indentation can reach approximately 1
second per indent (10,000 indents for ~2.5 hours). The special resolution can reach
the submicrometer regime (~300–500 nm). For the heat-treated TA15 sample, this
method provides the H and E distributions, clearly correlates the dual-phase (α + β)
microstructure, reveals the anisotropy of the hcp α phase. This method shows a lim-
ited mechanical-phase correlation for the as-deposited sample due to its ultrafine
microstructure (grain size ~ 1 μm).
High-Speed Nanoindentation Mapping of Additive Manufactured Titanium Alloys ... 81
Fig. 3. (a) BSE image of the heat-treated TA15 alloy. (b) The H and (d) E maps of the rectangle region
in 3(a). (c) The distribution of Al content in the rectangle region in 3(d), as measured by EPMA.
2. Using such a method, we observe that heat treatment leads to a concurrent hardening
of the α and β phases, increasing by 32% and 12%, respectively. This phenomenon
is associated with the redistribution of alloying elements: Al diffuses from the β
phase into neighboring α phase, while Mo and V diffused along the opposite direc-
tion. The element redistribution also results in the increment of the E, by 4% in the
β phase and 14% in the α phase. This mechanical distribution cannot be easily
revealed using conventional indentation methods.
3. Toward broad applications, the high-speed nanoindentation technique, together with
advanced microstructure characterization and machine learning tools, brings unique
opportunities to study and design new multiphase materials across length scales.
82 Z. Liu et al.
References
1. Phani PS, Oliver W. A critical assessment of the effect of indentation spacing on the measurement
of hardness and modulus using instrumented indentation testing. Mater Des. 2019;164:107563.
2. Liu Z, He B, Lyu T, Zou Y. A review on additive manufacturing of titanium alloys for aerospace
applications: directed energy deposition and beyond Ti-6Al-4V. JOM. 2021;73:1804–18.
3. Oliver W, Pharr G. Nanoindentation in materials research: Past, present, and future. MRS Bulletin.
2010;35(11):897–907.
4. Liu Z, Zhang J, He B, Zou Y. High-speed nanoindentation mapping of a near-alpha titanium alloy
made by additive manufacturing. J Mater Res. 2021;36(4):2223–34.
5. Chen H, Liu Z, Cheng X, Zou Y. Laser deposition of graded γ-TiAl/Ti2AlNb alloys: microstructure
and nanomechanical characterization of the transition zone. J. Alloys Compd. 2021;875:159946.
6. Chen H, Niu G, Wu H, Zou Y. Revealing non-uniform work-hardening in a heterogeneous struc-
tured steel using high-speed nanoindentation. JOM. 2022;74(6):2238–44.
Picosecond Laser Surface/Deep Patterning
of Alumina Ceramics
Hamidreza Yazdani Sarvestani1 (*), Warren Amsellem1, Zachary Katz1,

Ilhan Esmail1, Clement Beausoleil1, Javad Gholipour Baradari1, and Behnam Ashrafi1
1
Aerospace Manufacturing Technology Center, National Research Council Canada,
Montreal, QC, Canada
hamidreza.yazdani@nrc-cnrc.gc.ca
Abstract. Traditional machining techniques pose significant drawbacks when applied to

ceramics due to the material’s inherent brittleness. Specialized laser machining has been
known to solve these issues through higher precision and micrometer-scale feature con-
trol. In this study, a picosecond fiber laser has been used as a material removal system for
different engineering applications of industrial-grade alumina ceramics with a variety of
thicknesses and feature dimensions. This work explored picosecond laser process param-
eters such as focal position, linear speed, and wobble amplitude in order to control the cut
depth and optimize cut quality in terms of kerf width, kerf taper, and surface cleanness
while avoiding crack formation. The surface roughness was assessed to determine the
quality of the cuts. The optimal process parameters between the surface finish and mate-
rial removal rate were introduced. Using a circular wobble laser pattern, it was determined
that a greater cut depth can be achieved at lower linear speeds and wobble frequencies due
to the higher linear energy density. It has also been found that the kerf taper is dependent
on the cut depth and wobble amplitude, where the measured cuts follow the geometric
relation between these parameters accurately.
Keywords: Laser cutting · Alumina ceramics · Ablation · Surface profile ·

Picosecond laser
1 Introduction
Ceramics are known for their high compressive strength, stiffness, and thermal and
chemical resistance. As a result, these properties have increased the use of ceramics in
different applications [1]. However, one of the main challenges with ceramic process-
ing is the machinability of ceramics for producing complex parts. Ceramic cuts using
mechanical methods induce microcracks along the cut interface which results in crack
initiation sites as well as residual stresses [2–6]. This is due to the inherent brittleness
of ceramics and is the reason for the low fracture toughness of monolithic components.
With laser processing, the material is no longer subject to vibration or cutting forces
(shear forces) and instead is exposed to a high-energy beam which results in rapid abla-
tion of the material [7].

https://doi.org/10.1007/978-3-031-17425-4_15
84 H. Y. Sarvestani et al.
There are different types of laser cutting/engraving systems applied for several vari-
eties of materials where complex shapes demand precise, fast, and force-free process-
ing. For example, traditional CO2 lasers offer high-beam quality but are rarely used for
ultra-short processing applications [8]. Nd:YAG lasers are capable of producing higher
peak power than gas lasers in the form of extremely short bursts, but suffer from poor
energy efficiency and beam quality [1, 9]. Nanosecond lasers are preferable for material
removal or ablation [10, 11]; however, there is risk of forming a significant heat-affected
zone and microcrack formations in ceramics, which would affect the ultimate quality of
cut geometries [12, 13]. In order to minimize the heat-affected zone and induced micro-
cracks [14], pico-femtosecond lasers were utilized to engrave/cut alumina ceramics
[15, 16]. Picosecond (short pulse duration) and femtosecond lasers (ultra-short pulse
duration) yield high-quality results, appropriate for the production of high-precision
micro- and nanomachining [17].
In this study, a systematic approach was developed to predict the cut profile (i.e.,
kerf taper and cut depth) of the ablated, smooth, and defect-free laser cuts on industrial
grade alumina ceramics. The approach was based on the various input process param-
eters including wobble amplitude, wobble frequency, linear speed, and number of
passes of the picosecond laser system.
The laser used in this study is an Ytterbium Picosecond fiber laser (YLPP-25-3-
50-R, IPG Photonics, USA) with a maximum average power of 50 W and a laser
beam waist diameter of 17 μm. It produces a Gaussian spatial profile beam with 3 ps
long 25 μJ pulses of 1030 nm wavelength and is capable of a repetition rate of up to
1.83 MHz. Following the results obtained from the previous study [18], it was deter-
mined that optimal cut quality could be achieved by using a circular wobble pattern.
Wobble is the optional beam oscillation pattern drawn perpendicular to the laser
trajectory. It is defined by pattern shape in the laser software (line, circle, eight (8),
and infinity (∞)), repetition frequency (Hz), and wobble amplitude (mm). This
scheme directs the laser to produce circles parallel to the substrate surface, as shown
by the schematic in Fig. 1a along with a true visualization of the wobble pitch using
a 1 mm amplitude. The ceramics used are 1.016, 1.524, and 2.540 mm (0.04, 0.06,
and 0.1 in, respectively) thick industrial nonporous 96% alumina tiles (McMaster-
Carr, CA) with a density and porosity of 3875 kg/m3 and 0%, respectively. Cuts of
10 mm in length were generated to study the effect of laser process parameters on the
quality of the cuts. The cut depth, kerf taper, and kerf width were then measured
using a KEYENCETM 3D Laser Scanning Confocal Microscope, the results of
which are represented in the diagram in Fig. 1b, c for different cut profiles (i.e., deep
and shallow).
Picosecond Laser Surface/Deep Patterning of Alumina Ceramics 85
Fig. 1. (a) Schematic and microscope image of the circular wobble pattern illustrating the laser pulses
and direction, wobble amplitude, wobble pitch, kerf width, linear speed, and vector speed. Schematics
of the cut measurement nomenclature of (b) Deep, developed cut profiles (>40 passes) and (c) Shallow
cut profiles (<40 passes).
3 Results
The kerf taper of ablated cuts with developed “V-shape” can be determined using
θ = arctan (W/2D) (where θ, W, and D are the kerf taper, kerf width, and cut depth,
respectively), which is a function of the cut depth and wobble amplitude. Data gathered
from various ablation cut experiments are plotted against the geometric approximation,
as shown in Fig. 2. Plotted experimental data shows a strong correlation with the theo-
retical formulation. Here, the kerf taper is primarily controlled by the wobble amplitude
if a through-cut is required for the sample. By modifying this input, the theoretical kerf
taper range can be predicted using this approach. Making a through cut on a 2.54-mm-
thick alumina tile can produce 3.4–21.5° kerf tapers using a 0.3–2.0 mm range of wob-
ble amplitudes. However, there exist a physical limitation of manufacturing through
cuts on the 2.54 mm samples for the smaller amplitudes up to 1.0 mm. To achieve this
depth using the 0.5 mm wobble amplitude, the power-law relationship estimated the cut
time to be in the order of 107 s, thus limiting its practical feasibility. This result is due
to clipping of the incident laser spots by the tapered surfaces of the cuts, thereby
decreasing the energy density on the cut area. For smaller amplitudes, this phenomenon
is more significant, which hinders the effectiveness of the laser to ablate material within
the cut region. Therefore, the long processing times, which is a result of the decreasing
efficiency does not make the kerf taper predicted by the 0.3 mm and 0.5 mm wobble
amplitudes viable for ablating depths greater than 1.2 mm. However, it is important to
note that this limitation may be addressed by lowering the focal position during the cut-
ting process as preliminary investigations have shown. Finally, as the amplitude of the
circular wobble increases, the beam clipping decreases due to the wider cut geometry.
Therefore, deeper cuts can be achieved for the larger wobble amplitudes over the 220
Fig. 2. Geometric dependence of kerf taper to the cut depth for the V-shaped cut profiles.
passes that were required to cut through the 2.54-mm-thick alumina ceramic tile using
the 1.5 mm wobble amplitude.
Finding a metric to evaluate the viability of each of these processing schemes was
the final step toward implementing the circular wobble pattern for manufacturing near-
net geometries. The quality of ablated cuts deteriorates significantly with wobble
pitches above the 30-μm threshold as was illustrated by the large increase in surface
roughness in Fig. 3. A visually noticeable decrease in cut quality is determined for
wobble pitches above 30 μm. This value is in agreement with the laser spot size at the
chosen focal position of 400 μm below the surface of the ceramic. At wobble pitches
above the laser spot diameter, the lack of spot over-lap results in a decrease of total
incident energy, particularly at the center of the cut. This creates the large peaks and
valleys where the edges of the cuts experience large overlap due to the nature of the
circular cut pattern. Significantly larger pitches (e.g., 70 μm) result in shallow circular
cuts due to the least spot overlap of the tested pitches.
Furthermore, the viability of the process is determined by the material removal rate
after a set number of passes. The data in Fig. 4 illustrate this evaluation using the 1 mm
wobble amplitude after 20 passes. An optimal balance between the quality of the cut
and material removal rate was found at Number 1 (300 Hz wobble frequency, 8 mm/s
linear speed and 26.67 μm wobble pitch), which was the scheme used later to manufac-
ture various architectured geometries. The proposed approach to predict an ablated
high-quality cut geometry (i.e., kerf taper and cut depth) was based on the wobble
amplitudes, wobble frequency, linear speed, and the number of passes. First, by
50
10 µm pitch
30 µm pitch
40
Surface roughness - Sa (µm)
40 µm pitch
70 µm pitch
30
20
10
0
0.5 1.0 1.5 2.0
Amplitude (mm)
Fig. 3. Surface roughness of laser-cut profiles comparing the cut quality for different wobble pitches
and amplitudes.
choosing a kerf taper or cut depth, a set of practically feasible wobble amplitudes can
P Ep · fp
be calculated based on EL (see Fig. 2). The desired depth can either be a
Afw
through cut based on the thickness of the ceramic tile or a partial cut to create a net-
shaped cavity. To prioritize material removal rate over processing time, a set of low
wobble frequencies are chosen, and corresponding linear speeds such that the wobble
pitch is less than or equal to 30 μm to maintain acceptable cut quality (see Fig. 4). The
specific number of passes is then determined based on the amplitude to obtain the final
cut depth and kerf taper. Finally, the experimental cut depth and kerf taper can be com-
pared with the initial inputs to adjust the wobble amplitude or number of passes.
The present work delineates that high precision, deep ablated cuts can be achieved
on alumina ceramics through the implementation of low wobble frequencies and linear
speeds. Since other types of ceramics have similar characteristics (e.g., material removal
by melting vs. ablation), the developed methodology for deep and angle cutting can be
applied with similar trends; however, different energy densities should be applied.
I. Lower values of wobble frequencies and linear speeds increase the linear energy
density of the picosecond laser that results in a higher material removal rate. This
is an important consideration for upscaling the deep cutting process.
II. Both depth and wobble amplitude measurements as functions of wobble frequency
were in agreement with the power-law relationship of these parameters with the
linear energy density formulation.
III. By setting a maximum wobble pitch (in this study 30 μm), the wobble frequency
and linear speed should be optimized for each specific ceramic.
IV. In addition, larger wobble amplitudes can deliver deeper cuts due to decreased
beam clipping which is caused by the kerf edges of the cut geometry. However, an
important consideration is that higher wobble amplitudes have lower linear energy
densities and, when set large enough, may not yield efficient cuts (partial melting
70
60
5
Wobble pitch (µm) 50
2
40
30
1
20
3
10 4
0
0 1 2 3 4 5 6 7
Material removal rate (mm3/min)
36.63µm 34.12µm 39.29µm
-200 -200
-200
-400 -400
-400
-600 -600
-800 -800
-600
-1000 -1000
-1135.07 -839.59 -1168.9
1 2 3
33.6µm 80.23µm
0
-200
-400 -100
-600
-200
-800
-300
-1000
-1187.84 -426.11
4 5
Fig. 4. Evaluation of the wobble pitch and the material removal rate for the 1-mm wobble amplitude
cuts to assess the trade-off with low processing time while maintaining high cut quality.
and ablation instead of full ablation). In contrast, low wobble amplitudes (i.e.,
0.1 mm) have significantly larger linear energy densities initiating melting mecha-
nisms during manufacturing. There is high variability of the measured cut depth
geometry which also demonstrates a visible recast layer causing the initiation of
cracks.
V. A non-linear relationship was observed between the material removal rate and the
number of passes. There are a plethora of reasons for this outcome, a few of which
are: the beam is increasingly out of focus as ablation occurs at deeper regions in
the sample, the V shape created by the beam spreads the laser spot over a larger
region, clipping of the beam incurred by cut wall shadowing reducing the fluence,
increased dust and fume interactions with the laser beam.
VI. In addition to cut depth and kerf width, cut quality is an important parameter to
consider. Processing time and cut quality are mutually exclusive with respect to
wobble pitch. As the wobble pitch increases, the overall cutting time decreases
since a higher linear speed is used for the same input wobble frequency.
VII. By increasing the wobble pitch, the distance between the wobble circles increases
and the subsequent overlap of laser pulses decreases, resulting in the “valley and
hill” patterns at the center of the cuts.
4 Conclusions
This paper presented a novel method to control the final cut depth and kerf taper to
produce cavities and through cuts by implementing a circular wobble scanning pattern
with a picosecond laser. Using a range of input process parameters to control the linear
energy density and geometry of this pattern (i.e., wobble amplitude, wobble frequency,
wobble pitch, and linear speed), smooth, deep and defect-free cuts of desired geome-
tries were obtained with a high degree of precision. It was found that the smaller wob-
ble amplitudes and lower frequencies produce deeper cuts and smaller kerf tapers.
Lower wobble frequencies are preferential when considering the processing time, as a
greater cut depth can generally be achieved for the same linear speed but 100 Hz must
be considered as the lower threshold since significant melting of the ceramic and large
heat-affected zones was observed at this frequency. In addition, control of the wobble
pitch can be achieved by modifying the linear speed for a chosen wobble frequency.
The surface roughness, which is used to determine cut quality, increased significantly
for wobble pitches above 30 μm. A 70 μm wobble pitch limits the ablation to a shallow
cut profile and can be used for applications that require specific surface roughness’s.
The developed approach accurately defines the kerf taper as a function of the cut depth
for cut profiles that have a developed “V-shape.” This approach allows the design of
different advanced engineering applications such as manufacturing of topologically
interlocked ceramics.
Acknowledgments The authors would like to thank the security, materials, and technologies (SMT)
program at National Research Council Canada (NRC).
References
1. Rakshit R, Das AK. A review on cutting of industrial ceramic materials. Precis Eng. 2019;59:90–109.
2. Jianxin D, Taichiu L. Surface integrity in electro-discharge machining, ultrasonic machining, and
diamond saw cutting of ceramic composites. Ceram Int. 2000;26(8):825–30.
3. Li K, Liao TW. Surface/subsurface damage and the fracture strength of ground ceramics. J Mater
Process Technol. 1996;57(3–4):207–20.
4. Fenoughty K, Jawaid A, Pashby I. Machining of advanced engineering materials using traditional
and laser techniques. J Mater Process Technol. 1994;42(4):391–400.
5. Samant AN, Dahotre NB. Laser machining of structural ceramics—a review. J Eur Ceram Soc.
2009;29(6):969–93.
6. Parandoush P, Hossain A. A review of modeling and simulation of laser beam machining. Int J
Mach Tools Manuf. 2014;85:135–45.
7. Šugár P, et al. Laser beam milling of alumina ceramics-the impact on material removal efficiency
and machined surface morphology. In: Solid state phenomena. Zurich: Trans Tech Publ; 2017.
8. Duley W. CO2 lasers effects and applications. Elsevier; 2012. Academic Press, Inc. (London) Ltd.
9. Zhou B, et al. Efficient, frequency-stable laser-diode-pumped Nd: YAG laser. Opt Lett.
1985;10(2):62–4.
10. Knowles M, et al. Micro-machining of metals, ceramics and polymers using nanosecond lasers. Int
J Adv Manuf Technol. 2007;33(1–2):95–102.
11. Zhang H, et al. Ablation behavior of SiCp/AA2024 composites irradiated by a single-pulse nano-
second laser. Opt Laser Technol. 2020;126:106075.
12. Wang X, et al. Fabrication of micro-channels on Al2O3/TiC ceramics using picosecond laser
induced plasma micromachining. J Manuf Process. 2019;44:102–12.
13. Kim SH, Sohn I-B, Jeong S. Ablation characteristics of aluminum oxide and nitride ceramics dur-
ing femtosecond laser micromachining. Appl Surf Sci. 2009;255(24):9717–20.
14. López JML, et al. Droplet assisted laser micromachining of hard ceramics. J Eur Ceram Soc.
2016;36(11):2689–94.
15. Liu Y, et al. Effect of texture parameters on cutting performance of flank-faced textured carbide
tools in dry cutting of green Al2O3 ceramics. Ceram Int. 2018;44(11):13205–17.
16. Lin Q, et al. The effect of spot overlap ratio on femtosecond laser planarization processing of SiC
ceramics. Opt Laser Technol. 2020;129:106270.
17. Zhang Y, et al. Effects of laser repetition rate and fluence on micromachining of TiC ceramic.
Mater Manuf Process. 2016;31(7):832–7.
18. Beausoleil C, et al. Deep and high precision cutting of alumina ceramics by picosecond laser.
Ceram Int. 2020;46:15285–96.
Hybrid Manufacturing of IN718
Sila Atabay1 (*), Priti Wanjara1, Javad Gholipour Baradari1, Josh Soost2,
and Mathieu Brochu3
1
National Research Council Canada, Montreal, QC, Canada
sila.atabay@nrc-cnrc.gc.ca
2
Matsuura Machinery USA Inc, St Paul, MN, USA
3
Department of Mining and Materials Engineering, McGill University, Montreal, QC, Canada
Abstract. This study investigated the application of a hybrid manufacturing method to

fabricate Inconel® 718 (IN718) benchmarking coupons using in-envelop laser powder
bed fusion (LPBF) processing and subsequent high-speed machining. The density of the
as-fabricated coupons was measured using the Archimedes method. The surface finish of
coupons in the as-fabricated and post-machined conditions was then examined compre-
hensively to measure the average roughness using linear (profilometer) and areal/surface
(laser confocal microscopy) methods. Four different post-processing heat treatments
were applied to the as-fabricated coupons. The heat treatment response of hybrid (LPBF)
manufactured IN718 was investigated on the basis of the microstructure and microhard-
ness relative to the as-fabricated condition.
Keywords: Hybrid additive/subtractive manufacturing · Laser powder bed fusion ·

Ni-based superalloys · Inconel® 718 · Heat treatment · Microstructure · Mechanical
properties
1 Introduction
Inconel® 718 (IN718) is a precipitation-hardenable nickel-based superalloy, with

metastable disc-shaped γ″ precipitates having a body-centered tetragonal (BCT) struc-
ture and spherical or cuboidal γ′ precipitates with an ordered L12 structure, dispersed
in a face-centered cubic (FCC) γ matrix [1, 2]. In addition to these precipitates, NbC
and δ phases are observed in the microstructure of IN718. Among these, γ″ and γ′ are
the main strengthening precipitates, providing superior elevated temperature and
mechanical properties. By contrast, NbC and controlled precipitation of the δ phase
contribute to the microstructural stability at elevated temperatures by inhibiting grain
boundary sliding [3]. In addition to high strength and good resistance to creep and
fatigue at temperatures up to 700 °C, IN718 is also known to have good weldability due
to its relatively sluggish precipitation kinetics [4]. The combination of these properties
has made IN718 a popular alloy for aeroengine components, making up more than 30%
of the total weight of modern aircraft engines [5]. Traditionally, IN718 is processed by
casting, forging, and extrusion, which have very limited degrees of freedom especially
for fabricating complex geometries. Additionally, fabrication of these geometries

https://doi.org/10.1007/978-3-031-17425-4_16
92 S. Atabay et al.
requires extensive post-process machining, resulting in a significant amount of material

wastage [6]. Thus, the use of a near-net-shape processing method to fabricate IN718
parts would be beneficial both from economic and ecological perspectives.
Laser powder bed fusion (LPBF) is one of the main additive manufacturing (AM)
processes, which involves selective melting and solidification of a thin layer of powder
with the aid of a predefined laser raster path [7]. LPBF offers design flexibility by
enabling the fabrication of near-net-shape components having complex geometries
with reduced material waste [8]. However, LPBF AM still has certain drawbacks that
require further investigations and optimization for specific geometries and components.
A key concern is the poor surface quality of the LPBF-fabricated parts that, especially
for fatigue-critical applications, requires post-process machining, which can obliterate
the cost benefits associated with additive processing [9]. Hence, it would be of interest
to use a hybrid manufacturing technique – involving a single setup (in-envelop) that
applies both additive and subtractive processes – for the production of complex geom-
etry parts with high-quality surface finishes and tight dimensional tolerances.
Thus, this study investigated the application of a hybrid additive-subtractive method
(Matsuura LUMEX-Avance-25) to fabricate IN718 benchmarking coupons. The cou-
pons were then examined comprehensively for surface finish both with and without
high-speed machining. The microstructure of the hybrid manufactured IN718 samples
was investigated thoroughly in the as-fabricated condition and following post-
processing heat treatment. Specifically, four different heat treatments were applied in
this study based on material standards and/or recent procedures for IN718. Hardness
testing was then used to characterize the response of the IN718 microstructure to these
heat treatments.
IN718 powder provided by Matsuura USA was used in this study. The powder mor-
phology was characterized using a Hitachi SU3500 scanning electron microscope
(SEM). As shown in Fig. 1a, the powder particles were mainly spherical in morphology
with a few satellites and irregularities. The particle size distribution of the powder was
measured using an LA-920 Horiba particle size analyzer and the D10, D50, and D90 val-
ues were, respectively, 26 μm, 37 μm, and 55 μm. Flowability of the powder was
assessed using Hall and Carney funnels and the respective flow rates were 25 ± 1 s and
4 ± 2 s for 50 mg of powder, indicating that the powder has suitable flowability to be
used in the LPBF process. A Matsuura LUMEX Avance-25 hybrid LPBF and high-
speed machining system was used to fabricate the sample geometry represented in
Fig. 1b. The base plate was heated to 50 °C prior to printing. LPBF process was per-
formed using a laser power of 240 W, laser spot diameter of 200 μm, a laser scanning
speed of 700 mm/s for the infill scans and 500 mm/s for the contour scans, a hatch
distance of 110 μm, and a layer thickness of 50 μm with 90° rotation between the lay-
ers. After the part was printed, milling was performed at a feed rate of 2000 mm/min on
the profiles along two side faces of the reduced cross-section, as well as on the final top
surface of the entire part.
Hybrid Manufacturing of IN718 93
Fig. 1. (a) SEM micrograph of the IN718 powder used for hybrid manufacturing and (b) geometry of
the fabricated specimens.
Table 1. Summary of the applied heat treatments.

Temperature Time
Designation Standard Treatment (°C) (h)° Cooling
HT1 AMS 5663 Solution 980 1 Air cooling (AC)
[11] Age 720 8 Furnace cooling to 620 °C
(FC) (55 °C/h)
620 8 AC
HT2 AMS 5383 Homogenize 1080 1.5 AC
[12] Solution 980 1 AC
Age 720 8 FC to 620 °C
620 8 AC
HT3 AMS 5664E Homogenize 1065 1 AC
[13] Age 750 10 FC to 620 °C
650 8 AC
HT4 Non-standard Solution 1020 0.25 Water quenching (WQ)
[14] Age 720 24 AC
Electro-discharge machining (EDM) (FANUC ROBOCUT C400iB, Oshino, Japan)

with a brass wire having a diameter of 0.2 mm was used to remove the samples from
the base plate and for extracting the metallographic, hardness, and tensile specimens.
Samples having dimensions of 10 × 10 × 10 mm3 were sectioned for microstructural
analysis and hardness testing. Also, sub-size tensile samples with a thickness of 3 mm
were extracted, according to ASTM E8M-04 [10] for a gauge length of 25 mm.
Linear roughness (arithmetic mean height (Ra)) and surface roughness (Sa) were
measured using a portable profilometer (MITUTOYO SURFTEST SJ-210 4MN) and a
Keyence VK-X250 3D laser scanning confocal microscope, respectively. The density
of the as-fabricated sample was measured using the Archimedes method. Samples were
then subjected to four different heat treatments, which are detailed in Table 1. Afterward,
samples were cold mounted and ground planar using 600-grit SiC paper. The samples
were then polished with 9 μm, 3 μm, and 1 μm diamond suspensions, followed by a
final polishing using a 0.05-μm silica polishing suspension (OP-S). To reveal the micro-
structure, electroetching was performed using a 10% oxalic acid solution and 6 V for
5–10 s. A Keyence VK-X250 3D laser scanning confocal microscope was used for
microstructural analysis. The bulk hardness was measured using Rockwell testing
94 S. Atabay et al.
(Wilson 4TT-BB) with a diamond rounded-tip cone indenter (ground at 120°) with a
diameter of 0.2 mm and a minor load of 10 kg, followed by a major load of 150 kg, in
accordance with guidelines for the C-scale. Five independent measurements were
undertaken for each specimen condition to calculate the reported averages.
3 Results and Discussions
Roughness measurements using the surface profilometer (linear) and laser confocal
(area) methods resulted in Ra and Sa values of 5.2 μm and 3.6 μm, respectively for as-
fabricated condition. Scans aligned parallel to the building direction showed the highest
surface roughness. These values are in agreement with the previously reported values
(Sa = 4.5 μm) for LPBF fabricated IN718 alloy [15]. After machining, the Ra and Sa
values were 0.66 μm and 0.46 μm, respectively. These values are nearly an order of
magnitude lower than the as-fabricated roughness. Also, both measurement methods
(linear and surface) resulted in similar roughness values; the slight differences between
the obtained results can be attributed to the difference in the size of the measurement
area. The line method has a smaller number of data points over a larger sample length,
whereas the confocal method has a greater quantity of data points collected over a
smaller sample area.
The as-fabricated IN718 samples had near-full density. Specifically, the Archimedean
density of the as-fabricated samples was measured to be 8.15 g/cm3, which is 99.73%
of the theoretical bulk density of IN718 (8.17 g/cm3) [16]. This indicates that the com-
bination of LPBF parameters selected in this study produce highly dense IN718 with
minimal defects as seen in the as-fabricated polished micrograph (Fig. 2). Overall, the
defects in the as-fabricated IN718 samples were limited to small, spherical pores, which
may be attributed to pre-existing gas porosity in the alloy powder.
Figure 3 shows the preliminary microstructural analysis of the hybrid manufactured
IN718 samples in different conditions. As depicted in Fig. 3a, the as-fabricated micro-
structure consists of columnar grains elongated along the build direction over multiple
build layers. Molten pool boundaries are also visible in the as-fabricated state. Figure 3b
depicts the sub-grain structure of the as-fabricated IN718, which consisted of fine cel-
lular structures. No precipitation could be observed in the as-fabricated IN718 sample
within the resolution limits of the laser optical microscope. Dissolution of the molten
pool boundaries and precipitation were observed when the samples were subjected to
the different heat treatments (specifically HT1, HT2, and HT3). All four of the heat-
treated samples showed two different precipitate morphologies: larger light gray pre-
cipitates at the grain boundaries and smaller black precipitates within the grains. Grain
boundary precipitates have been demarcated in Fig. 3 with red arrows. Even though,
high-resolution microscopy is required to identify and further comment on the mor-
phology, phase composition and size distribution of the precipitates certain differences
can be identified from the representative optical micrographs. As depicted in Fig. 3c,
the size of the grain boundary precipitates is smaller after the HT1 compared to the
other heat treatments. In contrast, samples in the HT2 (Fig. 3d) and HT3 (Fig. 3e) con-
ditions both showed larger discrete precipitates distributed over the grain boundaries. It
Fig. 2. As-polished micrograph of the hybrid manufactured IN718 in as-fabricated condition.
Fig. 3. Optical micrographs of the hybrid manufactured IN718 samples in (a, b) as-fabricated, (c)
HT1, (d) HT2, (e) HT3, and (f) HT4 conditions.
is noteworthy that after HT4, the cellular sub-grain morphology of the as-built sample
was preserved contrary to the other heat treatments.
Rockwell hardness measurements were conducted as a preliminary mechanical
characterization method to understand the effect of the applied heat treatments on the
mechanical properties of the IN718 samples. Average hardness values for each sample
condition are shown in Table 2. The average hardness value of the as-fabricated (AF)
sample was measured as 29.6 HRC; this value is similar to the hardness values reported
for LPBF fabricated IN718 [17]. After heat treatment, the hardness of the IN718 sam-
ples increased by 42–51%. All of the applied heat treatments resulted in a hardness
value within the range of 42–46 HRC, which has been reported in previous studies for
fully heat-treated IN718 fabricated by standalone LPBF processing. HT2 and HT3
96 S. Atabay et al.
Table 2. Rockwell C hardness values of hybrid manufactured IN718 for different sample conditions.
Sample condition AF HT1 HT2 HT3 HT4
Hardness (HRC) 29.6 ± 1.4 42.0 ± 2.2 45.7 ± 1.3 44.8 ± 0.8 43.1 ± 0.7
c onditions showed similar hardness values that were slightly higher than the other two
heat treatments.
4 Conclusion
Hybrid additive/ subtractive manufacturing and post-processing heat treatment of

IN718 was investigated in this study. The as-fabricated roughness values measured in
this study were comparable with standalone LPBF printing technologies. Also, integra-
tion of in-envelop machining to LPBF processing facilitated the application of finishing
operations to meet surface roughness requirements of high-quality/performance prod-
ucts. Preliminary microstructural characterization revealed that the precipitation behav-
ior is affected by the selected heat treatments. The measured hardness of the IN718
samples increased from 29.6 HRC in the as-fabricated condition to 42–46 HRC after
heat treatment. Future research will target establishing a more comprehensive under-
standing of the heat treatments on the microstructure (precipitates, cellular structure)
and mechanical property relationships for hybrid (LPBF) manufactured IN718 through
high-resolution microscopy and tensile testing.
Acknowledgments The authors wish to thank X. Pelletier and M. Guerin of the National Research
Council Canada for their technical assistance in sample preparation, measuring the density, surface
roughness, and tensile properties.
References
1. Reed RC. The superalloys fundamentals and applications. Superalloys Fundam Appl.
2006;9780521859:1–372. https://doi.org/10.1017/CBO9780511541285.
2. Ramakrishna M, Koppoju S, Telasang G, Korla R, Padmanabham G. Effect of solutionizing
temperature on the microstructural evolution during double aging of powder bed fusion-addi-
tive manufactured IN718 alloy. Mater Charact. 2021;172:110868. https://doi.org/10.1016/J.
MATCHAR.2020.110868.
3. DuPont JN, Lippold JC. Welding metallurgy and weldability of nickel-base alloys. J Chem Inf
Model. 2009;53:456. https://doi.org/10.1017/CBO9781107415324.004.
4. Blackwell PL. The mechanical and microstructural characteristics of laser-deposited IN718. J
Mater Process Technol. 2005;170:240–6. https://doi.org/10.1016/j.jmatprotec.2005.05.005.
5. Gruber K, Stopyra W, Kobiela K, Madejski B, Malicki M, Kurzynowski T. Mechanical properties of
Inconel 718 additively manufactured by laser powder bed fusion after industrial high-temperature
heat treatment. J Manuf Process. 2022;73:642–59. https://doi.org/10.1016/j.jmapro.2021.11.053.
6. Li J, Zhao Z, Bai P, Qu H, Liu B, Li L, Wu L, Guan R, Liu H, Guo Z. Microstructural evolution and
mechanical properties of IN718 alloy fabricated by selective laser melting following different heat
treatments. J Alloys Compd. 2019;772:861–70. https://doi.org/10.1016/J.JALLCOM.2018.09.200.
7. Herzog D, Seyda V, Wycisk E, Emmelmann C. Additive manufacturing of metals. Acta Mater.

2016;117:371–92. https://doi.org/10.1016/j.actamat.2016.07.019.
8. Frazier WE. Metal additive manufacturing: a review. J Mater Eng Perform. 2014;23:1917–28.
https://doi.org/10.1007/s11665-014-0958-z.
9. Paccou E, Mokhtari M, Keller C, Nguejio J, Lefebvre W, Sauvage X, Boileau S, Babillot P, Bernard
P, Bauster E. Investigations of powder reusing on microstructure and mechanical properties of
Inconel 718 obtained by additive manufacturing. Mater Sci Eng A. 2021;828:142113. https://doi.
org/10.1016/J.MSEA.2021.142113.
10. A. International. Standard test methods for tension testing of metallic materials. 2018. https://doi.
org/10.1520/E0008.
11. A.C.H.R.A. Committee. Corrosion and heat resistant, bars, forgings, and rings 52.5 Ni 19Cr 3.0
Mo 5.1 Cb 0.90 Ti 0.50 Al 18Fe, consumable electrode or vacuum induction melted 1775 F (968
C) solution heat treated, precipitation hardenable. SAE Int. 1965.
12. AMSF Corrosion Heat Resistant Alloys Committee. Corrosion and heat-resistant, investment cast-
ings, 52.5 Ni-19Cr-3.0 Mo-5.1 Cb (Nb)-0.90 Ti-0.60 Al-18Fe, vacuum melted homogenization and
solution heat treated, AMS5383. Available online https://www.sae.org/standards/content/ams5383.
n.d. Accessed 21 July 2021.
13. A.C.H.R.A. Committee. Alloy bars, forgings, and rings, corrosion and heat resistant nickel base –
19Cr – 3.1Mo – 5.1(Cb + Ta) – 0.90Ti – 0.50Al solution and precipitation heat treated consumable
electrode or vacuum induction melted. 1965. https://doi.org/10.4271/AMS5663. UI – AMS5663.
14. Gallmeyer TG, Moorthy S, Kappes BB, Mills MJ, Amin-Ahmadi B, Stebner AP. Knowledge
of process- structure-property relationships to engineer better heat treatments for laser pow-
der bed fusion additive manufactured Inconel 718. Addit Manuf. 2020;31:100977. https://doi.
org/10.1016/J.ADDMA.2019.100977.
15. Premium EOS NickelAlloy IN718 material data sheet metal solutions. n.d.
16. Wang HP, Zheng CH, Zou PF, Yang SJ, Hu L, Wei B. Density determination and simulation of
Inconel 718 alloy at normal and metastable liquid states. J Mater Sci Technol. 2018;34:436–9.
https://doi.org/10.1016/J.JMST.2017.10.014.
17. Hosseini E, Popovich VA. A review of mechanical properties of additively manufactured Inconel
718. Addit Manuf. 2019;30:100877. https://doi.org/10.1016/J.ADDMA.2019.100877.
Part II Deep Decarbonization Pathways
for Pyrometallurgical Processes: Opportunities
& Challenges
Electrically Heated Fluid Bed as a Low-Carbon
Option for Pyrometallurgical Processes
Kamal Adham1 (*), S. Francey1, and C. McIntyre1

1
Hatch Ltd, Mississauga, ON, Canada
KAdham@Hatch.ca
Abstract. Using a “green electricity” heating source, produced from renewable

resources such as wind/solar/hydro, pyrometallurgical processes can lower their carbon
footprint significantly. Fluid beds are commonly used for drying ores, calcining concen-
trates, and decomposing metallic salts, normally with energy input from fossil fuel com-
bustion. Electrically heated fluid beds can be used instead of the fuel-burning units, as a
low-carbon option for pyrometallurgical processes. Heating can be provided in several
ways, including fluidizing gas-heating, in-bed indirect heating, and in-bed direct heating.
These methods and their applicability to different metallurgical processes are described in
this paper.
Keywords: Fluid bed · Electrical heating · Low carbon
1 Introduction
Carbonaceous fuels are commonly used to heat solids in metallurgical processes such
as drying, calcination, and various high-temperature reactions. Fluid beds are com-
monly used for many of those high-temperature processes, in part due to their excellent
heat transfer capability. Fluid beds can also be heated electrically, and when green
electricity is used, the carbon footprint of these processes can be lowered significantly.
This paper describes three methods of electric heating, fluidizing gas heating, in-bed
indirect heating and in-bed direct heating, and the applicability of these methods to dif-
ferent metallurgical processes.
2 Heating the Fluidizing Gas
In a conventional fluid bed dryer, the fluidizing gas is heated by the combustion of a
carbonaceous fuel in an external burner, sometimes referred to as a hot gas generator.
The flue gas from the hot gas generator is directed to the wind box of the fluid bed dryer.
The hot gas passes through the tuyeres or distributor plate, fluidizing and drying the
solids. As an alternative to combustion, electric resistance heating can be used to heat
the fluidizing gas upstream of the fluid bed dryer, as depicted in Fig. 1. A variety of
https://doi.org/10.1007/978-3-031-17425-4_17
102 K. Adham et al.
electric resistance gas heater configurations are available on the market, such as duct or
in-line heaters and circulation heaters.
The maximum temperature of the gas output is limited by the maximum continuous
operating temperatures that can be sustained by the heating elements in a given gas
environment. For certain gas mixtures, there may be other limitations associated with
the decomposition of gaseous components at high temperatures and potential fouling/
deposition on heating element surfaces.
Efficiencies of conversion of electric energy to useful heat in excess of 95% can be
achieved in electric resistance heating systems. The cost and heater size will depend on
the amount of energy input, the target temperature, operating pressure, and the gas flow
rate and composition. As the large-scale heavy industry seeks out electrification solu-
tions to support decarbonization efforts, the development of higher capacity, high-
temperature electric resistance gas heating systems is being pursued.
In fluid bed dryer applications, there are limitations on the maximum temperature of
the fluidizing gas that can be accepted, and these limits tend to be within the tempera-
ture range that the electric resistance gas heaters can provide. For example, mechanical
constraints for a conventional fluid bed dryer wind box, tuyeres, and distributor plate
typically limit the maximum fluidizing gas temperature to approximately
800 °C. Process-specific considerations may also limit the maximum gas temperature
at the tuyere discharge. In the case of potash drying, for example, gas temperatures in
excess of 600 °C are usually avoided to prevent the sintering of the potash fertilizer.
Similar principles can be considered for fluid bed applications other than drying that
also require a supply of hot fluidizing gas.
Fig. 1. Depiction of fluidizing gas heating upstream of a fluid bed dryer, using (a) fuel-fired hot gas
generator or (b) electric resistance gas heater.
Electrically Heated Fluid Bed as a Low-Carbon Option for Pyrometallurgical Processes 103
3 In-Bed Indirect Heating
3.1 Conventional In-Bed Combustion
In conventional fluid beds, carbonaceous fuels (gaseous, liquid, or solid) can be injected
into the hot solids and made to burn with the upward flow of fluidizing air. Such in-bed
combustion is very effective and efficient, as the energy release is almost completely
and instantaneously absorbed by the solids. For example, gaseous combustion gener-
ally occurs within the first 25–30 mm of bed material [1]; however, this combustion
only occurs when bed temperatures are sufficient to sustain fluidized bed combustion.
Critical bed temperatures are generally above 650 °C [1, 2] before which combustion
will not occur within the particle bed, but instead immediately above it, requiring the
generated heat to pass downwards through the particles. Combustion within the bed
will stabilize at higher temperatures, generally over 750 °C [2].
Fluid beds involving liquid or solid fuels require specialized injectors to supply the
fuel to the bed. Moreover, solid fuel combustion results in ash residue following com-
bustion which demands ash handling systems and potentially solids separation tech-
niques downstream of the fluidized bed [3].
Careful design of fluid beds implementing in-bed combustion is required since the
fluidizing velocity is inherently tied to the desired temperature. As the temperature
set point increases, more fuel is required, thereby changing the hydrodynamic char-
acteristics of the bed [1]. Furthermore, the combustion of large amounts of fuel will
increase the size of the fluid bed reactor, and the downstream off-gas treating
equipment.
3.2 In-Bed Indirect Electric Heating
Electrical heating using in-bed elements (Fig. 2) can be as efficient as combustion, but
its effectiveness depends on the maximum element temperature and its lifespan. An in-
bed electrical heater operates by converting electrical energy to heat energy through the
Joule effect (also known as resistive heating). This heat is transferred to the particles
Fig. 2. Immersion heating element.

104 K. Adham et al.
either directly through particles contacting the element’s surface or indirectly through
convection with the fluidizing gas.
Immersing electric elements into the bed will naturally overcome several of the dif-
ficulties associated to in-bed combustion. By decoupling the temperature control from
the process material balance, off-gas volumes are reduced shrinking downstream equip-
ment, and there is no effect on gas compositions, fluidization regime, or residuals like
ash or coke. In addition, tubular electrical elements (Fig. 3) can offer wider ranges of
temperature control since there is no required minimum combustion temperature.
However, immersing surfaces inside a fluid bed exposes them to high rates of erosion.
Erosion concerns can be mitigated through material selection, placement of the ele-
ments with adequate space above the distributor plate, and maintaining lower velocity
fluidization regimes (ex. bubbling).
While certain electric elements can reach very high temperatures, upwards of
1400 °C, temperatures must be limited within a fluidized bed due to the increased
mechanical loading from the particles. Surface temperatures for the electrical element
should therefore be typically limited to 1000 °C to avoid mechanical damage to the
element.
There must also be careful consideration for local hot spots near the element causing
particle sintering due to constrained movement within the coils. Conventional immersed
tube banks in fluid beds generally use center-to-center spacings of 2–4 times the outer
diameter of the tubes to allow for free-flowing of particles over the surfaces. Additionally,
electric heating elements are not recommended to be used with sticky particles due to
the increase of fouling associated with these materials.
Fig. 3. Cross-section of tubular electric heating element.

4 In-Bed Direct Heating
A number of methods can be used for the electrical in-bed direct heating of a fluidized
bed, including microwave, induction, and ohmic heating. Here, we concentrate on the
more proven ohmic heating, also known as the electrothermal fluid bed, which can be
used for conductive material (e.g., graphite or petroleum coke).
4.1 Electrothermal Fluid Bed
Passage of electricity through a bed of electrically conducting particles can generate a

significant amount of heat. However, the particulate bed should have reasonable resis-
tance to the flow of current and not be prone to arcing. The bed resistance depends on
many parameters, including the chemical nature of the solids, the surface properties and
shape of the particles, the fluid bed voidage, the fluidization velocity, and the properties
of fluidizing gas.
For a given gas (e.g., nitrogen or CO) if the fluidizing velocity is increased, the resis-
tance of the bed increases, but if the fluidizing velocity is too low and the bed is not
uniformly fluidized, localized hot spots can emerge. Ideally, the resistance offered by
the conducting particles generates enough heat within the bed to maintain it in an iso-
thermal condition. A typical electro-thermal fluidized bed reactor [4] equipped with
two electrodes (one in the center and another in a graphite crucible) is depicted in
Fig. 4.
Electrothermal fluidized bed reactors are very energy efficient, as a minimum
amount of external heating or transfer of heat would be required, while the electrical
current heats up the bed in situ. However, a mandatory requirement for a reactor of this
type is a bed of solids which is electrically conducting and does not volatilize during
heating. This problem can sometimes be overcome by mixing a nonconducting solid
feed (e.g., iron ore) with a conducting solid additive (e.g., coke as reductant).
This direct method of heating and reacting minerals in an electrothermal fluidized
bed has been used in a number of pyrometallurgical operations, mostly for the treat-
ment of carbonaceous material. However, significant limitations for applying the elec-
trothermal fluid beds to new processes exist and include:
• Voltage differential leading to arcing and particle sintering
• Conductivity of material required for electrothermal heating
• Material of construction considerations for the electrodes versus the fluidized bed’s
environment, to avoid chemical or physical degradation of the electrodes.
4.2 Treatment of Carbonaceous Material
Many carbonaceous materials are good candidates for the electrothermal fluid beds,
being conductive particulates with low volatilization at high temperatures. Here we
consider two industrial examples of petroleum coke and particulate graphite.
Petroleum coke, as received from the refiners, may not be suitable for making satis-
factory electrodes, unless volatiles are removed by high-temperature calcination. The
106 K. Adham et al.
Fig. 4. General features of a graphite electrothermal fluid bed.
energy necessary for the calcination of petroleum coke can be supplied by burning part
of the coke itself, but it can also be provided by a green electricity source. Based on the
tests in a 1 t/h pilot plant [5], an example case is considered here to demonstrate some
of the key parameters governing the operation of an electrothermal fluid bed for petro-
leum coke. Assuming a maximum electrode voltage differential of 400 V (to avoid
arcing) and an AC power factor of 0.8, a current load of 3125 amp would be required
for every 1 MW energy input. Using approximately 2.2 in2/kW electrode surface area,
the required electrode surface would be about 2200 in2 (1.4 m2).
Natural graphite requires purification, for use in many existing and emerging battery
applications. Purification of particulate graphite, to achieve +99.9% carbon content
with minimum metallic impurities, can be achieved in electrothermal fluid beds, with a
chlorine reagent or with temperature alone.
While thermal purification of graphite at temperatures over 2800 °C is a known
option, such ultra-high temperature furnaces are expensive to build and operate [6].
Ultra-high temperatures are required to volatilize most of the silica and other metallic
oxides, as shown in Table 1.
For the ultra-high temperature fluidized beds used for graphite purification, it is
essential to have excellent thermal insulation and a proper temperature-resistant reactor
vessel design. For most ultra-high temperature reactors, graphite is chosen as the con-
struction material and the column is generally insulated with suitable refractory bricks
or granules.
Table 1. Typical natural graphite impurities’ melting and boiling points.
Oxide impurity Melting point (°C) Boiling point (°C)

Silica (SiO2) 1670 2230
Iron oxide (FeO) 1420 3400
Calcium oxide (CaO) 2570 2850
Fig. 5. Lower electrothermal reactor temperatures can be used with the chlorine reagent, as predicted
by this HSC® chemical equilibrium simulation.
Using chlorine in a lower temperature electrothermal fluid bed is another option to

purify graphite [7]. Chemical reactions that occur between the graphite impurities and
the chlorine gas are generally categorized as carbochlorination, where some of the car-
bon value that is present in excess in the graphite structure, reduces the oxides (turning
into CO) and the chlorine gas reacts with the metallic values to turn them into volatile
chlorides (as shown in Fig. 5). Most of the metallic impurities can be chlorinated and
volatilized at lower temperatures than 1000 °C, but some (e.g., Si) require higher tem-
peratures (up to 1500 °C). However, previous processes have achieved limited com-
mercial success because of the costs associated with chlorine usage (cost and equipment
damage) and other chemicals required to treat the reactor’s offgas. Chlorine treatment
of natural graphite resources can only be commercially successful if these drawbacks
are addressed through process and equipment improvements.
References
1. Gao W, Kong L, Hodgson P. Fluidized-bed heat treating equipment. 2014. https://doi.org/10.31399/

asm.hb.v04b.a0005927.
2. Friedman J, Li H. Natural gas combustion in a fluidized bed heat-treating furnace. Combust Sci
Technol. 2005;177(11):2211–41. https://doi.org/10.1080/00102200500240844.
3. Yang WC. Handbook of fluidization and fluid-particle systems. 1st ed. Boca Raton: CRC Press;
2003. https://doi.org/10.1201/9780203912744.
4. Gupta CK, Sathiyamoorthy D. Fluid bed technology in materials processing. Boca Raton: CRC
Press; 1998. p. 310–1.
108 K. Adham et al.
5. Paquet JL, Foulkes PB. Calcination of fluid coke in an electrically heated fluidized bed. Can J
Chem Eng. 1965;43:94–6.
6. Fedorov SS, for Brookhaven National Laboratory, et al. Ultra-high temperature continuous reac-
tors based on electro-thermal fluidized bed concept. J Fluids Eng-Trans ASME. 2016:3–47.
7. Adham K, Bowes G. Natural graphite purification through chlorination in fluidized bed reactor. In:
Extraction 2018. The minerals, metals & materials series. 2018. p. 2505–12.
Circored Fine Ore Direct Reduction Plus DRI
Smelting – One Route Towards Green Steel
Sebastian Lang1, Max Köpf1 (*), and Timo Haimi2
Outotec GmbH & Co KG, Oberursel, Germany

1
2
Metso Outotec Oyi, Helsinki, Finland
sebastian.lang@mogroup.com; maximilian.koepf@mogroup.com;
timo.haimi@mogroup.com
Abstract. The hydrogen-based direct reduction of iron ore combined with EAF smelt-
ing is being widely discussed as a possible replacement for the commonly used BF/
BOF route in steelmaking when targeting carbon footprint reduction. One alternative
to shaft furnaces is the Circored process of Metso Outotec, which uses fine ore as feed
for fluidized bed reactors and is eliminating so the cost and energy-intensive pelletizing
step. As a direct reduction process using 100% hydrogen as the reductant, Circored has
already proven its functionality in an industrial-scale demonstration plant.
For the smelting of DRI Metso Outotec can use its rectangular six-in-line smelt-
ing furnace, which is based on a flash smelter body and Söderberg electrodes. This
DRI smelting solution is able to replace small-/medium-sized BFs and produce hot
metal with the desired carbon content in existing steel plants with BOF converters.
The large furnace volume enables the processing of low-grade iron ore with high
gangue content and thus the use of BF-grade feed for DRI production.
Keywords: Hydrogen · Direct reduction · DRI · HBI · DRI smelting
1 Introduction
In 2019, the United Nations announced that over 60 countries had committed to carbon
neutrality by 2050. These targets were confirmed in 2021 at the COP 26 meeting, sig-
nificantly accelerating the dynamics for the development of carbon-neutral or green
steelmaking technologies. With limited investment cycles left until 2050, the steelmak-
ing industry must decide within the next 5–10 years in which alternative technology for
the reduction of iron ore to invest.
The hydrogen-based direct reduction of iron ore combined with EAF smelting is
being widely discussed as a possible replacement for the commonly used BF/BOF
route in steelmaking to decrease the carbon footprint over the steel value chain.
However, applying this approach will generate the need for several hundred million
tons of additional high-quality DR-grade pellets annually, with the associated impacts
on the demand for high-grade iron ore concentrates, pellet availability, and conse-
quently pellet prices. One alternative to shaft furnaces is Metso Outotec’s Circored
https://doi.org/10.1007/978-3-031-17425-4_18
110 S. Lang et al.
process, which uses fine ore as feed for fluidized bed reactors, eliminating the cost and
energy-intensive pelletizing step. As a direct reduction process using 100% hydrogen as
a reductant, Circored has already proven its functionality in an industrial-scale demon-
stration plant. Direct charging of hot DRI to a smelter would further increase the energy
efficiency of the process and opens up for low-grade iron ores as feed materials.
As frequently discussed steelmaking route the direct reduction in shaft furnaces fol-
lowed by EAF-steelmaking relies on the use of high-grade iron ore, it is evident that
this will increase the already visible shortage of such feed materials. The Metso Outotec
rectangular six-in-line smelting furnace combines a flash smelter body with Söderberg
electrodes. The large furnace volume enables the processing of pre-reduced low-grade
iron ores with, respectively, high gangue content and consequently big slag volumes. So
it is a way to use BF-grade feed for DRI production and delivering pig iron into the
present steel value chain. This DRI smelting solution can replace small-/medium-sized
BFs and produce hot metal with the desired carbon content in existing steel plants with
BOF converters.
This concept is new in the global framework and can be very competitive regarding
both investment costs and technological risks and is an important enabler in decarbon-
izing steel production, thus contributing directly to the transition toward green steel.
2 The Circored Process
Circored is based on fluid bed technology, a field in which Metso Outotec has been
leading the development now for several decades. The technology is state-of-the-art for
industrial processes such as coal combustion in circulating fluidized beds (CFB) and
alumina calcination. The main advantages of fluidized bed processes are excellent heat
and mass transfer conditions, precise temperature control, short solids retention times
leading to higher plant capacities, and lower investment and operating costs.
The technology can also be applied for the direct reduction of iron ore to eliminate
the cost and energy-intensive agglomeration step of iron ore fines in the form of sinter-
ing or pelletizing.
The generic Circored process is based on the reduction behavior of iron ore, as
shown in Fig. 1.
For the direct reduction of fine iron ore, Circored applies a two-staged reactor con-
figuration with a CFB followed by a bubbling fluidized bed (FB) downstream. Below is
a brief description of the Circored process, which is also visualized in the simplified
process flow diagram in Fig. 2.
The preferred grain size for the process is 0.1–2.0 mm, though depending on decrep-
itation behavior, a grain size of up to 6 mm might also be acceptable. Ultrafine concen-
trate and in-plant fines can be microgranulated. The iron ore fines are dried and
preheated in a CFB preheater to a temperature of approximately 850–900 °C before
being introduced into the first-stage CFB reactor.
The initial reduction step is fast and controlled by the outer mass transfer of the
reductant to the iron oxide particle. The CFB is the ideal reactor to achieve a prereduc-
tion degree of 65–80%, offering the following characteristics: high gas velocities of
Circored Fine Ore Direct Reduction Plus DRI Smelting – One Route Towards Green Steel 111
Fig. 1. Performance of H2-based direct reduction process in fluidized beds.
Fig. 2. A simplified process flow diagram of the Circored process including microgranulation.
112 S. Lang et al.
4–6 m/s and high differential velocities between gases and solids leading to short solids
retention times of 20–30 minutes, and the optimum lateral and vertical mixing of solids
and gases ensuring uniform temperature distribution throughout the reactor.
For processing ultrafine ores (<50 μm) or scrubber dust, Metso Outotec has patented
a very simple microgranulation process. In this process, the ultrafine particles are
agglomerated under the addition of a binder to microgranules of an average size of
approximately 350 μm. This process variant does not require any additional equipment
as the hardening of the granules takes place in the preheating section of the Circored
plant (also shown in Fig. 2).
After prereduction, the material is discharged into the secondary FB reactor. For the
prolonged diffusion-controlled final reduction step, the bubbling FB reactor offers opti-
mum conditions for achieving reduction degrees in excess of 93–95%, with low gas
velocities of 0.5–0.6 m/s and longer solids retention times of 60–180 minutes depend-
ing on the nature of the ore.
For the Circored process, hydrogen was selected as the sole reducing agent. Its spe-
cific reduction reaction temperature characteristic lowers the temperature of the process
to 630–650 °C and avoids particles sticking in the reduced material. Because of this, at
the time of its development, hydrogen was the enabler for a new process; today, with
the steel industry under pressure to decarbonize, the use of hydrogen as the sole reduc-
tant is nowadays getting much more significance. From this perspective, one could say
that the hydrogen-based Circored process was developed 20 years too early.
As the reduction of iron ore with hydrogen is an endothermic reaction, the energy
must be supplied by heating both the ore and process gas. In the past, this was done
using natural gas and process bleed gas. To achieve a totally carbon emission-free pro-
cess, it is now foreseen to replace natural gas with electric heating powered by (green)
renewable energy.
A further significant advantage of the CFB reactor is the ability to inject dust-laden,
partially reacted gas from the second-stage FB reactor, which creates a counter-current
flow of gases and solids throughout the dual-reactor system. This has a twofold advan-
tage: the nozzle grate of the CFB reactor is only exposed to clean recycled process gas,
thereby minimizing nozzle wear, and the overall gas utilization of the entire process is
maximized. The Circored process generates no residues or relevant emissions besides
water and unavoidable heat release.
Key Features of the Circored Process

• Preheating the iron ore fines to 850–900 °C in a separate CFB reactor for cal-
cining prior to charging in the primary CFB reduction stage. This method also
allows the processing of magnetite ores, which can be difficult to reduce as the
preheating stage oxidizes the magnetite to hematite. In addition, the micro-
granules produced can be hardened to avoid the generation of ultrafine parti-
cles in the reduction stage.
• Prereduction in a CFB in about 20–30 minutes to a reduction degree of
65–80%.
• Final reduction in a compartmentalized FB reactor to 93–95% reduction.

• The use of hydrogen as the sole reductant enables low temperatures of 630–
650 °C in the CFB and the FB; this low temperature avoids particle sticking
and means that the reaction is easy to control.
• A zero-carbon process variant is possible with electrical heating from renew-
able sources to provide the energy for the endothermic reduction reactions.
2.1 Comparison of Different Steelmaking Routes
When comparing Circored/EAF steelmaking with other state-of-the-art routes, espe-

cially in terms of CO2 emissions and cost, common battery limits need to be defined.
We look here at the conversion of an iron ore concentrate to raw steel with all the
required intermediate steps, forming the references for this comparison (see Fig. 3).
Both Circored and shaft furnace direct reduction technologies produce DRI or HBI
which can directly be used in an EAF. The product from the blast furnace, pig iron, is
fed to an oxygen converter (BOF). All these downstream processes are considered in
the CO2 and cost figures.
If an ultrafine concentrate is used, a microgranulation step should be applied to nar-
row down the particle size distribution and make the material properly fluidizable.
Along with this option comes the opportunity to recycle any kind of dust or fines origi-
nating at different points in the plant. In a classical set up, the Circored process would
use a natural gas reformer to provide hydrogen for the reduction. In future scenarios,
green hydrogen produced from renewable sources is used for direct reduction and elec-
tric heating.
It is apparent that the BF/BOF route is by far the largest emitter of CO2; furthermore,
the technical solutions to minimize emissions are limited for this route. While the CO2
Fig. 3. A comparison of the main components, feeds, and products of different steelmaking routes.
114 S. Lang et al.
emissions of the Circored process and shaft furnace direct reduction are in the same
order of magnitude, Circored benefits from the omission of the pelletizing step. As
Circored+ (fully decarbonized process variant) is designed to produce real green steel,
the CO2 emissions of the process become negligible.
When it comes to cost, Circored is a very competitive route, partly because the CO2
taxation costs are lower than for the BF/BOF route and because there is no pellet
premium compared to the shaft furnace route. In terms of CO2 emissions and cost, the
“regular” Circored process is already competitive today, and the Circored + variant is
predicted to be even more so in future.
3 The Circored Plant in Trinidad
There were two main drivers for the development of the Circored process: the growing
demand for a direct reduction process utilizing iron ore fines directly to decrease HBI
production cost by avoiding a costly agglomeration step, and Metso Outotec predeces-
sor’s desire to apply its vast experience of utilizing fluidized bed technology also for
direct reduction.
In 1996, the contract for the first Circored plant, to be built in Trinidad, was awarded
by Cliffs and Associates Limited. The plant started operation in May 1999; after the
discharge system was modified it reached its process design parameters in March 2001
and was operating at up to 105% of its design capacity.
Despite functioning normally, the plant was unfortunately idled after a short period
of successful operation, which produced 300,000 tons of high-quality HBI (Fig. 4).
This was due to several changes in ownership and to economic and political reasons
including steel-market developments and the lack of availability of natural gas. The
Trinidad plant fulfilled process expectations in terms of a high and uniform degree of
metallization achieved from the outset.
Key Achievements of the Circored Plant in Trinidad

• Over 300,000 tons of high-quality HBI were produced and were subsequently
processed in electric arc furnaces located in the US.
• Plant design HBI production of 63 t/h periodically exceeded.
• High HBI product quality with maximum metallization degrees greater than
95% and constant briquette densities above 5.2 g/cm3.
Since the design, erection, and operation of the first Circored plant in Trinidad,
numerous modifications to the original setup were investigated and developed:
• Plant capacity increased to 1.25 Mt/year per line, now considered the technical and
economical optimum for an industrial-scale plant.
• Replacing natural gas with green electricity for preheating to achieve complete car-
bon neutrality.
Fig. 4. Trinidad plant with HBI stockpile.
• Microgranulation: for processing ultrafine (<50 μm) ores and scrubber dust, Metso
Outotec has patented a very simple microgranulation process. In this process, the
ultrafine particles are agglomerated to microgranules to an average size of approxi-
mately 350 μm with the addition of a binder. The process does not require any addi-
tional heat-hardening equipment as the hardening of the granules takes place in the
preheating section of the Circored plant.
• Direct feeding of hot DRI into an EAF to further improve energy efficiency.
• For low-grade iron ores, a combination of a single reduction stage Circored process
(metallization degree of 75–85%) with smelting reduction in an electric smelter for
hot metal production is feasible.
4 DRI Smelting
By utilizing the existing products and wide experience of the Metso Outotec Ferroalloys
and Non-Ferrous Smelting organization, a new solution has been developed for smelt-
ing DRI that has been produced from low-grade iron ore to produce hot metal for fur-
ther use in BOFs. Such DRI feed would be a challenge for standard steelmaking EAFs
because of the excess slag volume, which causes unacceptable losses of Fe units to the
slag. In addition, incorporating this type of process into existing steel plant facilities
using BOF converters can be a challenge, at least on a larger scale and for more demand-
ing steel grades.
116 S. Lang et al.
The Metso Outotec solution, therefore, aims to replace the blast furnace and achieve
continuous production of suitable hot metal quality for an existing steel plant, to enable
the use of existing BOF converters. The main idea is to use a closed EAF for the smelt-
ing of the pre-reduced iron ore feed (DRI from Circored). For fossil-free production,
sustainable sources of carbon reductant would be needed for the smelting furnace, since
the final reduction takes place there.
The main driver of this solution is that there is no need to use high-quality DRI-
grade pellets or to increase the usage of external (high-quality) scrap. It can use DRI
produced from standard BF-grade pellets by direct reduction in a shaft furnace or
Circored-DRI produced from low-grade iron ore fines and even utilize recycling of
waste material like dust briquettes. Also, the slag can be modified to remove impurities
and achieve the desired slag chemistry, targeting very good metal yields and good-
quality slag products that can be used as raw materials by the cement industry. Furnace
off-gas can be either utilized as combustible off-gas (in a similar way to coke oven gas/
blast furnace gas in the possible existing power plant) or combusted to produce steam
for other heating purposes.
The Metso Outotec DRI smelting furnace design uses six self-baking Söderberg
electrodes, which are standard in ferroalloy and non-ferrous processes. The design of
the furnace body is well known from in flash smelting furnaces and proven in numerous
installed plants. Therefore, all equipment is proven technology and has many reference
projects. Figure 5 shows an illustration of the furnace.
Fig. 5. Metso Outotec DRI smelting furnace design.

The feed can be fed into the feed bins on top of the smelting furnace, either hot or
cold. There are several possibilities for the feeding system, but the most commonly
used systems are conveyors and buckets. The hot feeding is highly efficient in terms of
electrical energy consumption since a feed temperature of <600 °C reduces specific
smelting electrical energy consumption by as much as 20%. In the case of hot feed, the
reductants and fluxes need to be fed from separate bins to the hot DRI.
The furnace is lined with refractories and cooled with water via different copper
cooling elements.
The carbon and silicon levels of the hot metal produced with the furnace need to be
configured to the needs of each steel plant. Additionally, all DRI pellets use different
iron ore feed, which means that the solution needs to be optimized according to the
needs of each individual steel plant.
5 Perspectives
With the Circored process, Metso Outotec provides an alternative process route for the
reduction of iron ores, using hydrogen as the sole reductant and fine ore instead of pel-
lets as feed material. Circored is currently the only process for iron ore direct reduction
based on pure hydrogen and has proven its functionality and performance in an
industrial-scale demonstration plant with a capacity of 500,000 t HBI/year.
Circored technology is flexible in its production setup. Besides merchant cold HBI,
which can easily be shipped, hot and cold DRI can be produced and directly linked to
EAFs and BOFs as a substitute for hot metal and/or other virgin iron units. This guar-
antees the production of high-quality steel products that have traditionally been the
strength of integrated steel plants.
As the frequently discussed steelmaking route direct reduction in shaft furnaces fol-
lowed by EAF-steelmaking relies on the use of very clean raw material, it is evident that
this will increase the already visible shortage of high-quality, clean raw material. The
Metso Outotec DRI smelting furnace to produce hot metal would be able to handle
large slag volumes without excess Fe-unit losses, thereby allowing the use of lower-
quality raw materials. This DRI smelting solution can replace small-/medium-sized
BFs and produce hot metal with the desired carbon content in existing steel plants with
BOF converters.
High-Temperature Hydrogen Reduction of Bauxite
Residue for Iron Recovery
Arman Hoseinpur1 (*), Per Ivar Friborg1, Casper Van Der Eijk2, and Jafar Safarian1
1
Department of Materials Science and Engineering, Norwegian University of Science and
Technology (NTNU), Trondheim, Norway
arman.h.kermani@ntnu.no
2
SINTEF Industry, Trondheim, Norway
Abstract. In order to develop a low-carbon footprint process for industrial waste valori-
zation of bauxite residue (BR), the hydrogen reduction of a mixture of BR (61 wt.%) and
limestone (39 wt.%) was studied for the purpose of simultaneous production of iron and
calcium aluminates. In this research, pellets were made and subsequently sintered at
1200 °C. The pellets were then reduced in pure hydrogen in a vertical tube furnace at
1170 °C, where hydrogen was introduced from the bottom to the pellets bed. The prod-
ucts of the reduction were characterized by X-ray diffraction (XRD) technique, which
revealed the reduction of iron oxide phases to metallic iron. The co-existing phases were
found to be 12CaO.7Al2O3 (Mayenite), CaSiO4, and CaTiO3 phases. Results indicated the
fluxing of the BR by lime prevented the iron trapping in aluminate phases and hence all
the iron contents can be reduced to metallic iron. Microstructural analysis was done by
electron microscopy (SEM) on the sintered pellets and the corresponding reduced pellets.
The preliminary results presented in this paper indicated the feasibility of iron production
out of BR by this method.
Keywords: Iron recovery · Bauxite residue · Hydrogen · Solid-state · Reduction
1 Introduction
Bauxite residue (BR) or red mud has been a long-lasting environmental problem in the
alumina production from bauxite via the Bayer process [1–3]. It is said that the current
world stockpile of BR is 4 billion tons and almost 175.5 million tons of BR adds annu-
ally to these stockpiles [4]. BR has high PH and hence is harmful if it leaks to the nature.
The environmental issues related to the BR, and the catastrophes happened over last
decades could be a driving force to think about the valorization of BR; however, consid-
ering the high amount of metal oxides such iron oxide, alumina, and titanium oxides, and
rare earth elements, and the fact that processing of BR has no mining costs, the research-
ers consider BR as a main secondary source for these elements. Especially when it
comes to the rare earth elements, whose market is monopolized by China (over 78%) [4].
Iron oxide is the main constituent of BR for many stockpiles in the world and those
that can be found in Europe are rich in iron oxide such as 44.8% in Germany and 42%
in Greece [4]. Therefore, recovery of iron will be an integrate part of any scenario for
https://doi.org/10.1007/978-3-031-17425-4_19
120 A. Hoseinpur et al.
valorization of BR. However, the application of hydrogen as a green substitute for car-
bon (or CO gas) as the reductant to reduce iron oxide of BR makes the hydrogen reduc-
tion a sustainable BR valorization alternative. When it comes to the valorization of the
BR, the other compound of high importance to be recovered is alumina. Considering
the depletion of the high-concentration bauxite mines (with alumina content >30%), the
recovery of alumina from BR has become economic. It is reported that the alumina
content in several BR stockpiles over the world is higher than 20 wt.% and in some
cases even up to be 25 wt.%. Considering the amount of the piled BR in each stock, the
BR should be respected as a secondary source for alumina in close future. Recovery of
alumina via the Pederson route has been reinvestigated by NTNU and SINTEF to estab-
lish a modernized Pedersen process for the production of alumina from calcium alumi-
nates slags [1–3, 5]. Using the experiences from the previous activities in ENSUREAL
project [6], an approach for valorization of BR is being perused in the HARARE
Eu-project [7], with the hydrogen reduction of BR that is already mixed with limestone
leading to the production of iron and calcium aluminates. In HARARE, the solid-state
hydrogen reduction of BR that is already mixed with limestone is being studied as an
approach to valorize BR and recover its iron content. The calcium aluminates produced
in hydrogen reduction is further hydrometallurgical processed to produce alumina and
REEs in HARARE [2]. Addition of limestone to the BR may provide a full reduction of
iron unless some amount of iron oxide may remain trapped in the form of iron alumi-
nate phases (FeAl2O4) [8]. This paper presents the preliminary results of the hydrogen
reduction of BR fluxed with limestone for the purpose of iron recovery. The overall
scheme for the valorization of the bauxite residue in the HARARE approach via a
Ca-route is presented in Fig. 1. As shown in this figure, the produced pellets from BR
and limestone will go through the pelletizing and sintering processes. In the sintering
process, CO2 will be the main gas emitting from the samples, which is mainly the result
of the calcite decompaction. The leaching of the similar oxide mixtures including
Mayenite phase is already investigated [2] and it is shown the injection of CO2 is
required for the decomposition of calcium aluminate to calcite and high-quality alu-
mina. The mass balance calculations indicate when starting with 1000 kg bauxite resi-
due and 640 kg calcite, about 355 kg CO2 emits to environment, from which 195 kg will
be used in the leaching processes for regeneration of calcite. Considering a close loop
system, a part of the generated CO2 in the sintering step can be used in the final leaching
process. This reduces the carbon footprint of the HARARE process. The synthesized
calcite in end of the loop will be streamed back to the beginning of the process to be
mixed with the bauxite residue. Mass balance calculations indicate that 595 kg of the
Fig. 1. The HARARE approach for valorization of bauxite residue.

High-Temperature Hydrogen Reduction of Bauxite Residue for Iron Recovery 121
640 kg limestone can be returned from the end of the process if it has low number of
impurities, hence dependency of the process to the raw material reduces. Another prod-
uct of the leaching processes will be approximately 375 kg grey mud, potentially
cement clinker raw material (55 wt.% CaO, 22 wt.% SiO2, Al2O3 wt.%, 16 wt.%) and
0.1 kg of the valuable scandium oxide. The hydrogen reduciton of the lime fluxed
BR pellets can be done in isothermal and non-isothermal regimes. Results from iso-
thermal reduction regime is presented in reference [9] and this study presents results
from non-isothermal reduction regime.
In this research, bauxite residue (BR) was provided by Mytilineos S.A, Metallurgy
Business Unit (former Aluminum of Greece). BR (1000 g) was mixed with limestone
(640 g) and pelletized in a lab-scale pelletizing drum. Preliminary experiments showed
when doing the reduction directly on the pellets, lots of gas (humidity) releases in the
reduction furnace over 400–600 °C leading to pressure built-up, pulverizing of the pel-
lets, and practical difficulties in the hydrogen reduction process. Hence, a sintering step
was found beneficial for making the pellets rigid and solid enough so they tolerate high-
temperature reduction step. Hence, the green pellets were dried at 60 °C overnight and
then were sintered in a muffle furnace at 1200 °C at ambient atmosphere for 2 hours.
Thermogravimetry analysis (TGA, Linseis PT1600, Germany) was used for analysis of
the sintering process. For this purpose, a pellet was heated in Ar atmosphere while heat-
ing by 5 °C/min and up to 1200 °C. Table 1 shows the XRF analysis of a sample from
the pellets after the sintering step. The sintered pellets were then subjected to hydrogen
reduction in a vertical tube furnace with an alumina tube chamber. In the hydrogen
reduction experiments, 25 g of the sintered pellets was charged into a molybdenum
crucible (Fig. 1) with several holes in the bottom for hydrogen flow (0.15 NL/min) to
the pellet bed. The off-gas valve on top of the furnace was kept open during the experi-
ments so that pure hydrogen was introduced below the pellet bed and moved upward
through the bed. The reduction experiments were carried out with a heating rate of
10 °C/min and holding at the 1170 °C for 1 hour. Once the reduction was finished, Ar
was blown to the bed to cool down the reduced sample with about 10–15 °C/min. The
products of the reduction experiments were characterized by X-ray diffraction (XRD)
and scanning electron microscopy (SEM) techniques (Fig. 2).
3 Results
3.1 TGA Analysis of Pellets
The preliminary experiments with the dried pellets (39 wt.% limestone and 61 wt.%
BR) made it clear that a sintering step was required. The TGA analysis shown in Fig. 3
indicates that the sample has lost approximately 20 wt.% of its initial weight when
Table 1. The XRF analysis of the sintered pellets.
Oxides Wt.%
CaO 33.20
Fe2O3 30.60
MgO 0.33
SiO2 6.69
Al2O3 19.80
MnO 0.04
Cr2O3 0.21
V2O5 0.16
TiO2 4.08
NiO 0.09
Na2O 3.54
SrO 0.11
ZrO2 0.11
Fig. 2. Schematic of the reduction bed, inner diameter of the crucible is 30 mm.
heating the dried pellets up to 1200 °C. As the TGA shows, there is no any significant
weight change beyond 800 °C and the total mass loss up to 1200 °C was recorded
20 wt.%. The TGA analysis makes it clear that the mass loss takes place in three differ-
ent steps, over three various temperature ranges, and each can be explained as
follows:
(a) Step I: 1 wt.% loss over 270–350 °C, which could be associated with the decompo-
sition of iron oxyhydroxide (FeO.OH) which takes place at 275 °C [10].
(b) Step II: 2 wt.% loss over 350–550 °C; this could be associated with the decomposi-
tion of diaspore (AlO.OH) taking place between 400 and 500 °C [10].
(c) Step III: 16.63 wt.% loss over 550–800 °C, which can be associated with the
decomposition of calcite (CaCO3) taking place between 600 and 830 °C.
During this heat treatment, the calcium carbonate decomposes to CaO and CO2 and
metal oxyhydrates decompose to metal oxides and water vapor. This leads to the forma-
tion of lots of micropores and cavities in the pellets’ bodies.
3.2 Sintering and Reduction Steps
The same mass loss obtained in TGA analysis was observed when sintering 600 g of the
pellets in a muffle furnace at ambient atmosphere. Figure 4a shows the XRD pattern of
the sintered sample, which consists of Gehlenite, Brownmillerite, Larnite, and silica
phases. Further investigations of the microstructure were carried out by SEM. The SEM
analysis of the sintered pellets (Fig. 5) reveals the porosities and cavities in the pellets’
bodies. As shown in Fig. 5, various phases were detected in Back Scattered Electron
(BSE) modes with different morphologies and contrasts. Characterizing these phases
with EDS point analysis technique made it clear that the matrix phase with the brightest
color is Ca-Fe-O with 25 At.% of Fe and minor amounts of Al, Na, and Ti dissolved into
the structure, and Brownmillerite (CaFe2O5) peaks nominated in the XRD pattern could
represent this phase. The spherical phases contained Ca-Ti-Fe-O with 13–14 At.% of Fe
and minor amounts of Si and Al dissolved in the structure. The Ca3Fe2TiO8 peaks
detected in the XRD pattern (Fig. 4a) can then represent this phase. Finally, the dark
1 Heating rate: 5°C/min

0
-1 Mass change -19.6063 %
-3
-5
-6
-7
Rel. mass change (%)
-8
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
0 200 400 600 800 1000 1200

Temperature (°C)
Fig. 3. The TGA analysis of the mixture (39% limestone and 61% Bauxite residue), under argon
atmosphere.
briquets and rectangular phases were characterized as Al-Ca-Si oxides with 3.5 At%
dissolved Fe in the structure and can be represented by the Gehlenite (Al2Ca2SiO7) in
the XRD pattern shown in Fig. 4a. Therefore, the investigations of the sintered pellets
made it clear that Fe is mainly in the Brownmillerite and then Gehlenite phases, while
small amount of iron was also possible to be detected in the matrix phase.
Ca3Fe2TiO8 peaks detected in the XRD pattern of the sintered sample can then rep-
resent this phase. Finally, the dark briquets and rectangular phases were characterized
as Al-Ca-Si oxides with 3.5 At% dissolved Fe in the structure and can be represented by
the Gehlenite (Al2Ca2SiO7) in the XRD.
XRD pattern of the reduced sample by hydrogen at 1170 °C and 1 hour can be seen
in Fig. 4b.When comparing the reduced sample with the sintered sample, it is obvious
that the reduction has yield iron in the sample, and the reduced sample also consists of
Mayenite, and calcium titanate, indicating on the reduction of iron oxide. From the
XRD pattern, it seems all the iron oxides are reduced, but further studies were carried
out by SEM, as presented in Fig. 5. As shown in Fig. 5, the produced iron grains can be
seen in form of clusters close to each other in the oxide matrix. Other bright phases are
seen in the matrix with a needle shape (named as phase 1 on the figure) and the EDS
point analysis indicated a chemical composition of 24 At.% Ca, 8 At.% Fe, 8 At.% Al,
7 At.% Ti, and 50 At.% O, which then gives the chemical formula of Ca3FeAlTiO6. As
shown in the SEM image, this phase comprises a considerable amount of the oxide
matrix. It is worth to mention that it was not possible to find a XRD pattern representing
Ca3FeAlTiO6 phase in the pattern diffraction file directory (version PDF 4). The XRD
pattern of Ca3Fe0.4Ti0.6O2.8 was well fitted with some the peaks of the reduced sample,
and the Ca3Fe2TiO8 phase was not able to be recognized after the reduction. This shows
that the reduction of Ca-Fe-Ti-O was not completed after 1 hour reduction, and longer
times are required for a full reduction. In addition, mostly next to the iron grains a
bright phase could be detected which is marked as 2 on the SEM image depicted in
Fig. 4. XRD patterns of the sintered pellets (a) compared with the samples after reduction at 1170 for
1 hour (b).
Fig. 5. Microstructure of the sintered pellets characterized by SEM, the major phases are named.
(a) microstructre at lower magnification showing the pores and cavities, (b) microstructure at higher
magnifications representing the sintered phases
Fig. 6. The EDS point analysis revealed that this phase is Wüstite (FeO) phase with dis-
solved Mg (approximately 10 At.%) in its structure. It should be mentioned here that at
some regions of the pellets after the reduction there was no intermediate phases as
Wüstite and phase #1 and the microstructure was mainly iron and Mayenite.
These results indicate the feasibility of the iron production from the mixture of BR
and limestone by hydrogen reduction and it shows that the iron reduction process is
time consuming. Results indicate there are three phases in the sintered pellets that con-
tain iron (Brownmillerite, Gehlenite, and Ca3Fe2TiO8), and after 1 hour of hydrogen
reduction at 1170 °C, the reduction process is still not finished. The Ca-Fe-Ti phases are
the last phases to be fully converted during the reduction course. Further studies on the
hydrogen reduction can be done by changing parameters such as temperature, time,
ratio of limestone over BR, mode of reduction (isotherm / non-isotherm), and also trial
in hydrogen reduction rotary furnaces.
4 Conclusions
Hydrogen reduction of sintered pellets, made from bauxite residue and limestone, was
carried out and the below main remarks can be concluded:
Fig. 6. The microstructure of the reduced pellets at 1170 °C and for 1 hour.
• The products of the hydrogen reduction are metallic iron and Mayenite phases (the
alumina content of the bauxite residue was mostly converted to Mayenite).
• The results indicated after 1 hour of hydrogen reduction at 1170 °C, the reduction is
not completed and considerable Ca-Fe-Ti oxides and small amounts of Wüstite
phase with dissolved Mg can be found in the matrix; longer reduction time is neces-
sary for complete reduction.
• The reduced iron grains were formed mostly adjacent to the evolved Mayenite
grains; however, most of them were close together in cluster shapes.
Acknowledgments This research was funded by European Union’s Horizon 2020 research and inno-
vation program under grant agreement No 958307. The authors of this paper are thankful to Steinar
Prytz and Birgitte Sofie Carlsen from SINTEF for making the pellets and preparing the metallography
samples.
References
1. Lazou A, Van Der Eijk C, Tang K, Balomenos E, Kolbeinsen L, Safarian J. The utilization of baux-
ite residue with a calcite-rich bauxite ore in the Pedersen process for iron and alumina extraction.
Metall Mater Trans B Process Metall Mater Process Sci. 2021;52:1255–66. https://doi.org/10.1007/
s11663-021-02086-w.
2. Azof FI, Vafeias M, Panias D, Safarian J. The leachability of a ternary CaO-Al2O3-SiO2 slag
produced from smelting-reduction of low-grade bauxite for alumina recovery. Hydrometallurgy.
2020;191:105184. https://doi.org/10.1016/j.hydromet.2019.105184.
3. Azof FI, Kolbeinsen L, Safarian J. Characteristics of calcium-aluminate slags and pig iron pro-
duced from smelting-reduction of low-grade bauxites. Metall Mater Trans B Process Metall Mater
Process Sci. 2018;49:2400–20. https://doi.org/10.1007/s11663-018-1353-1.
4. Archambo M, Kawatra SK. Red mud: fundamentals and new avenues for utilization. Miner Process
Extr Metall Rev. 2021;42:427–50. https://doi.org/10.1080/08827508.2020.1781109.
5. Azof FI, Tang K, You J, Safarian J. Synthesis and characterization of 12CaO⋅7Al2O3 slags: the
effects of impurities and atmospheres on the phase relations. Metall Mater Trans B Process Metall
Mater Process Sci. 2020;51:2689–710. https://doi.org/10.1007/s11663-020-01969-8.
6. ENSUREAL – sustainable alumina production. 2021. https://ensureal.com/. Accessed 31 Dec 2021.
7. h2020harare.eu. Hydrogen-based metal recovery from waste. https://h2020harare.eu/. 2021.
Accessed 31 Dec 2021.
8. Lazou A, van der Eijk C, Balomenos E, Kolbeinsen L, Safarian J. On the direct reduction phenom-
ena of bauxite ore using H2 gas in a fixed bed reactor. J Sustain Metall. 2020;6:227–38. https://doi.
org/10.1007/s40831-020-00268-5.
9. Skibelid OB, Velle SO, Vollan F, Van Der Eijk C, Hoseinpur-Kermani A, Safarian J. Isothermal
Hydrogen Reduction of a Lime-Added Bauxite Residue Agglomerate at Elevated Temperatures for
Iron and Alumina Recovery, Materials. 2022;15(17):6012. https://doi.org/10.3390/ma15176012.
10. Yang H, Hu Y, Yang W, Ao W, Qiu G. Alteration of diaspore by thermal treatment. J Cent S Univ
Technol. 2004;11:173–5. https://doi.org/10.1007/s11771-004-0036-z.
Is Hydrogen the Silver Bullet for Decarbonization?
Mansoor Barati1 (*)

1
University of Toronto, Toronto, ON, Canada
mansoor.barati@utoronto.ca
Abstract. The production of metals is directly responsible for producing up to 10% of

anthropogenic carbon dioxide. To decarbonize the industry with an eye to carbon neutral-
ity by 2050, a roadmap is emerging that involves interim measures such as carbon capture
and storage. The use of hydrogen as both fuel and reductant is, however, viewed as the
ultimate solution toward a net-zero extraction of metals. In this talk, this view is analyzed
from technical and economic perspectives for the case of steel. The potential to use hydro-
gen as a reductant and the challenges against it are discussed.
Keywords: Hydrogen metallurgy · Metal extraction · Decarbonization
1 Introduction
Carbon has been the backbone of the metals industry as both fuel and reductant, making
the production of metals the second largest source of industrial greenhouse gases, only
after cement. Steel manufacturing, for example, is estimated to account for as low as
6% [1] and as high as 11% [2] of the global anthropogenic carbon dioxide generation
and about a quarter of the industrial emissions. Production of other metals (Ni, Mg, Al,
and ferroalloys), although generating significantly smaller amounts of emissions
because of their lower annual production, is still responsible for a substantial amount of
GHGs compared to other industries with a similar economic impact. Consequently, the
metallurgical industry is at the fore of reshaping its processes and technologies to meet
the climate goals such as those set in COP26.
A pathway to net-zero steel production is emerging which includes improving
energy efficiency and recycling of metals in the short term, reliance on natural gas and
biomass-based production of iron units together with carbon capture and
sequestration/utilization in the medium term, and using hydrogen or electrolysis of
slags in the long term. Currently, only few metals and in very limited quantities are
commercially produced using hydrogen [3]. In the fast-paced paradigm shift to hydro-
gen metallurgy, numerous fundamental and technological aspects need to be investi-
gated and addressed. This article compares the cases of carbon- and hydrogen-based
production of steel from technical perspectives of thermodynamics, kinetics, and heat
transfer to demonstrate the opportunities and constraints.
https://doi.org/10.1007/978-3-031-17425-4_20
130 M. Barati
2 H2 versus C and CO
Hydrogen-based production of steel from primary sources is dominantly viewed as an

extension of the current DR-EAF method of steelmaking which involves solid-state
reduction of iron ores to metallic iron followed by the melting and refining in an electric
furnace. Alternatively, smelting of iron ores in a slag-bath using hydrogen may be an
option. The former method is favorable because of a 60-year proven record of technolo-
gies that use a mixture of CO + H2 for reduction of iron ores. Table 1 compares the
thermodynamic strengths of reduction of FeO by C, CO, and H2 by presenting the
equilibrium constants and reaction enthalpies. In comparison to CO, hydrogen is a
stronger reducing agent at temperatures exceeding 827 °C, which is on the lower range
of the typical temperatures in direct reduction processes for iron ore. Therefore, from
thermodynamic perspective, hydrogen must be able to outperform CO in a direction
reduction process. On the other hand, the reduction strength of carbon is significantly
greater than H2, particularly at smelting temperatures (>1400 °C). These suggest that a
switch from CO to H2 in direct reduction processes should not negatively affect a direct
reduction process from thermodynamic point of view (i.e., utilization of gas, metalliza-
tion degree, etc.) but hydrogen may not be as effective reductant as C in processes
involving smelting (e.g., blast furnace or slag-based smelting). Another interesting
observation is that both the equilibrium constant of reduction by C and endothermicity
of reduction are much greater than H2, this is due to the strong effect of the positive
entropy of the reduction by carbon. Further, the reduction by CO is exothermic while
that with H2 is endothermic, requiring a greater supply of heat to the reactor. This heat
may be supplied by either increasing the gas flowrate or increasing its temperature, both
of which have certain implications. A higher gas flowrate results in a smaller utilization
of the H2 in the process and larger circulation load for the gas. On the other hand,
increasing the gas temperature is highly limited due to the sticking problem it can create
for the pellets/iron ore. However, these are aspects that are seen as design consider-
ations rather than a fundamental barrier, for example, Midrex has successfully operated
DR plants with H2/CO in a wide range of 0.37–3.8 feed gas [4].
The higher diffusivity of hydrogen compared to CO typically results in faster reduc-
tion kinetics, unless there are localized heat transfer restrictions. This is evident in
Fig. 1, showing that the reduction of iron ore pellets by H2 can be double that of CO. The
implication of faster kinetics and endothermic reaction is that heat supply and transfer
will be the likely limiting step when gas is switched from CO or CO + H2 to H2.
The heat transfer between hydrogen and solids could be substantially higher than
that with CO, despite the lower density of H2. This is because of the much greater
Table 1. Equilibrium constant and standard enthalpy of FeO reduction by C, CO, and H2.

H800 C H 900C H1600C
Reaction K eq800°C K eq900°C 1600°C
K eq kJ kJ kJ
FeO + H2 = Fe + H2O 0.48 0.56 0.72 16.1 15.8 −8.4
FeO + CO = Fe + CO2 0.53 0.45 0.18 −17.9 −17.2 −35.7
FeO + C = Fe + CO 3.6 15.5 3500 152 151 125
Is Hydrogen the Silver Bullet for Decarbonization? 131
100
90
80
70
Reduction (%)
60
50
40
30 Hydrogen
20 Midrex gas
10 Carbon monoxide
0
0 5 10 15 20 25 30
Time (min)
Fig. 1. Kinetics of reduction of hematite pellets at 850 °C by different gases.
thermal conductivity of hydrogen. For example in a set of conditions pertaining to

direct reduction of iron ore pellets, the heat transfer between gas and pellets with H2 is
three times that of CO. Further, on volume basis, the sensible heat of CO and H2 is
comparable. Consequently, heat transfer rate does not appear as a limiting factor. On
the other hand and as discussed earlier, a greater amount of heat must be supplied to the
reactor because of the endothermic nature of the reactions.
A thermodynamic evaluation of the reduction of oxides by hydrogen shows that
hydrogen cannot be a suitable reductant for all the oxides. For example, if a minimum
H2 utilization of 10% is considered (i.e., H2O/H2 of 0.1 in the off-gas), oxides such as
Cr2O3, MnO, V2O3, SiO2, TiO2, MgO, and Al2O3 cannot be reduced using hydrogen. On
the other hands, hydrogen is a thermodynamically favorable reductant for reduction of
oxides of metals such as Zn, Co, and Ni.
3 Techno-economic Considerations
Hydrogen cannot be the universal solution to full decarbonization of the metals indus-
try. While it has a huge potential for replacing fossil fuels in the iron and steel sector
(the metal with largest share of GHG emissions), its application for other high-impact
metals such as Al has thermodynamic limitations unless ultra-high temperatures (e.g.,
plasma) are utilized. Even for some reactive metal oxides reducible by hydrogen, its
low utilization can result in uneconomic or technically challenging conditions, such as
large gas recirculation load and high-energy consumption. The different reaction kinet-
ics and process heat balance will require design improvements to realize massive
deployment of hydrogen in the metals industry. More importantly, however, low-cost
green hydrogen is the biggest hurdle against a rapid transition to hydrogen metallurgy.
For example, while the reductant cost of removing a kilogram of oxygen from iron ores
132 M. Barati
is ~ $0.2–0.6 for carbon, it is currently $4–14 for blue hydrogen and $24–64 for green
hydrogen. The transition thus must be enabled by the other industrial sectors paving the
way for large-scale and low-cost production of green hydrogen.
References
1. Fan Z, Friedmann SJ. Low-carbon production of iron and steel: technology options, economic
assessment, and policy. Joule. 2021;5(4):829–62. https://doi.org/10.1016/j.joule.2021.02.018.
2. Swalec C. “These 553 steel plants are responsible for 9% of global CO2 emissions,” carbonbrief.
org. 2021. https://www.carbonbrief.org/guest-post-these-553-steel-plants-are-responsible-for-9-
of-global-co2-emissions. Accessed 6 Jan 2022.
3. Rukini A, Rhamdhani MA, Brooks GA, Van den Bulck A. Metals production and metal oxides
reduction using hydrogen: a review. J Sustain Metall. 2022:0123456789. https://doi.org/10.1007/
s40831-021-00486-5.
4. Chevrier V. Midrex H2: ultimate low-CO2 ironmaking and its place in the new hydrogen economy.
In: AISTech – iron and steel technology conference proceedings. Vol 2018. Philadelphia: AIST;
2018. p. 725–9.
Electrical Resistivity of Charge Materials in the Si
Process and Its Effect on Energy and CO2 Emissions
Haley Hoover1 (*), Meggi Głowacki2, Gudrun Sævarsdottir1, and Merete Tangstad1
1
Norwegian University of Science and Technology, Trondheim, Norway
haley.hoover@ntnu.no
2
Poznań University of Technology, Poznań, Poland
Abstract. The resistivity of mixtures of charge materials in the Si furnace was investi-
gated from 25 to 1600 °C. Char was measured first, followed by char + quartz, and finally
char + quartz + woodchips in a 1:1.8:0.5 weight ratio. The results show that the resistivity
of char + quartz is roughly four times higher than with char alone and that the resistivity
rises rapidly after 1500 °C. Char + quartz + woodchips had a similar resistivity to char +
quartz after the woodchips had transformed to charcoal. The transformation of the wood-
chips to charcoal led to a large volume change and high variability at lower temperatures.
Both char + quartz and char + quartz + woodchips were nonconductive at lower tempera-
tures. Charcoal made from the woodchips during the resistivity measurements had a
slightly higher resistivity than industrial-made charcoal, but it was conductive over the
whole temperature range. The resistivity of the charge mix has the ability to affect the
quality of the furnace operations through power consumption and Si recovery, and there-
fore CO2 emissions.
Keywords: Resistivity · Silicon production · Power consumption
1 Introduction and Background
Industrial production of silicon requires large amounts of both electrical energy and
carbon materials [1]. As such, both the energy consumption and the CO2 emissions
associated with the process have become a topic of interest as the ferroalloy industry
seeks to reach targeted climate goals [2]. The materials needed to produce silicon metal
are quartz and carbon materials (coal, char, charcoal, and woodchips). The mass bal-
ance of the ideal process is shown in Eq. (1); however, the process has intermediate
steps and produces byproducts such as SiO gas and SiC, in addition to the Si. The rel-
evant chemical reactions are printed in Eqs. (2–5) [3]. Equation (2) shows the metal-
producing reaction in the high-temperature area of the furnace. Equation (3) shows the
production of the intermediate SiC, which is necessary for Eq. (2). Equations (4 and 5)
show the production of SiO gas, and the reverse of these reactions are known as the
“condensation reactions” as SiO will condense at lower temperature areas in the
furnace.
https://doi.org/10.1007/978-3-031-17425-4_21
134 H. Hoover et al.
SiO2 s 2C s Si l 2CO g (1)
SiOg SiC 2Si COg (2)
SiOg 2C SiC COg (3)
SiO2 Si 2SiOg (4)
2SiO2 SiC 3SiOg COg (5)
Silicon production requires large amounts of electrical energy to drive the carboth-
ermic reduction of quartz. The energy is delivered via carbon electrodes, which strike
an electric arc in the lower part of the furnace, producing enough heat to drive the nec-
essary reactions [4]. The electrical properties of charge materials in the silicon process
are important in determining the energy distribution in the furnace. Highly resistive
charge is desirable to ensure sufficient current is directed through the arc in the lower
part of the furnace [5]. Extensive work has been done on the resistivity of various car-
bon materials at high temperatures [6–14]; however, no studies to the authors’ knowl-
edge have incorporated quartz and woodchips. Work has also been done on the resistivity
of partially transformed SiC, an essential intermediate in the process (Eqs. 2 and 3) [8,
9, 11, 15, 16]. A summary of the bulk resistivity of various carbon materials and SiC at
high temperature is shown in Table 1. Although the resistivity of the carbon materials
can vary due to differences in composition and structure, based on previous work, it is
reasonable to expect char (semicoke) to be between 9 and 16 mΩm and the charcoal
Table 1. Bulk resistivity of materials for the Si process from the previous literature.
Material Temperature (°C) Resistivity Particle size (mm)
Various cokes [6] 1500 4–14 mΩm 10–30
Petcoke [6] 1450 8–20 mΩm 10–30
Charcoal [6] 1600 7–16 mΩm 10–30
Anthracites [6] 1400 9 and 41 mΩm 10–30
Coke [12] 1600 20–36 mΩm 3–6
Char [12] 1600 16.6 mΩm 3–6
Coke [10] 1600 70.9 mΩ 5–25
Char and gas coke [10] 1600 164 mΩ 5–25
Coke [7] 600 110 mΩm
Semicoke [7] 600 1250 mΩm
Charcoal [14] 1600 10.8 mΩm 4.5–9.5
Char (treated) [14] 1600 9.1 mΩm 4.5–9.5
Coal (treated) [14] 1600 7.8 mΩm 4.5–9.5
69% SiC [15] 1500 20–42 mΩm 4–8
72% SiC (7% Si) [15] 1500 23–31 mΩm 4–8
60% SiC (6% Si) [8] 1500 16–17 mΩm 4–8
60% SiC (36% Si) [8] 1500 14–18 mΩm 4–8
63% SiC (2% Si) [8] 1500 13 mΩm 4–8
Electrical Resistivity of Charge Materials in the Si Process and Its Effect on Energy… 135
(woodchips after conversion) to be between 7 and 16 mΩm. Quartz is a well-known

insulator with a high resistivity. This work investigates the electrical resistivity of mix-
tures of raw materials in ratios similar to industrial furnaces from 25 to 1600 °C. How
this affects Si yield and therefore CO2 emissions are also discussed.
2 Experimental Methodology
The bulk resistivity is measured using the apparatus shown in Fig. 1. This four-point
setup has been previously featured in works concerning the bulk resistivity of carbon
materials and partially transformed SiC material with good results [8, 9, 13–16]. An
alumina cylinder contains the sample within a larger graphite crucible. Molybdenum
wires are threaded through the alumina cylinder with a distance of 10 cm with the
sample covering both wires. A graphite block sits atop the sample and acts as the top
electrode, with the graphite crucible acting as the bottom electrode. The electrodes are
connected to a power source which provides current (I) to the system in a series of
pulses every 25 °C. The molybdenum wires measure the voltage drop (∆V) across the
sample during the pulses, and the electrical resistivity (ρ) is calculated using Ohm’s law
(Eq. 6) and the resistivity equation incorporating the measuring zone (l) and the area (A)
(7). The sample is compacted manually before each experiment to reduce variations in
the void, and a stainless steel weight is added on top to provide even pressure. The
series of experiments is shown in Table 2. First the carbon material, in this case, char
(semicoke), heat treated (HT) to 1200 °C was run to establish a reliable baseline. Next
quartz was added in a 1:1.8 or 1:1.2 ratio. Char, quartz, and woodchips were then
Weight
Weight
Graphite
Graphite block block
Mo
Sample wires
Graphite crucible
Mo
Alumina cylinder
wires
Thermocouples
Fig. 1. Schematic of resistivity setup (left) and image from lab (right).
Table 2. List of experiments.
Experiment no. Material Ratio (wt%)

2 CharB (HT 1200 °C)
3 CharB (HT 1200 °C)
1 CharA (untreated) [15]
2 CharA (HT 1600 °C) [9]
1 CharA (HT 1200 °C) [14]
2 CharB + quartz 1:1.8
1 CharB + quartz + woodchips 1:1.8:0.5
2 CharB + quartz + woodchips 1:1.8:0.5
3 CharB + quartz + charcoal 1:1.8:0.15
1 Charcoal (from woodchips)
1 Charcoal
1 Charcoal [14]
1 Charcoal [9]
measured in a 1:1.8:0.5 weight ratio; this mixture is shown in Fig. 2. Table 3 summa-
rizes the composition of the carbon materials, while the quartz was 99.8% SiO2.
V I R (6)
V A
r (7)
I l
Furnace efficiency is expressed as a function of either power (P) or resistance. The

unitless furnace efficiency factor (η) is shown in Eq. (8). This term shows the ability of
the furnace to utilize its electrical power with minimal losses. Typical furnace effi-
ciency is 85% [17]. Si recovery (μ) is the amount of Si in the metal product as a percent
of the amount of Si in the quartz. This ratio is shown in Eq. (9). Typical Si recovery is
as high as 95% in Norway [17]. The specific power consumption is a function of both
the furnace efficiency and the Si recovery. This is shown in Eq. (10), where (K1) and
(K2) are constants based on the enthalpy of the total mass balance equations for produc-
tion of Si (Eq. 1) and SiO after unit conversion, assuming the reactants are 25 °C and
the products are 1600 and 1400 °C for solids and gases, respectively. Typical power
consumption is 11–13 MWh/tSi [17]. Specific CO2 emissions in Norway are typically
calculated based on the carbon in the raw materials and the consumption of the elec-
trodes only, because emissions associated with energy production are excluded due to
Norway’s hydropower [18]. In this work, the consumption of the electrodes is neglected
and the CO2 emissions are based on the mass balance of carbon (excluding volatiles and
fines) in the raw materials only.
Fig. 2. Image of the mixtures of char and quartz (left) and woodchips (right).
Table 3. Proximate analysis of carbon materials.

(Wt. %) Char B (HT 1200 °C) Char B (HT 1600 °C) Charcoal (from woodchips)
Fixed carbon 96.2 98.8 91.4
Volatiles 2.6 0.9 3.7
Moisture 1.4 0.1 <0.01
Ash 1.2 0.1 4.9
P PLost
Furnace efficiency factor , (8)
P
Si in metal
Si recovery = (9)
Si in raw material
1 K1
Specific Power Consumption K2 (10)
h m
To establish a baseline and confirm that the char used in this work falls within the same
range as previous resistivity measurements, the resistivity of the char was measured
three times. The results are shown in Fig. 3. The results show that at 1600 °C the char’s
Fig. 3. Bulk resistivity versus temperature for char experiments (blue lines with markers) compared
to literature values using the same apparatus [9, 14, 15] shown as lines without markers.
resistivity falls within the literature range of about 10 mΩm. It is directly compared to
the measurements of other chars using the same apparatus [9, 14, 15]. There is a good
agreement based on the respective heat treatment temperatures. The first experiment’s
results were excluded due to equipment issues during this period in the lab.
Quartz was then added to the char. The results from char mixed with quartz in a 1:1.8
ratio and 1:1.2 ratio are shown in Fig. 4. Regarding the 1:1.8 ratio experiments, one was
excluded due to equipment failures. Experiments 2–4 show reasonable agreement
around 1400 °C. All three experiments show that the resistivity when the quartz is
added is approximately four times the value for the char alone. In the smaller 1:1.2 ratio
experiments, the resistivity generally appears lower after 1400 °C, likely due to the
lower amount of quartz. The mixture in both ratios is nonconductive at lower tempera-
tures. The resistivity begins to rise rapidly after 1500 °C for all but one of the experi-
ments. There is larger variation in these samples possibly due to the fact that quartz is
an insulator. Therefore, a small amount of concentrated quartz particles in the measur-
ing zone would disrupt the current path very effectively. A simplified example of this is
shown in Fig. 5, where a bed of two conductive particles and one insulating particle are
fixed and possible current paths is shown in red.
Furthermore, as quartz begins to approach it’s melting point, it will soften and
expand. This expansion could affect the current paths and thus the bulk resistivity in
two possible ways. The first is that the softening and expansion pushes the conducing
particles closer together and increases their contact area. The second is that as it softens
and expands, it starts to envelop the edges of its neighboring conductive particles and
thus pushes them further away. During one experiment, the results of which are not
shown due to instrument failure, the setup was heated past the melting point of quartz.
The resulting mass extracted from the setup afterwards is shown in Fig. 5. The quartz
has cooled into a mass with pieces of the char attached throughout, but rarely in contact
with one another.
Fig. 4. Bulk resistivity versus temperature for char + quartz in 1:1.8 weight ratio (orange lines) and
1:1.2 weight ratio (blue lines).
Conducting particles
Conducting particles
Insulating particles
Possible current path
Char Melted
quartz
Fig. 5. Schematic showing a possible current path through a bed of conducting and insulating particles
(left) and a retrieved mass of char and quartz after an experiment in which the quartz melted (right).
Woodchips were then added to the char + quartz mixture in a 1:1.8:0.5 weight ratio.
The results of the resistivity of the char + quartz + woodchips from 1000 to 1600 °C are
shown in Fig. 6. The mixture is nonconductive at lower temperatures. The addition of
the woodchips posed additional challenges as the wood will be converted to charcoal as
the apparatus is heated. This will result in a significant volume change and a change in
resistivity as charcoal is a good conductor and woodchips are not. The mixture before
and after the experiment is shown in Fig. 7. The process for charcoal making is sum-
marized in Fig. 8, where the moisture and volatiles will be released first endothermi-
cally and then exothermically [19]. The temperature range for this process is between
100 and 500 °C, so it is assumed that over 1000 °C the woodchips are fully converted.
For experiment 3, the used woodchips from the previous experiments, which are now
Fig. 6. Bulk resistivity versus temperature for char + quartz + woodchips and char + quartz + charcoal
made from the woodchips (gray line).
Fig. 7. Image of the char + quartz + woodchips before (left) and after (right) the measurements where
the woodchips have transformed to charcoal.
considered charcoal, were used in a similar fixed carbon ratio (assuming woodchips are
10% and charcoal 80% [20]) to see the resistivity without the conversion and accompa-
nying volume change. Experiments 2 and 3 show good agreement, and while experi-
ment 1 has a lower resistivity than the others and has a spike between 1300 and 1500 °C,
after that it seems to agree with the other two. This deviation could be due to shifting
material as any remaining volatiles are released. Additionally, the resistivity appears to
Fig. 8. Summary of the charcoal making process, where blue boxes are endothermic and red boxes are
exothermic [19].
Fig. 9. Bulk resistivity versus temperature for the charcoal made from woodchips in this work (blue
dotted line) compared to industrially made charcoal and charcoal from the previous literature [9, 14],
shown as lines without markers.
be similar to that of the 1:1.8 char + quartz mixture at high temperature, around
40–50 mΩm.
The charcoal made from woodchips in this work was recovered and the resistivity
was measured on its own. This was then compared to other industrially made charcoals.
The results from 900 to 1600 °C are shown in Fig. 9. The resistivity of the “homemade”
charcoal is higher than the industrial samples after 1200 °C; however, it is conductive
over the entire temperature range. Industrially made charcoal is nonconductive below
around 900 °C before it rapidly decreases in resistivity, as seen in multiple studies [9,
14, 15].
The electrical properties of the charge mix can have significant effects on the quality
of operations of the furnace. The specific energy consumption (kWh/t Si) at 80–86%
furnace efficiency as a function of Si recovery is shown in Fig. 10. Also, the specific
Fig. 10. Specific power consumption (at various furnace efficiencies) and minimum specific CO2
emissions (red line) versus Si recovery.
CO2 emissions (kgCO2/kgSi) as a function of Si recovery are shown. As recovery is

increased both the power consumption and the CO2 emissions decrease. The theoretical
minimum based on the carbon balance in the furnace excluding the electrodes is 3.14
kgCO2/kgSi. The resistivity of the charge mix has a direct effect on both of these rela-
tionships. Firstly, it is understood that high resistivity in the charge is desirable to direct
the current through the arc [21]. The results in this work have shown that above 1000 °C
the charge mix is conductive, and the resistivity will decrease with increasing tempera-
ture. This has several implications for the operations of the furnace. For example,
overcoking could lead to electrical problems by increasing the power consumption
because the carbon materials are at least four times more conductive than the mixtures
that contain quartz. Similarly, overcoking leads to increased CO2 emissions as Si recov-
ery is lowered as carbon consumption is increased. Additionally, overcoking leads to
SiC accumulation. Work by Hoover [8, 9, 15, 16] on the conductivity of transformed
SiC also shows it to be conductive over 1000 °C and only slightly more conductive than
the charge mix (see Table 1).
4 Conclusion
The resistivity of mixtures of charge materials in the Si furnace was investigated from
25 to 1600 °C. Char was measured first, followed by char + quartz, and finally char +
quartz + woodchips. The results show that the resistivity of char + quartz is higher than
with char alone, and the resistivity rises rapidly after 1500 °C. Char + quartz + wood-
chips had a similar resistivity to the char + quartz after the woodchips had transformed
to charcoal. The transformation of the woodchips led to a large volume change and high
variability at lower temperatures. Both char + quartz and char + quartz + woodchips
were nonconductive at lower temperatures. Additionally, the charge mixtures have
higher resistivity than the results for partially transformed SiC (0–4 times higher) and
for carbon materials (2–4 times higher). Charcoal made from the woodchips during the
resistivity measurements had a slightly higher resistivity than industrial-made charcoal,
but it was conductive over the whole temperature range. The resistivity of the charge
mix has the ability to affect the quality of the furnace operations through power con-
sumption and Si recovery, and therefore CO2 emissions.
Acknowledgments This publication has been partly funded by the SFI Metal Production, (Centre for
Research-based Innovation, 237738). The authors gratefully acknowledge the financial support from
the Research Council of Norway and the partners of the SFI Metal Production. This paper is also pub-
lished as part of the project “Electrical Conditions in Submerged Arc Furnaces – Identification and
Improvement,” in short SAFECI, with financial support from The Research Council of Norway (Pr.Nr.
326802) and the companies Elkem, Eramet Norway, Finnfjord, and Wacker Chemicals Norway. The
authors would also like to thank Gerrit Surup for his experimental guidance and helpful discussions.
References
1. Schei A, Tuset JK, Tveit H. Production of high silicon alloys. Trondheim: Tapir Forlag; 1998.
2. Elkem. Important milestone on the road to carbon-neutral metal production. Elkem
AS. Sustainability Highlights 2017 Magazine. p. 7; 2017.
3. Bø SM. Phases and zones in the silicon process. MS thesis, Department of materials technol-
ogy. Trondheim: Department of Material Science and Engineering, Faculty of Natural Sciences,
Norwegian University of Science and Technology; 2016.
4. Sævarsdottir G, Bakken J. Current distribution in submerged arc furnaces for silicon metal/fer-
rosilicon production. In: Ferrosilicon smelting; 2010. p. 717–28.
5. Magnusson TY, Tangstad M, Sævarsdottir G. The effect of side arcs on current distributions in
a submerged arc furance for silicon production. In: Materials processing fundamentals. TMS
Springer; 2020. p. 177–88.
6. Eidem PA. Electrical resisitivity of coke beds. PhD Thesis, Faculty of Natural Sciences and
Technology, Department of Material Science and Engineering. Trondheim: Norwegian University
of Science and Technology; 2008.
7. Hasannezhad H, Meysami A. Effects of porosity and electrical resistance of metallurgical coke and
semicoke on silicon recovery in a electric arc furnace. In: Electric arc smelting; 2018. p. 336–41.
8. Hoover H. Electrical resistivity of materials in the Si process-an internal literature review.
Trondheim: Norwegian University of Science and Technology; 2022.
9. Hoover H, Pedersen T, Sævarsdottir G, Tangstad M. Bulk resistivity of SiC and carbon materials
in the silicon furnace. In: Silicon for the chemical and solar industry XV. Trondheim: Norwegian
University of Science and Technology; 2020. p. 65–74.
10. Krogerus H, Lintumma T, Jokinen P. Laboratory investigations of the electrical resistivity of cokes
and smelting charge for optimizing operation in large ferrochrome furnaces. In: Southern African
pyrometallurgy; 2006. p. 309–28.
11. Krokstad MA. Electric resistivity in industrial SiC-slag. MS thesis, Institute of Material
Technology, Department of Natural Sciences. Trondheim: Norwegian University of Science and
Technology; 2014.
12. Nurmukhanbetov Z, Privalov O, Prokopyev L. Reducibility and electric resistance of carbonaceous
materials within ferroalloy smelting processes. In: The thirteenth international ferroalloys con-
gress: efficient technologies in ferroalloy industry. Almaty; 2013.
13. Pedersen T. Method for measuring bulk resisitivity in particulate carbon materials. Specialization
project TMT4500. Department of Material Science and Engineering. Trondheim: Norwegian
University of Science and Technology. p. 1–8; 2019.
14. Pedersen T. Electrical resistivity in carbon materials converting to SiC. MS thesis, Institute of
Material Technology, Department of Natural Sciences. Trondheim: Norwegian University of
Science and Technology; 2020.
15. Hoover H, Saevarsdottir G, Tangstad M. Electrical resistivity of partially transformed silicon car-
bide made from coal. In: INFACON XVI. Trondheim; 2021.
16. Hoover H, Sævarsdottir G, Tangstad M. Electrical resistivity of transformed carbon materials in the
silicon furnace. In: TMS furnace tapping. Anaheim; 2022.
17. Tangstad M. Metal production in Norway. Trondheim: Akademika Publishing; 2013.
18. Monsen B, Lindstad T, Tuset J. CO2 emissions from the production of ferrosilicon and silicon
metal in Norway. In: Electric furnace conference proceedings. Norwegian University of Science
and Technology; 1998. p. 371–8.
19. Food and Agriculture Organization of the United Nations (FAO). Industrial charcoal making.
Rome: FAO Forestry Paper 63; 1985.
20. Myrvågnes V. Analyses and characterization of fossil carbonaceous materials for silicon produc-
tion. PhD thesis, Faculty of Natural Sciences and Technology, Department of Material Science and
Engineering. Trondheim: Norwegian University of Science and Technology; 2008.
21. Magnussen T. Basic paramters in the operation and design on submerged arc furnaces, with partic-
ular reference to production of high-silicon alloys. J South Afr Inst Min Metall. 2018;118:631–6.
Fossil-Free Carburization of Hydrogen-Based DRI
T. K. Sandeep Kumar1, Hesham Ahmed3 (*), Bo Björkman2, and Johanna Alatalo4

1
Luossavaara-Kiirunavaara Aktiebolag, Luleå, Sweden
2
Luleå University of Technology, Luleå, Sweden2
Hesham.ahmed@ltu.se
3
Central Metallurgical Research and Development Institute, Helwan, Egypt
4
Formerly at Luossavaara-Kiirunavaara Aktiebolag and presently at AFRY (ÅF Pöyry),
Luleå, Sweden
Abstract. The use of fossil-based reductants and replacing them with green H2 enables
the transition toward fossil-free iron and steel making. However, the produced carbon-
free DRI will make the refining and steelmaking operations to have a starting point far
from today’s operational practices. Therefore, in H2-based iron-making process, a way to
introduce carbon in a controlled manner is essential. Hence, the carburization of H2-
reduced DRI was studied at the single-pellet scale using gases simulating bio-based syn-
gas. The factors affecting the carburization process, such as gas composition, temperature,
and duration as well as the simulated syn-biogas were investigated thoroughly in the
temperature range from 600 to 800 °C. Further, the carbon introduced into the hydrogen-
reduced DRI can either be in the form of cementite, graphite (free-carbon), or a mixture
of both. Therefore, the effect of carbon content and the form in hydrogen-based DRI for
their respective melting characteristics have also been investigated and presented here.
Keywords: Carburization · Hydrogen-based DRI · Bio-syngas · Fossil free
1 Introduction
Ironmaking technology is on the brink of transformation from the use of fossil-based to

fossil-free (i.e., hydrogen) primary reductants [1–3], which will enable mitigating the
CO2 emissions in accordance with the Paris Climate Summit in 2015 [4, 5]. The reduc-
tion of iron ore by only hydrogen emits steam into the atmosphere instead of CO2
(Eq. 1) and results in the production of carbon-free DRI.
Fe 2 O3s 3H 2 g 2 Fe 3H 2 O g (1)
Though, it is advantageous to have C, preferably in the form of cementite (Fe3C) in

DRI (Eq. 2) for smooth and energy-efficient smelting in EAF. Therefore, it is necessary
to design the optimum thermal and gaseous atmosphere for introducing carbon into the
DRI, especially by fossil-free means in an exclusive/separate carburization furnace
after reduction.
https://doi.org/10.1007/978-3-031-17425-4_22
146 T. K. Sandeep Kumar et al.
3Fe s C Fe 3 Cs (2)
Bio-syngas generated from the gasification of biomass presents a viable fossil-free

source of carbon-containing gases for carburization. As bio-syngas might comprises
several gas species, such as CO, CH4, CO2, H2, and N2, it is intended in this study to (i)
systematically investigate the fundamental aspects of each of these gases and their mix-
tures, (ii) investigate the effect of varying bio-syngas compositions on the carburization
of DRI at single-pellet scale between 600 and 800 °C, and (iii) evaluate the carburized
DRI for its melting behavior with respect to carbon content.
2 Experimental Results and Findings
The DR-grade pellets from LKAB were pre-reduced to ~95% in the separate tubular
furnace with only hydrogen at 830 °C. These pre-reduced DRIs are used to investigate
the carburization behavior isothermally at single-pellet scale in thermogravimetric
analysis (TGA) by capturing its in situ weight gain during the process. According to the
Fe-Fe3C equilibrium phase diagram [6], the maximum weight gain that could be
attained for complete conversion to cementite is 6.67%; therefore, the experiments
were terminated either when DRI achieved the same weight gain or when the maximum
stipulated time of 180 min has been. After carburization, the resulting DRIs were exam-
ined for their macroscopic (mosaic) and microscopic features under a light optical
microscope at the Material Laboratory at LKAB, which were further analyzed by an
image analysis tool to quantify the cementite phase fraction.
Since there can be numerous probable gas mixtures and temperatures, the statistical
approach has been adopted to design the experiments (D-Optimal method) to reduce
the number of experiments to a reasonable number, considering the irregular experi-
mental region with CO and CH4 ranging from 0 to 100%, while limiting H2 and CO2 to
20% in the gas mix [7]. The significant variables which influenced the rate of carburiza-
tion to 2% C and the carbon content in the form of cementite are shown in Fig. 1.
It suggests that both CH4 and CO in the gas have a significant effect on increasing
the carburization rate as well the amount of cementite. The combined effect of param-
eters infers that it is preferable to have higher proportions of CH4 at higher carburizing
temperatures to achieve a higher rate as well as the amount of cementite, while CO is
more likely to deposit carbon on the surface. Also, H2 in the gas enhances the rate and
extent of carburization, while CO2 seems to lower the rate as well as carbon deposition,
and thus could be used in smaller proportions. It seems that higher temperature is ben-
eficial for the rate of carburization, especially with CH4 as the carburizing gas. However,
it is the end product – DRI or HBI, which defines the choice of carburizing temperature
and hence, determines the preferred gas composition. The information from these stud-
ies can be used to indicate what type of bio-syngas could be optimal.
The bio-syngas that can be generated from gasifiers comprises of CH4, CO, CO2, H2
and N2, and does not have the potential to carburize the DRI [8, 9]. However, with the use
of CO2 scrubber and enriching the gasifying gas with O2, there is a possibility to lower
(alter) the proportions of CO2 and N2 in order to attain favorable potential for cementite
formation. Thermodynamic simulations via FactSage have been carried out to estimate
Fossil-Free Carburization of Hydrogen-Based DRI 147
Fig. 1. Statistical analysis for the significance of factors. (a) Rate of carburization and (b) Amount of
carbon in cementite.
the carburizing potential of the thermal and gaseous atmosphere by keeping carbon for-
mation dormant, and subsequently the amount of cementite formation. The bio-syngas
compositions and temperatures where thermodynamic simulations suggest to form only
cementite and do not re-oxidize to Wüstite were chosen for further single-pellet carburi-
zation experiments in TGA followed by micro- and macrostructural image analysis [10].
The resulting weight gain curves from TGA for varying N2 and CO2 contents (at
three levels) in the carburizing gas, for example, at 700 °C are shown in Fig. 2. It was
found that lowering N2 and hence, compensating by an increase in the proportion of CO
in gas mix increased the rate and extent of carburization (cementite). It was re-affirmed
that CO2 in gas mix retards the decomposition of cementite, but at CO2 >approx. 7–8%
in gas mix, the rate of cementite formation decreases significantly, and could not able
to introduce 2% C (weight gain) even after carburizing for 180 min [10].
On analyzing the DRI macrostructurally, the cementite can be distributed in varying
forms across the cross-section, as represented in Fig. 3. It can be observed that as N2 is
lowered while maintaining a similar CO2 proportion in the gas mix, the cementite is
formed uniformly throughout at relatively rapid rate (Fig. 3c), while it is distributed
sparingly at higher N2 content (Fig. 3a) and seems to form sharp interface near the
periphery when carburized at 800 °C (Fig. 3b).
Further experiments have been carried out to evaluate the effect of carbon content on
the melting temperature of the hydrogen-reduced DRI [11]. This was done by using
TGA/ DTA with carburized iron at predefined carbon content (up to ~5%), as shown in
Fig. 4.
Fig. 2. Weight gain (%) during isothermal carburization in TGA exposed to various proportions of N2
and CO2 in the simulated bio-syngas.
Fossil-Free Carburization of Hydrogen-Based DRI 149
Fig. 3. Macroscopic phase distribution of cementite (bright) and metallic Fe (dark) in the carburized
DRI.
Fig. 4. DTA curves of DRI with varying carbon content with respect to temperature.
It can be observed that the first peak appears at approx. 735 °C, which corresponds
to the phase transformation from BCC to FCC structure, followed by a second endo-
thermic peak at ~1130 °C inferring the initial melting of the sample, and a third endo-
thermic peak at higher temperatures >1300 °C indicating the final stage of melting
stage, largely in accordance with the Fe-Fe3C phase diagram [6]. The amount of molten
fraction was found to increase by increasing the carbon content.
Acknowledgments The project members would like to thank the Swedish Energy Agency
(Energimyndigheten) for funding through the HYBRIT Research Program 1. Gratitude is also extended
to the Metallographic Laboratory (Metlab) of LKAB for the support for microscopic sample prepara-
tion and image analysis. The research has been carried out within the context of CAMM, Centre of
Advanced Mining and Metallurgy, at Luleå University of Technology, Sweden.
References
1. HYBRIT – towards fossil-free steel. 2020. https://www.hybritdevelopment.com/.

2. Pei M, Petäjäniemi M, Regnell A, Wijk O. Toward a fossil free future with HYBRIT: development
of iron and steelmaking technology in Sweden and Finland. Metals. 2020;10:972.
3. WorldSteel A Climate action: worldsteel member initiatives. 2021. https://www.worldsteel.org/
steel-by-topic/environment-climate-change/climate-action/climate-member-initiatives.html.
4. United Nations framework convention on climate change The Paris Agreement. 2020. https://
Unfccc.int/process-and-meetings/the-paris-agreement/the-paris-agreement.
5. Government of Sweden Paris climate agreement enters into force today. 2020. https://www.gov-
ernment.se/press-releases/2016/11/paris-climate-agreement-enters-into-force-today/.
6. Avner SH. Introduction to physical metallurgy. New York: McGraw-Hill; 1974.
7. Sandeep TK, Ahmed H, Alatalo J, Björkman B. Effect of temperature and gas mixtures on cement-
ite formation during carburization of hydrogen-reduced DRI. Submitted to J Sustain Metall. 2022.
8. Wennebro J, Wiinikka H, Molinder R, Lundgern M. Producing synthesis gas from biomass gasifi-
cation for direct reduction of iron. SP Energy Technology Center; 2015.
9. Grip C, Toffolo A, Salman H, Andersson LI, Ritzen O, Tottie M, Robinnsson R, Winnikka H,
Östman M, Sandberg E. Forestry meets steel: a technoeconomic study of the possible DRI produc-
tion using biomass. 2015.
10. Sandeep TK, Ahmed H, Alatalo J, Björkman B. Carburization behavior of hydrogen-reduced
DRI using synthetic bio-syngas mixtures as fossil-free carbon source. Submitted to J Sustain
Metall. 2022.
11. Ahmed H, Sandeep TK, Alatalo J, Björkman B. Effect of carbon concentration and carbon bonding
type on melting characteristics of hydrogen reduced iron ore pellets. Submitted to ISIJ Int. 2022.
Utilization of Carbon Fibers (CF) as an Alternative
Reduction Agent for Slag Valorization Processes
Ludwig Blenau1 (*), Manuela Wexler2, Jonathan Mahl2, Werner Baumann2,

Dieter Stapf2, and Alexandros Charitos1
1
TU Freiberg, Institute of Nonferrous Metallurgy and Purest Materials, INEMET,
Freiberg, Germany
ludwig.blenau@inemet.tu-freiberg.de; alexandros.charitos@inemet.
tu-freiberg.de
2
Karlsruhe Institute of Technology, Institute for Technical Chemistry, ITC, Karlsruhe, Germany
Manuela.wexler@kit.edu; jonathan.mahl@kit.edu; werner.baumann@
kit.edu; dieter.stapf@kit.edu
Abstract. We describe a method for utilizing and thereby “chemically recycling” carbon
fibers (CF) in a pyrometallurgical slag valorization process. Since the utilization of CF as
a lightweight construction material is rising, especially with windmills being decommis-
sioned, end-of-life carbon fibers with no established recovery route will accumulate.
Combustion in waste incineration plants has proven inefficient. We prove in crucible
experiments on different PAN-based CF types and sizes (powder to 2 cm fibers) that CF
can be utilized energetically and as a reduction agent in fayalitic copper slag reduction. In
this process, CF are consumed while a pig iron and a secondary glassy slag phase are
created. High temperatures (above 1400 °C) and constant contact with reactive molten
slag in a high turbulence environment (created via Ar-blowing lance) are critical for CF
disintegration. Reaction kinetics are monitored by off-gas measurements of CO and CO2.
A comparison with coke is drawn. While up to 90 wt.% of fibers can be consumed in one
batch, leftover CF can be back cycled into the same process. The produced pig iron is
utilizable in the steel industry and the secondary glassy slag for various glass fiber appli-
cations. A scale-up into a high turbulence Top Submerged Lance (TSL) furnace and tests
on end-of-life CF components are planned. The results shown open the way for CF use in
other metallurgical reduction processes and may be a promising solution to the current CF
disposal challenge.
Keywords: Carbon fibers (CF) · Alternative reduction agents · CTR · Slag

valorization · Pyrometallurgy · Fayalitic slag
1 Introduction
Since the utilization of carbon fibers (CF) as a lightweight construction material rises,
especially with windmills being decommissioned, end-of-life carbon fibers with no
established recovery route will accumulate. Currently, around 200 kt of CF is produced
https://doi.org/10.1007/978-3-031-17425-4_23
152 L. Blenau et al.
annually. Combustion in waste incineration plants and even special waste incineration
plants have proven inefficient [1]. The CF leaves the incineration chamber with the fly
ash, and while they remain only minimally altered, “WHO-fibers” (with a potentially
dangerous fiber geometry according to the world health organization) may be gener-
ated. Carbothermic reduction (CTR)-based pyrometallurgical processes have a high
demand for alternative carbon-based reduction agents since additional fossil CO2 emis-
sions can be avoided. A true closed loop could be created when using organic CF. This
benefits the environment and avoids CO2 taxes. CF, by definition, contain a minimum
of 92 wt.% carbon [2] and a high fixed carbon content. The resin covering the fibers can
be considered volatile. The pure fibers are, however, extremely temperature resistant.
Fayalitic slag is the by-product of the copper production and is a FeOx-SiO2-CaO-
Al2O3 slag,typically containing around 1 wt.% of residual copper. The slag contains
around 55 wt.% FeO. Around 50 Mt. of fayalitic slag is produced yearly [3, 4]. While
most of the slag is landfilled, it can be used as construction material for roadbeds or
dams, as abrasive material, or in the cement industry. Previous studies determined its
suitability for iron recovery [5–8]. Looking at the reaction kinetics, the main goal for
successfully reducing the slag is prolonged contact between the carbon source and the
slag [9, 10]. This paper describes a method for utilizing and thereby “chemically recy-
cling” carbon fibers (CF) as a reduction agent in a pyrometallurgical slag valorization
process.
2 Methods
Different Polyacrylonitrile (PAN)-based pure CF types (Sigrafil and Nippon) and sizes
(<150 μm powder, 6 mm fibers, and 6 cm fibers; 7 μm ⌀), as well as graphite powder
and petroleum coke powder are compared in crucible experiments. Additionally, the
quantity of CF is varied for the 6 mm Sigrafil fibers (33 and 66 g) and the residuals from
a previous experiment are reintroduced into a new run (29 g of recovered Sigrafil 6 mm
from one 66 g run). The experiments are prepared and evaluated using FactSage 8.1.
The reduction agent is placed at the bottom of a corundum crucible, and fluxed fayalitic
slag (500 g slag, 17 g CaO) is placed on top; a lid closes the system. The flux is benefi-
cial because it lowers the slag viscosity. The crucible is flushed with Ar (200 l/h) and
heated in an induction furnace to 1450 °C; the total experimental duration is 5 h. The
reaction kinetics are monitored (Fig. 1) by off-gas measurements of CO and CO2 via an
ABB AO2000 infrared detector. This also allows conclusions about the predominant
oxygen partial pressure since it corresponds to a certain CO/CO2 at a given temperature.
To increase the reduction rate of CF, turbulence is introduced to the system by submerg-
ing the Ar-blowing lance (200 l/h) during the experiment. The resulting iron phase is
analyzed via inductively coupled plasma optical emission spectrometry and carbon sul-
fur analysis, the secondary slag via X-ray fluorescence spectroscopy, and the leftover
fibers via optical microscope (not shown here) and fiber characterization.
Utilization of Carbon Fibers (CF) as an Alternative Reduction Agent for Slag… 153
25 1600
1500
1400
20 1300
1200
1100
Temperature (°C)
% CO and CO2
15 1000
900
800
700
10 600
500
400
5 300
200
100
0 0
00:00:00 01:00:00 02:00:00 03:00:00 04:00:00 05:00:00 06:00:00 07:00:00 08:00:00
Time (hh:mm:ss)
Fig. 1. Temperature (red, dotted), CO (green, broken), and CO2 (blue) emissions during a CTR experi-
ment using 33 g Sigrafil 6 mm CF on 500 g slag. The resulting reduction gas mixture changes from
CO2-dominated to CO-dominated with rising temperature. A first reaction peak is reached around
1:20 h; this marks the burn-off of the resin. At around 1:30 h (the slag starts melting above 1050 °C),
the fibers rise through the slag due to their lower density and react. The CO value reaches about 8%
before 2:00 h. The reaction rate subsequently slows down because the “swimming fibers” have less
contact with the unreduced slag. When adding turbulence via the Ar-lance around 3:40 h, the reaction
rate increases drastically, and the CO level peaks at 20 vol.-%. Altogether a CF conversion of
XCF = 82% and an iron oxide to iron conversion of XFe = 72% was achieved.
Table 1. CF conversion rates of selected experimental setups.
Material Conversion of CF
Sigrafil 6 mm (66 g) 50%
Nippon 6 cm (66 g) 67%
Sigrafil 6 mm (33 g) 82%
Sigrafil 6 mm (29 g reused) ~100%
All types of carbon fibers show significant and comparable reaction rates. The Sigrafil
6 mm fibers had the lowest conversion (cf. Table 1) of around 50%; the respective value
for Nippon 6 cm pieces was 67% when using 66 g of CF. The experiment with half the
number of fibers (33 g) showed an increase in conversion to 82% of the Sigrafil 6 mm
fibers due to the effective increase of the slag/reductant ratio. Furthermore, the reusage
of the aforementioned 29 g of these fibers lead to a conversion of nearly all available CF
(some smaller fiber fragments escaped to the off-gas). This is probably due to the
increased surface area of the altered fibers. The CF powder shows a much quicker reac-
tion, but dusting, especially during the turbulent stage of the experiment, leads to sub-
optimal results. The same applies to graphite and petroleum coke powder. Pelletization
might eliminate this aspect. In general, the CF reacts comparably fast to graphite; the
petroleum coke outperforms both in regard to reaction speed.
Fig. 2. Nippon CF 6 cm pieces before (left) and after (right) the CTR experiment (66 g fibers on 500 g
slag). Around 22 g fibers remain, which translates to a conversion rate of XC = 67%. The fiber size
distribution varies. While WHO fibers are present, most of the remaining fibers are still relatively intact
and presumably were caught above the turbulent melt. These fibers can be reused as reduction agents
and are even more effective due to their increased surface area.
An example of a reduction experiment using Sigrafil 6 mm is shown in Fig. 1 (the

different stages of the experiment being explained within the figure caption). A first
CO2 peak is visible before the reduction starts, which holds true also for experiments
using different fiber types. This is due to the resin burning off. Afterward, the longer
fibers show similar behavior in their reaction rates, the Nippon 6 cm pieces having the
longest reaction time. The reaction rate is assessed by monitoring the CO and CO2
curves. The geometry of the longer fibers forces them to stay at least partially sub-
merged in the melt, leading to higher reaction rates. The smaller fiber fragments can
accumulate on top of the melt even with heavy turbulence being applied. This shows
that creating a reaction mixture with constant contact between the slag and the fibers is
key to a successful reduction. The microscopic analysis of the remaining fibers shows a
wide range of decomposition degrees. While WHO fibers are present, recycling unuti-
lized fibers (Fig. 2) into the process in a closed system seems promising. The resulting
iron phase (Fig. 3) is carbon saturated as well as low in sulfur, lead and zinc (since these
elements are fumed out). The only undesired component for the steel industry is a rela-
tively high copper amount of 1.7 wt.% in the best case. A pre-reductive step and removal
of a copper-rich phase to the copper process could aid in resolving this issue [8]. The
secondary “glassy” slag (Fig. 3) is SiO2-rich and contains below 10 wt.% of FeO in the
best case. Since it is also free of toxic components and is compositionally suited, it can
be used for glass fiber production without further fluxing.
4 Conclusion and Outlook
CF can be utilized energetically and as a reduction agent for pyrometallurgical pro-

cesses, as demonstrated in fayalitic copper slag reduction. In this process, CF is con-
sumed while a pig iron and a secondary glassy slag phase are created. High temperatures
Utilization of Carbon Fibers (CF) as an Alternative Reduction Agent for Slag… 155
Fig. 3. The resulting secondary slag phase (left) from the Nippon 6 cm experiment (Fig. 2) is compo-
sitionally suited for glass ceramics and even glass fiber production because the FeO content drops
below 10 wt.%. This leads to a suitable viscosity-temperature profile for this slag. The resulting iron
phase (right) of said experiment is suitable for the steel industry when blended enough to compensate
for the high copper content of around 1.7 wt.%. Around 71 wt.% of all available iron is recovered.
(above 1400 °C) and constant contact with reactive molten slag in a high turbulence
environment create a reactive mixture critical for CF utilization. While up >80 wt.% of
fibers can be consumed in one batch, leftover CF can be recycled back into the same
process. The produced pig iron is utilizable in the steel industry and the secondary
glassy slag for various glass fiber applications. A scale-up into a high turbulence Top
Submerged Lance (TSL) furnace and tests on end-of-life CF and CF polymer compo-
nents are planned. TSL-technology is also advantageous because a continuous closed
system process can be designed, thus disallowing the escape of any WHO fibers gener-
ated. The results shown open the way for CF use in other metallurgical reduction pro-
cesses and may be a promising solution to the current CF disposal and chemical
recycling challenge.
Acknowledgments We would like to thank the Sächsische Aufbaubank – Förderbank (SAB) for
financing this research.
References
1. Stockschläder J, Quicker P, Thiel C, Beckmann M, Baumann W, Wexler M, Stapf D, Koralewska

R. Carbonfaserhaltige Abfälle in der thermischen Abfallbehandlung – Zwischenergebnisse aus
Untersuchungen im industriellen Maßstab. In: Energie aus Abfall. TK Verlag; 2019. p. 333–56.
2. Morgan P. Carbon fibers and their composites. 1st ed. CRC Press; 2005. https://doi.
org/10.1201/9781420028744.
3. Dhir RK, De Brito J, Mangabhai R, Qun Lye C. Sustainable construction materials: copper slag.
2017. http://www.sciencedirect.com/science/book/9780081009864.
4. Gorai B, Jana RK, Premchand. Characteristics and utilisation of copper slag—a review. Resour
Conserv Recycl. 2003;39:299–313. https://doi.org/10.1016/S0921-3449(02)00171-4.
5. Blenau L, Stelter M, Charitos A. Carbothermic reduction of fayalitic slag with graphite – under-
standing reaction kinetics for pig iron production. In Slag valorisation symposium 7 conference
proceedings; 2021.
6. Sarfo P, Wyss G, Ma G, Das A, Young C. Carbothermal reduction of copper smelter slag for recy-
cling into pig iron and glass. Miner Eng. 2017;107:8. https://doi.org/10.1016/j.mineng.2017.02.006.
7. Erdenebold U, Choi H-M, Wang JP. Recovery of pig iron from copper smelting slag by reduction
smelting. Arch Metall Mater. 2018;63:1793–8. https://doi.org/10.24425/amm.2018.125106.
8. Busolic D, Parada F, Parra R, Sanchez M, Palacios J, Hino M. Recovery of iron from copper
flash smelting slags. Miner Process Extr Metall. 2011;120(1):32–6. https://doi.org/10.117
9/037195510X12772935654945.
9. Mitrašinović A. Effect of temperature and graphite immersion method on carbothermic reduction
of fayalite slag. JOM. 2017;69:1682–7. https://doi.org/10.1007/s11837-017-2455-y.
10. Warczok A, Utigard TA. Fayalite slag reduction by solid graphite. Can Metall Q. 1998;37:27–39.
Deep Decarbonization and Pathways to Net Zero
Emissions: A Steel Industry Perspective
Ian Cameron1 (*)
1
Hatch Ltd., Mississauga, ON, Canada
ian.cameron@hatch.com
Abstract. Accounting for 6–7% of global greenhouse gas (GHG) emissions, the steel
industry and its supply chain emit more CO2e (CO2 equivalent) than any other metal
produced. Steel is a “hard-to-abate” industry due to its reliance on carbon to reduce iron
ore to iron. Steel demand will grow due to population growth, implementation of renew-
able energy systems, and a need for resilient infrastructure. The steel industry must reduce
GHG emissions while increasing output by more than 30% by 2050. This paper describes
the expected technology changes and related challenges that face the steel industry on its
journey to net zero GHG emissions.
Keywords: Agglomeration · Carbon capture · Hydrogen · Ironmaking · Iron ore

reduction · Steelmaking
1 Partial Decarbonization of Existing Steelmaking Technologies
Steel production falls into three subgroups:

The blast furnace and basic oxygen furnace route (BF-BOF), where coal reduces hema-
tite to Fe. This accounts for 70% of global steel production.
The melting of steel scrap in an electric arc furnace (EAF), sometimes with pig iron or
direct-reduced iron (DRI) to dilute scrap impurities, accounting for 25% of global
steel output.
The reduction of iron ore using reformed natural gas in a shaft furnace followed by
smelting in an EAF (DRI-EAF). This route accounts for about 5% of global steel
production.
GHG emissions vary from 0.2 t CO2e/t liquid steel (LS) for a scrap melting EAF
with access to green power to over 2.3 t CO2e/t LS for the BF-BOF route (Fig. 1).
For many steel producers, natural gas is not available to convert to DRI-EAF steel-
making with its lower GHG emissions. These blast furnace-based steelmakers will
optimize operations and improve raw material quality to achieve best-in-class perfor-
mance and reduce GHG emission rates to 1.7–2.0 t CO2e/t LS.
An innovative technical approach, blast furnace top gas recycling has the potential
to reduce GHG emissions to about 1.4 t CO2e/t LS [1]. Developed by the European
Union’s steel industry, top gas recycling reuses CO extracted from blast furnace top
https://doi.org/10.1007/978-3-031-17425-4_24
158 I. Cameron
2.4
2.2 Typical BF-BOF Route
(tonne CO2e/tonne Liquid Steel) 2.0 Operang Range

Scope 1+2 Emission Intensity
1.8
Typical EAF Route
1.6 Operang Range
1.4
1.2
1.0
0.8 Grid Emission
Factors:
0.6 0.96 kgCO2e/kWh
0.4 Typical DRI-EAF Route 0.42 kgCO2e/kWh
0.2 Operang Range
0.0
0 20 40 60 80 100
Ore based % Scrap Scrap based
Fig. 1. Comparison of CO2e emissions rates based on the amount of virgin iron units (Ore-based) and
recycled iron units (Scrap-based) in the liquid steel produced. Scope 1 + 2 CO2e emissions expressed
per tonne of liquid steel [1].
gas. After heating, the CO-rich gas steam is reinjected into the existing tuyeres and a
new set of mid-shaft tuyeres.
Hydrogen usage both as a direct tuyere injectant and in a top gas recycling blast
furnace can further reduce blast furnace GHG emission rates [1, 2]. With hydrogen
added to a top gas recycling blast furnace, GHG emissions can be decreased by over
50% and approach 1.0 t CO2e/t LS, matching the emission rate of the natural gas DRI-
EAF process route. As H2 reduction is strongly endothermic, coke combustion must
provide the needed sensible heat in addition to coke’s well-known role to provide per-
meability for gas and molten liquid flow essential to the blast furnace process.
In the USA, scrap-based EAFs with their low carbon footprint represent 70% of
American steel production. EAF steelmaking will continue to expand globally and,
especially in the USA where U. S. Steel, Nucor, ArcelorMittal, and others have
announced new EAF steelmaking projects. When considering Scope 2 power-related
GHG emissions and Scope 3 emissions attributed to EAF pig iron usage to dilute scrap
impurities, US-based EAF shops will have carbon footprints approaching 0.8 t CO2e/t
LS.
Steel producers in Europe and Canada, notably ArcelorMittal, Saltzgitter and
Dillinger have announced plans to convert their BF-BOF process to natural gas-based
DRI-EAF production as a transition step to hydrogen steelmaking. These steel produc-
ers anticipate that by 2040–2050, all European Union blast furnaces will close to meet
social pressures to decarbonize the steel industry. DRI shaft furnace ironmaking allows
a large and immediate reduction in GHG emissions using natural gas in place of coal.
Increased scrap usage in the new EAFs will further reduce GHG emissions, and the
degree of scrap usage will depend on the final steel quality requirements and local scrap
availability.
Deep Decarbonization and Pathways to Net Zero Emissions: A Steel Industry Perspective 159
The prevailing DRI-EAF economics depends on using extremely low gangue iron
ore pellets to minimize slag produced in the standard EAF. Due to the limited avail-
ability of DR-quality iron ore pellets and challenges in producing nitrogen-sensitive
steel grades in standard EAFs, large stationary electric smelting furnaces (ESF) can
address both issues. ESFs produce carbon-containing molten iron that will allow the
steel producer to continue to operate their existing BOF steelmaking plants. ESFs can
produce slag favored by the cement industry where the cement producers gain a GHG
benefit by avoiding limestone calcination. Both Tata Steel Netherlands and
ThyseenKrupp have annouced plans to smelt DRI in new ESF smelting plants.
2 Deep Decarbonization Using Emerging Steel Technologies
Conversion of natural gas DRI shaft furnaces to 100% green hydrogen service com-
bined with either EAF or ESF steelmaking will offer a pathway to emission rates less
than 0.3 t CO2e/t LS. Due to the endothermic nature of hydrogen reduction of iron ore,
efficient electrical heating technologies will heat hydrogen as part of the DRI process.
Large resistive electrical heaters are an emerging technology that can heat process
gases to suitable temperatures (~1000 °C) needed for hydrogen reduction of iron ore to
metallic iron.
High-pressure fluid bed reduction processes are in development to eliminate iron ore
pelletizing and its related GHG emissions [3, 4]. Fluid bed technologies may help steel
producers process iron ores that are not easily pelletized due to chemically bound water
present.
Scrap-based EAF producers that use merchant pig iron must find greener iron
sources to reduce Scope 3 emissions and achieve overall emissions of less than 0.8 t
CO2e/t LS for all three emission scopes. Options include increased use of green DRI
produced using hydrogen, use of sustainable charcoal to produce pig iron in mini blast
furnaces, or pig iron produced using smelting reduction technologies that smelt fine
iron ore with coal and sequester the GHG-rich off-gas. EAF operators prefer pig iron to
dilute scrap impurities due to its inherent chemical energy and fast melting behavior.
Smelting reduction combined with carbon capture and storage (CCS) offers a way to
produce pig iron with a low GHG footprint. The HIsarna™ process smelts fine iron ore
and coal with intensive oxygen usage. The process provides an off-gas CO2 content of
>85% to facilitate CCS [5].
Rio Tinto is exploring the use of raw biomass such as agricultural wastes, and micro-
wave technology to produce solid metallic iron from its iron ores [6]. This avoids bio-
mass charring common to other uses of biochar as a GHG reduction strategy. Vale is
implementing its Tecnored™ process in which sustainable biochar, produced from
sugar cane leaves (bagasse), is briquetted with iron ore as self-reducing briquettes and
as biocarbon fuel briquettes and charged to a specially designed rectangular blast
furnace [7]. Boston Metal and ArcelorMittal are developing electrolysis processes that
reduce hematite to Fe using green electricity with elevated and aqueous temperature
technologies at the heavy pilot plant stage.
160 I. Cameron
3 Achieving Net Zero Green Steel
With 100% green electricity widely available, and a substantial supply of carbon-free
iron metallics, steelmaking challenges remain to reach zero GHG emissions. Carbon
emissions from graphite, pre-baked or Soderberg electrodes, and charge carbon to pro-
mote heat transfer and reduce residual iron oxides present in the EAF and ESF charge
materials is an outstanding challenge to overcome.
End-of-pipe biotechnologies are potential solutions to reduce steelmaking emissions
to zero. The production of bioethanol from CO-rich off-gas from an ESF represents one
method to capture carbon. Algae produced using photosynthesis can sequester CO2
from weak EAF or residual ESF off-gas and produce algae products for sale that seques-
ter the contained carbon such as soil remediation.
3.1 A Path to Net Zero for Steel
Pathways to net zero by 2050 are possible; the key requirements for steel industry
decarbonization are access to massive amounts of green electrical power, low gangue
iron ores, and ironmaking technologies that are GHG-free and allow society to recycle
steel scrap in the most efficient way possible. New pyrometallurgical technologies are
in the pipeline that can adapt to these fast-changing requirements to aggressively reduce
steel industry-related GHG emission rates.
Acknowledgments The enthusiastic support from Hatch colleagues Nicholas (Nick) Aubry, Brittany
(Britt) MacKinnon, Richard Elliott, and Maéva Chrzaszcz was appreciated as I prepared this keynote
presentation.
References
1. Sukhram M, Lefebvre K, Aubry N, Cameron I, Ellis B, Liu X, Honeyands T. Top gas recycling
revisited to reduce blast furnace CO2 emissions. Nashville: AISTech; 2021.
2. Cameron I, Aubry N, Sukhram M. Blast furnace hydrogen usage – challenges and opportunities.
AIST hydrogen webinar, June 10, 2021.
3. Primetals. https://www.primetals.com/press-media/news/hyfor-pilot-plant-under-operation-the-
next-step-for-carbon-free-hydrogen-based-direct-reduction-is-done. Last accessed 2022/03/08.
4. Metso Outotec. https://www.mogroup.com/portfolio/circored-hydrogen-based-reduction/. Last
accessed 2022/03/08.
5. Meijer K, Zeilstra C, Hage H, Broersen P, van Boggelen J. Various roads to CO2 reduction with
HIsarna technology. Dusseldorf: METEC; 2019.
6. Rio Tinto. https://www.riotinto.com/news/stories/decarbonising-steel-making. Last accessed
2022/02/27.
7. Potter S, Filho H, Maia L. LeafIron – an innovative contribution to eliminate CO2 emissions from
steel production. Dusseldorf: METEC; 2019.
Utilization of Pretreated Mn-Ore in a Pilot-Scale
Ferromanganese Furnace: Effect of Ore Pretreatment
on Carbon and Energy Consumption
Tichaona Mukono1 (*) , Jonas E. Gjøvik2, Heiko Gaertner2,

Trine Asklund Larssen2, Maria Wallin1, Eli Ringdalen2, and Merete Tangstad1
1
Norwegian University of Science and Technology, Trondheim, Norway
tichaona.mukono@ntnu.no
2
SINTEF, Trondheim, Norway
Abstract. The production of high-carbon ferromanganese (HCFeMn) alloys is an

energy-intensive process where manganese ores are smelted in a submerged arc furnace
(SAF) using carbon reductants thereby generating CO2 emissions. In the prereduction
zone of the SAF, higher manganese oxides in the ore are reduced to MnO through solid–
gas exothermic reactions and at a temperature around 800 °C, the unwanted endothermic
Boudouard reaction is also active. The Boudoaurd reaction consumes both carbon and
energy. As such, the total coke and energy consumption are highly dependent on if the
prereduction occurs by CO gas or solid C. Improvement of existing SAF ferromanganese
process in resource and energy efficiency as well as reduction of CO2 emission through
ore pretreatment in a separate unit is being explored in the EU H2020-funded PreMa
project. A successful pretreatment limits the extent of Boudouard reaction thereby reduc-
ing the carbon and energy footprint of the process. Pilot-scale experiments have been
conducted at SINTEF/NTNU in a 440 kVA AC electric furnace utilizing different feed
mixtures of untreated manganese ore, manganese ore preheated in air, and manganese ore
prereduced with solid carbon. The accounting material and energy balance calculations
for the three pilot experiments were then calculated in HSC Chemistry software. In addi-
tion to a stable furnace operation, pretreatment of Comilog resulted in an increased
energy consumption in the SAF mainly due to reduced exothermic reactions. Prereduction
with solid carbon in separate unit was found not to be beneficial as there was insignificant
difference in ore oxygen level compared to preheated ore.
Keywords: Ferromanganese production · Ore pretreatment · Pilot · Carbon ·

Energy consumption
1 Introduction
In the ferromanganese industry, several efforts to reduce the energy consumption and
environmental CO2 footprint emanating from the direct process have received extensive
attention. The ferromanganese furnace process consists of a prereduction zone and coke
bed zone. The prereduction zone is characterized by a set of exothermic chemical
https://doi.org/10.1007/978-3-031-17425-4_25
162 T. Mukono et al.
reactions where high manganese oxides (MnO2, Mn2O3, Mn3O4) present in the ore are
reduced in the solid state by CO gas generated from Mn metal formation reaction in the
coke bed located at the bottom of the submerged arc furnace (SAF) [1]. These reactions
generate CO2 which, if it occurs at temperatures above 800 °C, participates in the endo-
thermic Boudouard reaction (C + CO2(g) = CO(g)), consuming energy and extra-
carbon. Hence, the gas–solid reactions in the prereduction zone are to a large extent
decisive for the variations of both coke and energy in ferromanganese production [2].
This will hence also affect the total CO2 emissions from the process.
One of the main focus areas in order to reduce direct process-related CO2 emissions
and energy consumption is the optimization of the prereduction process and raw mate-
rial’s usage. The usage of carbonated materials contributes to the CO2 emissions both
through releasing the bound CO2 through an endothermic decomposition reaction and
through the released CO2 reacting with solid carbon according to the Boudouard reac-
tion. As such removal of carbonated fluxes is one potential way of reducing the CO2
emissions from the furnace. In a recent industrial trial on the elimination of carbonate
fluxes in silicomanganese production, an 8.5% decrease in CO2 footprint was achieved
in comparison to utilization of fluxes [3]. The presence of potassium in charge materials
which recirculates and accumulates in the furnace has been reported to increase the
kinetics of the Boudouard reaction and thus causing excessive carbon and energy con-
sumption of the ferromanganese process [4, 5]. According to Swamy [5], the carbon
consumption in an industrial furnace is between 305 and 380 kg per ton of alloy due to
the Boudouard reaction. Typical energy requirements for the ferromanganese process
range between 2.0 and 3.5 MWh per ton alloy [1, 6–8]. Several studies [8–12] have also
been conducted on the prereduction behavior of various ores with the main goal of
optimizing the prereduction reactions. The prereduction behavior differ and are signifi-
cantly dependent on the ore type used. Prereduction of Comilog and Nchwaning ores in
CO/CO2 atmospheres has been extensively studied by Larsen et al. [9, 10]. Investigated
variables included temperature, particle size, and CO concentration in thermogravimet-
ric experiments. It was reported that an increasing partial pressure of CO in CO/CO2
atmosphere promoted the reaction rate with an order of 0.7 and 1.5 for Comilog and
Nchwaning ores, respectively.
In PreMa project [13], pretreatment of Mn-ores in a separate unit prior to feeding the
SAF is an avenue to reduce overall energy consumption and CO2 emissions from the
ferromanganese process currently under investigation. Several aspects of this process
under investigation include design and optimization of the separate unit, use of sustain-
able energy sources, prereduction behavior of ores, and effect of pretreatment on SAF
furnace performance through pilot investigations [14–17]. In this paper, energy and
carbon consumption from utilization of pretreated and untreated ore are discussed on
the basis of accounting mass and energy balances from pilot work.
Utilization of Pretreated Mn-Ore in a Pilot-Scale Ferromanganese Furnace: Effect… 163
2 Mass and Energy Balance for Pilot Experiments
The detailed description of the 440 kVA AC electric pilot furnace at SINTEF/NTNU
and its operation have been previously presented [17]. The pilot-scale experiments were
conducted with untreated and pretreated Comilog ore. The ore had been pretreated by
Eramet Ideas [15] in a rotary kiln in two separate campaigns, one on preheating and
another on prereduction with solid carbon. Therefore, three separate pilot experiments
were conducted at SINTEF/NTNU with untreated (green ore) Comilog, preheated
Comilog, and prereduced Comilog. Typically, the oxygen level in ore referred to as O/
Mn ratio for Comilog is 1.98 and the pretreatment resulted in change in an O/Mn ratio
to 1.6 and 1.5 for preheated and prereduced ore, respectively as shown in Fig. 1. The
small difference in O/Mn ratio between preheated and prereduced ore indicates that
there is not much change due to presence of solid carbon for the pretreatment used.
Table 1 shows the detailed chemical composition of the raw materials. Raw materi-
als were chemically analyzed by SINTEF Norlab using X-ray fluorescence, while the
oxygen content, given by the MnO2 amount, was obtained by titration for ore and sinter.
The manganese ores were utilized in combination with other raw materials such as
sinter, flux, and coke as charge mixtures in the pilot experiments. The charge mix was
calculated aiming at 30% MnO in slag and the calculated charge mixture per pilot
experiment is shown in Table 2. The pilot furnace was fed approximately a total of 22
batches weighing 30 kg per batch in all the pilot experiments. The tapped slag and alloy
in the experiments were at a slag/alloy ratio equal to 0.6. During the pilot runs, the
process off-gas was continuously measured throughout the experiment to quantify CO,
CO2, H2, O2, N2, and CH4 contents. The accounting material and energy balance calcu-
lations for the three pilot experiments were then calculated in HSC Chemistry software
[18] using the actual slag and alloy chemistry established from the products tapped
during the pilot experiments and the input charge blend shown in Table 2. The composi-
tion of the tapped slag and alloy is as shown in Table 3.
2
1.8
1.6
1.4
O/Mn ratio
1.2
1
0.8
0.6
0.4
0.2
0
Untreated Preheated Prereduced
Comilog ore types
Fig. 1. O/Mn ratio of untreated and pretreated Comilog ore.

Table 1. Composition of the raw materials (wt. %).

Raw materials Mn, tot Fe, tot MnO2 SiO2 Al2O3 MgO CaO K2O Fix. C O/Mn
Untreated 51.2 2.9 79.6 6.4 5.8 0.1 0.4 0.7 – 1.98
Preheated 55.1 3.0 51.8 4.9 6.6 0.2 0.4 1.2 – 1.59
Prereduced 54.4 3.2 43.1 4.5 6.4 0.2 0.2 1.1 4.5 1.50
Sinter 57.5 3.9 19.3 6.4 7.4 0.0 0.1 0.8 – 1.21
Burnt lime – 0.6 – 0.5 0.2 0.6 95.5 – – –
Polish coke – 0.7 – 5.1 3.3 – – – 88.6 –
Table 2. Charge mixtures for experiments conducted with untreated and pretreated Comilog.
Pilot experiment charge blend (%)
Raw materials Untreated Pretreated Prereduced
Comilog ore 29.3 29.0 29.6
Sinter 46.0 45.5 46.5
Coke 16.0 16.2 14.8
Limestone 8.7 9.2 9.1
Total 100 100 100
O/Mn 1.51 1.32 1.32
In mass balance calculations, the MnO in tapped slag was fixed according to Table 3.
The Mn in the alloy was obtained by the difference between the total Mn from the
charge mixture and the one reported in the slag as MnO. The carbon content in the alloy
was saturated and fixed at 7% in the alloy, while the total input of Fe was considered to
report to the alloy phase. The unreducible oxides reported to the slag phase and SiO2 was
calculated by balancing the amount of Si input from the charge mixture and Si reported
in the alloy phase. In calculating the energy balance, raw material’s input temperatures
were assumed to be 25 °C, whereas slag and alloy were considered to be tapped at
1500 °C and the off-gas was assumed to be 200 °C. The relationship between the spe-
cific carbon consumption, the CO2/(CO2 + CO) ratio, and energy consumption to pro-
duce 1 ton of HCFeMn alloy was established based on the calculations from 0 to 100
degree of prereduction. The degree of prereduction has been previously defined by
Tangstad [19] as a measure of the extent of the gas reduction of higher manganese
oxides, which is reflected in the CO2/(CO2 + CO) ratio of the off-gas and is inversely
related to the coke and energy consumption. The degree of prereduction is defined to be
100% when all the Mn3O4 from the reduction of higher manganese oxides is reduced at
low temperatures before the Boudouard reaction is activated. If all CO2 from the reduc-
tion of Mn3O4 is consumed by the Boudouard reaction, the degree of prereduction is 0%.
The accounting energy balances were calculated between 0% and 100% degree of prer-
eduction, and the actual degree of prereduction for the three pilot experiments was
calculated based on the CO/CO2 ratio from the off-gas measurements of the last stable
Table 3. Compositions of tapped alloy and slag from the pilot experiments.
Alloy chemistry Slag chemistry
% Untreated Preheated Prereduced % Untreated Preheated Prereduced
Mn 85.8 85.5 84.3 MnO 33.0 32.1 31.0
Fe 6.9 6.1 6.6 SiO2 14.8 14.2 14.4
Si 0.4 0.9 0.9 Al2O3 19.4 19.8 20.1
C 7.0 7.1 7.3 CaO 28.9 31.4 32.0
Mn/Fe 12.4 13.9 12.8 MgO 3.9 2.6 2.4
Table 4. Measured range of CO/CO2 in off-gas and calculated CO2/(CO2 + CO) ratio.
Comilog ore type Off-gas component (%) Calculated CO2/(CO2 + CO)
CO CO2 (Maximum–minimum)
Untreated 40–55 60–45 0.6–0.45
Preheated 55–65 45–35 0.45–0.35
Prereduced 65–70 35–30 0.35–0.30
period toward the end of the experiments. The CO/CO2 off-gas measured is summa-
rized in Table 4. In addition, Table 5 shows, as an example, the mass balance at 0%
degree of prereduction for untreated Comilog charge mixture per 100 kg input material,
calculated in HSC Chemistry 10® [18].
As seen in Table 4, untreated Comilog charge mixture gave the highest CO2 concen-
trations during the experimental runs, followed by charge mixture with preheated
Comilog and lastly, prereduced Comilog with the lowest observed CO2 concentration.
The opposite trend will be observed for CO concentrations. The relationship between
the specific carbon consumption, the off-gas CO2/(CO2 + CO) ratio, and energy con-
sumption to produce 1 ton of HCFeMn alloy for the three different pilot-scale scenarios
is shown in Figs. 2, 3, and 4 for untreated, preheated, and prereduced Comilog,
respectively.
It can be seen that as the CO content of the off-gas increases, the energy consump-
tion due to the Boudouard reaction increases. The main conclusion is that for a fixed
charge the lower CO2 to (CO + CO2) ratio in the off-gas indicates higher CO2 consumed
in the Boudouard reaction and consequently higher carbon and energy consumption as
has been observed previously [5, 20, 21]. The minimum and maximum carbon con-
sumption for the three pilots shown by Figs. 2, 3, and 4 are within the same range
mainly because chemistry of the slags and alloys tapped presented in Table 3 is quite
similar with insignificant variations. However, based on the composition of the mea-
sured furnace off-gas, differences in carbon consumption are observed. Untreated
Comilog has a wide range of carbon consumed and this decreases for preheated and
prereduced Comilog. The narrowing of the range of carbon consumption as is reflected
in off-gas composition shows that the process was more stable for pretreated Comilog
in comparison to untreated. The energy consumption when utilizing untreated Comilog
ranging from 1970 to 2060 kWh is lower compared to preheated Comilog (2140–2210
kWh) and prereduced Comilog (2170–2210 kWh). If the prereduction reactions com-
mences in the prereduction zone of the SAF, higher manganese oxides will reduce to
lower manganese oxides from MnO2 to Mn2O3 to MnO3O4 to MnO in addition to reduc-
166
Table 5. Calculated material’s balance for pilot experiment with untreated Comilog in charge mix.
T. Mukono et al.
Raw materials Mass (kg) MnO2 Mn2O3 Mn3O4 MnO Fe2O3 FeO SiO2 Al2O3 CaO MgO K2O CO2 Fix C
Comilog 29.33 23.81 0.69 1.38 1.04 1.73 0.35 0.35 –
Sinter 46.0 – – 24.3 12.32 2.40 3.08 3.42 0.14 0.34 –
Lime 8.67 8.67
Coke 16.0 0.68 0.68 14.6
Sum 100 23.81 0.69 24.3 12.32 1.38 2.40 4.80 5.83 8.67 0.49 0.69 – 14.6
Kmol 0.27 0,004 0.11 0.17 0.01 0.03 0.08 0.06 0.16 0.01 0,01 – 1.22
Products Mass (kg) Mn Fe C Si MnO SiO2 Al2O3 CaO MgO K2O CO CO2
HCFeMn 40.78 34.92 2.83 2.85 0.17
Slag 29.99 9.90 4.43 5.83 8.67 0.49 0.69
Gas 29.25 19.53 2.47
Sum 100 34.92 2.83 2.85 0.17 9.90 4.43 5.83 8.67 0.49 0.69 19.53 2.47
Kmol 0.64 0.05 0.24 0.01 0.14 0.07 0.06 0.16 0.01 0.01 0.87 0.11
Metal composition (mass %) Slag composition (mass %) Gas composition (%)
Mn Fe C Si MnO SiO2 CaO MgO Al2O3 K2O CO CO2
85.6 6.9 7.0 0.4 33 14.8 28.9 1.6 19.4 2.3 88.8 11.2
Fig. 2. The relationship between energy consumption, CO2/(CO2 + CO) ratio, and carbon consump-
tion for production of 1 ton of alloy utilizing untreated Comilog.
tion for production of 1 ton of alloy utilizing preheated Comilog.
tion of iron oxides. The reduction of these reactions is exothermic. Pretreatment of

Comilog ore through preheating and prereduction lowers the O/Mn ratio of the ores and
as such, this will eliminate a part of exothermic reactions resulting in increased energy
consumption in the SAF.
The measured off-gas was used to calculate the degree of prereduction for the differ-
ent charge mixtures in the pilot scenarios and these are summarized together with
energy consumption and carbon consumption in Table 6.
To evaluate the benefit of charging hot pretreated Comilog ore, calculations were
done assuming that the ores are pretreated in a separate unit and fed into the furnace at
tion for production of 1 ton of alloy utilizing prereduced Comilog.
Table 6. Summarized degree of prereduction, carbon, and energy consumption for the pilot scenarios.
Degree of prereduction Carbon consumption Energy consumption
Pilot experiment (%) (kg C/t alloy) kWh/t alloy
Untreated Comilog 55–82 271–293 1970–2060
Preheated Comilog 65–82 273–288 2140–2210
Prereduced Comilog 70–80 279–288 2170–2210
600 °C. The calculation was made using the highest degree of prereduction summarized
in Table 6, for each ore type. The off-gas was assumed to exit the furnace at 700 °C,
while the slag and alloy are tapped at 1500 °C. The energy comparison of the scenarios
is as shown in Fig. 5. It can be seen that pretreatment of Comilog ore and charging it hot
into the furnace will significantly decrease the energy consumption by about 300 kWh
in the SAF. The observed trend is in close agreement with the previous reported calcu-
lations of different pretreatment scenarios [22]. However, sustainable energy sources
should be utilized in preheating the ore in a separate unit in order to achieve reduced
overall energy consumption for the ferromanganese process.
4 Concluding Remarks
Pilot-scale experiments have been conducted at SINTEF/NTNU in a 440 kVA AC elec-

tric furnace utilizing different feed mixtures of untreated manganese ore, manganese
ore preheated in air, and manganese ore prereduced with solid carbon. Energy and
carbon consumption of a HCFeMn furnace were performed for untreated and pretreated
Prereduced Comilog (25 ℃)
Prereduced Comilog (600 ℃)
Preheated Comilog (600 ℃)
Preheated Comilog (25 ℃)
Untreated Comilog (25 ℃)
1700 1800 1900 2000 2100 2200 2300

kWh/t alloy
Fig. 5. Energy consumption per ton of alloy for untreated Comilog mix (25 °C), preheated Comilog
(600 °C), and prereduced Comilog (600 °C).
Comilog based on pilot experimental data. The main conclusions can be summarized as
follows:
• The minimum carbon consumption is fixed for a given metal composition and was
calculated at 100% degree of prereduction for the reduction of MnO and SiO2 to Mn
and Si in metal, respectively, and carbon dissolution in the metal. This was found to
be within similar values due to the closeness of the composition of the tapped mea-
sured alloy and slag from the pilot runs.
• The off-gas measurements show that the use of untreated Comilog ore leads to con-
siderably lower CO/CO2 off-gas composition compared to pretreated Comilog ores,
largely related to the high oxygen level of Comilog ore. Hence, the use of pretreated
ore will keep the CO2 emissions low.
• Pretreatment of Comilog will lead to an increased energy consumption in the SAF. In
addition, prereduced Comilog did not have any significant difference from preheated
Comilog with regard to energy and carbon consumption owing to insignificant
change in oxygen level when Comilog is prereduced with solid carbon.
• Within the variations of the experiments, the degree of prereduction is the same for
the three experiments. Hence, the CO2 content will be higher for untreated ore and
the energy consumption of untreated ore will be lower due to more exothermic reac-
tions. However, charging hot pretreated ore will result in lowering of the energy
consumption in SAF by 300 kWh/ton alloy.
Acknowledgments This work is part of the PreMa project, which has received funding from the
European Union’s Horizon 2020 Research and Innovation Programme under Grant Agreement No
820561 and industry partners: Eramet, Ferrogroble, Transalloys, OFZ, and Outotec.
References
1. Olsen SE, Tangstad M, Lindstad T. Production of manganese ferroalloys. Trondheim: Tapir

Academic Press; 2007.
2. Ishak R, Tangstad M. Degree of prereduction without coke consumption in industrial furnaces. In:
11th international ferroalloys congress. New Dehli, India; 2007. p. 268–80.
3. Davidsen J. Reducing the CO2 Footprint from SiMn production by optimization of fluxes. In: 16th
international ferroalloys congress; 2021. p. 27–9. https://doi.org/10.2139/ssrn.3926019.
4. Lindstad T, Syvertsen M, Ishak RJ, Arntzen HB, Grøntvedt PO. The influence of alkalis on the
boudouard reaction. In: 10th international ferroalloys congress; 2004. p. 261–71.
5. Swamy KN, Robertson DGC, Calvert P, Kozak D. Factors affecting carbon consumption in the
production of high carbon ferromanganese. In: 9th international ferroalloys congress. Quebec City,
Canada; 2001. p. 293–301.
6. Eissa M, El-Faramawy H, Ahmed A, Nabil S, Halfa H. Parameters affecting the production
of high carbon ferromanganese in closed submerged arc furnace. J Miner Mater Charact Eng.
2012;11:1–20. https://doi.org/10.4236/jmmce.2012.111001.
7. Steenkamp JD. FactSage-based design calculations for the production of high-carbon ferromanga-
nese on pilot-scale. In: 11th international symposium on high-temperature metallurgical process-
ing. San Diego, California, USA; 2020. p. 757–71. https://doi.org/10.1007/978-3-030-36540-0_67.
8. Ringdalen E, Gjøvik JE, Larssen TA, Tangstad M. Pretreatment of manganese ores in different gas-
atmospheres- a method to reduce energy consumption and CO2 emissions in Mn-alloy production.
In: 16th international ferroalloys congress; 2021. p. 27–9. https://doi.org/10.2139/ssrn.3930059.
9. Larssen TA, Senk D, Tangstad M. Reaction rate analysis of manganese ore prereduction in CO-CO2
atmosphere. Metall Mater Trans B Process Metall Mater Process Sci. 2021;52:2087–100. https://
doi.org/10.1007/s11663-021-02162-1.
10. Larssen TA, Senk D, Tangstad M. Reduction of manganese ores in CO-CO2 atmospheres. Metall
Mater Trans B Process Metall Mater Process Sci. 2021;52:363–81. https://doi.org/10.1007/
s11663-020-02018-0.
11. Larssen TA, Tangstad M. Off-gas characteristics for varying conditions in the prereduction zone of
a ferromanganese furnace –a basis for energy recovery. In: 16th international ferroalloys congress;
2021. p. 27–9. https://doi.org/10.2139/ssrn.3926228.
12. Turkova K, Slizovskiy D, Tangstad M. CO reactivity and porosity of manganese materials. ISIJ Int.
2014;54:1204–8. https://doi.org/10.2355/isijinternational.54.1204.
13. PreMa project. n.d. https://www.spire2030.eu/prema. Accessed 15 Aug 2021.
14. Kazdal T, Haas-Wittmuess R, Richter S, Lang S, Binder C, Reuter M. Process design for the pre-
treatment of manganese ores. In: 16th international ferroalloys congress; 2021. p. 27–9. https://doi.
org/10.2139/ssrn.3926619.
15. Julia N, Hecquet A, Nussbaum G, Blancher S, Amalric A. Pre-heating manganese ore in a pilot-
scale rotary kiln. In: 16th international ferroalloys congress; 2021. p. 27–9. https://doi.org/10.2139/
ssrn.3926242.
16. Hamuyuni J, Saarenmaa J, Mäkelä P, Pekkala O, Binder C, Rannantie S, Lindgren M. Pretreatment
of manganese ore for improved energy efficiency and smelting furnace stability. In: 16th interna-
tional ferroalloys congress; 2021. p. 27–9. https://doi.org/10.2139/ssrn.3926120.
17. Mukono T, Gjøvik JE, Gærtner H, Wallin M, Ringdalen E, Tangstad M. Extent of ore prereduc-
tion in pilot-scale production of high carbon ferromanganese. In: 16th international ferroalloys
congress. Trondheim; 2021. p. 1–14. https://doi.org/10.2139/ssrn.3926275.
18. Roine A. HSC Chemistry Software. In: HSC Chem. Software, v 10.0.4.3. Pori: Metso Outotec;
2021. www.mogroup.com/hsc.
19. Tangstad, M. Ph.D. thesis. Trondheim, Norway: The Norwegian Institute of Technology; 1996.
20. Tangstad M, Olsen SE. The ferromanganese process – material and energy balance. In: 7th interna-
tional ferroalloys congress INFACON VII. p. 621–630; 1995. https://www.pyrometallurgy.co.za/
InfaconVII/621-Tangstad.pdf.
21. Ahmed A, Haifa H, El-Fawakhry MK, El-Faramawy H, Eissa M. Parameters affecting energy
consumption for producing high carbon ferromanganese in a closed submerged arc furnace. J Iron
Steel Res Int. 2014;21:666–72. https://doi.org/10.1016/S1006-706X(14)60103-5.
22. Tangstad M, Ichihara K, Ringdalen E. Pretreatment unit in ferromanganese production. In: 14th
international ferroalloys congress: energy efficiency and environmental friendliness are the future
of the global ferroalloy industry. Kiev, Ukraine; 2015. p. 99–106.
Technology Developments and the Metals Industry
in a Low Carbon Economy
P. J. Mackey1 (*)
1
P.J.Mackey Techjnology Inc., Kirkland, QC, Canada
Abstract. The energy transition now underway to renewables recognizes that global
carbon emissions need to be reduced. A more sustainable economy largely based on
renewable energy will take decades to achieve and will require mining and metallurgical
operations to provide the base for new energy sources. The metals industry will therefore
play a crucial role in this transition to a low-carbon world economy. Historically, Canada
with its relatively small population and important mineral resources and land area has
played a major role in the technology and development of mineral exports of metals such
as nickel and copper. Today, Canada can capitalize on the growing demand throughout the
world for the critical minerals and metals required and our country has the technology and
resources for a range of these metals for the renewable energy age. Technology develop-
ments for metal production processes having a lower CO2 footprint are also required.
Keywords: CO2 emissions · Renewable · Hydrogen · Energy · Iron and steel ·

Aluminum · Copper · Zinc · Nickel · Cement · Canadian technology · Low carbon
economy
1 Introduction
The energy transition now underway to renewables recognizes that global carbon emis-
sions need to be reduced. A more sustainable economy largely based on renewable
energy will take decades to achieve and will require mining and metallurgical opera-
tions to provide the base for new energy sources. The metals industry will therefore
play a crucial role in this transition to a low-carbon world economy. Historically,
Canada with its relatively small population and important mineral resources and land
area has played a major role in the technology and development of mineral exports of
metals such as nickel and copper. Today, Canada can capitalize on the growing demand
throughout the world for the critical minerals and metals required and our country has
the technology and resources for a range of these metals for the renewable energy age.
Technology developments for metal production processes having a lower CO2 footprint
are also required.
This brief paper, based in part on an earlier presentation [1], firstly summarizes
global CO2 emissions for a number of key metals. Then, the likely timeline to decarbon-
ize the world steel industry is illustrated. Canada can capitalize on these developments
https://doi.org/10.1007/978-3-031-17425-4_26
172 P. J. Mackey
and this is briefly discussed. The role of research and development in this transition is
also briefly discussed.
2 Global Metals and CO2 Emissions
Table 1 presents data for the global production and value of five key metals (iron/steel,
aluminum, copper, zinc, and nickel) and also includes data for cement for comparison,
also providing average global CO2 emissions per tonne of metal/material, total CO2
emissions for each and also as a percent of global CO2 emissions.
It is seen that the global iron and steel industry contributed significantly to global
CO2 emissions; together with the four non-ferrous metals in Table 1, it is seen that these
five metals contribute to 13.2% of global CO2 emissions – of this percentage, about
three-quarters is due to iron and steel.
Efforts to significantly lower CO2 emissions in iron and steel production have been
underway in a number of countries for some time and recently accelerated on several
fronts. Thus, the use of biofuels and increased use of natural gas where available are
examples of adapting existing technologies for lower CO2 emissions. There are also
exciting new developments underway such as examining the use of hydrogen in a shaft
furnace for DRI production and electrolysis of iron ore in a molten salt bath. Efforts is
also underway to reduce CO2 emissions in many metal mining operations by changing
from diesel vehicles to electric or the use of hydrogen.
Of interest, efforts are also underway to reduce CO2 emissions in cement kilns –
including electrically heated kilns, the possible use of plasma burners, exterior shell
heating by natural gas also with carbon sequestration of the separate, richer-CO2 laden
process gas, etc.
Table 1. CO2 emissions data for five metals and cement production (2020).
CO2 for
Metal Value metal
(Average global ore Production (~US$ Global CO2 Proportion of total t CO2/t
grade) mill. t mill.) mill. t CO2 CO2, % metal
Steel (57–62% Fe) 1870 4,138,000 3380 9.70 1.8
Aluminum (~40% Al) 64 169,000 1000 2.90 15.7
Copper (0.62% Cu) 21 190,000 82 0.24 4.5
Zinc (range 2–6% Zn) 14 41,000 52 0.15 3.8
Nickel (1.6%Ni-L, 2.4 42,000 77 0.22 32.0
1.5–2% Ni-S)
Total CO2 as % of 13.2
global
Included for reference
Cement 4100 510,000 2870 8.3 0.7
Note: 1. The global CO2 emissions considered here are generally energy-derived emissions; 2. The CO2
emissions for each metal are typical global averages. It is noted that the CO2 emission per tonne of a
given metal varies widely from one plant, smelter, or mine/mill to another and from one country to
another. For example, for copper, data show that for just the mine/mill area only, CO2 emissions vary
from 0.4 to over 22 t CO2/t Cu, with the average being 3.8 t CO2/t Cu; 3. For nickel, L refers to limonite
ores, and S refers to saprolite ores
Technology Developments and the Metals Industry in a Low Carbon Economy 173
Example of possible rate of world-wide take-up of low carbon technology in iron and steel making
% of world iron-steel by low CO2 technology worldwide
120.0
Global stock of EVs and % of vehicles on road (IEA, Paris)
10 mill (1-2%) 125 mill (9%)
100.0
Note: Application rate shown here is based in the rate of applicaton

world wide of BOF technology in the period 1950 to 1980
80.0
60.0
40.0
Testing and piloting - H2 for DRI
and other technologies
2022 to 2030
20.0 Canada-% of new
4% 10% 50% vehicles to be
100%
zero-emissions
0.0
2020 2025 2030 2035 2040 2045 2050 2055 2060 2065
Year
Fig. 1. Example of possible rate of worldwide take-up of low carbon technology in iron and steel
making. Note: “Low carbon” here is assumed to be less than about 200 kg CO2/t steel.
Figure 1 presents the possible rate of worldwide take-up of low-carbon technology

in iron and steel making, expressed as the % of world steel by low-CO2 technology
worldwide by year up to 2065.
Developed by the author based on his experience and data, it is projected here that by
2050 – a year that net zero is to be reached as proclaimed by many companies – about
65% of world steel might potentially be produced by various low-CO2 technologies. It
is thought that the timelines for the four non-ferrous metals listed in Table 1 might be
less than that shown for steel in Fig. 1 – that is, a significant proportion of these metals
globally might possibly be produced by low-CO2 technologies by say 2050 – and also it
may occur in the general sequence of zinc, copper, aluminum, and nickel.
Also shown in Fig. 1 is the projected global stock of EVs for 2020 and 2030 (upper
left area of the graph), and the projected % of new vehicles in Canada as EVs for a
number of years (lower area). However, it is not certain that in nearly 7 years’ time that
50% of new vehicles in Canada will be EVs. These projections likely need review.
3 Opportunities for Canada
Canada once led the world in nickel production, Fig. 2. Thus, from about 1910 to 1950,
Canada produced over 70% of the world’s nickel; as recently as 1970, the country pro-
duced about 50% of the world’s nickel (primarily in Sudbury, Ontario and in Thompson,
Manitoba).
Recently, Canadian nickel production relative to world output has declined to about
7% of world Ni, principally due to flatlining of Canadian nickel and significant increases
174 P. J. Mackey
Chart Title
10,000,000 100
World Ni production, t
90
1,000,000
80
Canadian Ni production, % of world

100,000 70
60
Nickel production, tonnes
10,000
50
1,000
40
Canada Ni production
% of World
100 30
20
10
10
1 0
1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 2020 2040
Year
Fig. 2. World nickel production tonnes (LH axis) and Canadian nickel production as a % of world
production (RH axis) from 1840 to 2018.
elsewhere in the world, notably in Indonesia. There are a number of potential new proj-
ects in Canada, and Canadian nickel output is likely to rise in future years.
While the country no longer has the large historical nickel output, it is worth noting
that this former large nickel production was accompanied by significant and successful
research and development efforts over a long period so that Canada has a huge knowl-
edge base in nickel processing and that of by-products such as cobalt and precious
metals. There is a similar situation regarding copper processing and by-products, also
with a large knowledge base; all these metals today are referred to as “critical metals.”
This store of knowledge in the area of critical metals, plus limited but still important
metal reserves of many metals, suggests that Canada can and should capitalize on the
growing world demand for these critical metals. As with past initiatives, Governments
can help in this quest. While a number of programs are in place, it is believed much
more is needed for true economic growth. There needs to be more support for universi-
ties and research centers, specifically in the mining, metals, and materials areas, and
also support for Junior mining companies in the critical mineral sectors. Such activities
actually parallel the approach for change to boost the country’s wealth and growth as
discussed in a recent Globe and Mail article [2].
References
1. Mackey PJ. The importance of the metals industry in a low-carbon economy: a look at future tech-
nologies. John F Elliott Lecture, MS&T, Rolla, MO, USA, 11 November 2021.
2. Bradshaw J, Willis A. The Big rift: corporate Canada to focus on real growth not “sugar pops”.
Globe and Mail, Saturday 2 April 2022.
Predictive Furnace Refractory Maintenance
Procedures to Extend Campaign Life and Reduce
Waste
Afshin Sadri1 (*), Matthew Cramer1, and Wai Li Winnie Ying1

1
Afshin.Sadri@hatch.com
Abstract. As furnace lining concepts and designs have improved over the last 100 years,
the smelter management and maintenance approach has also changed significantly within
this period. One of the fundamental approaches that have changed is the view of the fur-
nace lining. In traditional thinking, the lining was viewed as a consumable with unpredict-
able service life, but a new and modern idea is to see the furnace lining as a piece of
equipment that should be maintained. Simply burning through the lining and hoping for
the best is no longer acceptable or economically viable. Modern actively managed freeze-
lining approaches have, in theory, infinite life and must be proactively managed. If refrac-
tory linings need active management, it flows logically that the causes and costs must be
understood so that the appropriate measures and strategies can be developed to manage
the equipment. The strategies must be turned into operating and maintenance procedures
that are followed on a regular basis, and the data generated should form the basis of any
decisions. Throughout the campaign life of a furnace, successful refractory maintenance
procedures will ultimately reduce relining costs, furnace downtime, waste disposal vol-
ume, and carbon footprint of the smelter. In this paper, the authors will discuss the reasons
and costs of a furnace run-out and the benefits of data-driven predictive maintenance in
decision-making leading to the extension of a furnace campaign life.
Keywords: Refractory waste · Predictive maintenance · Nondestructive testing

(NDT) · Furnace integrity monitoring (FIMS) · Run-out · Maintenance strategy ·
Risk-based decision-making
1 Introduction
Reduction of waste and recycling is becoming one of the main focuses in the metal
processing industry. Approximately, 3 million tons of waste refractory materials are
generated annually in the United States by the ferrous, nonferrous, and glass industries
[1]. The used refractory linings constitute a major percentage of the industrial waste
generated by metal process plants and virtually all of it ends up in landfill sites [2]. The
refractory waste contains silicates, aluminates, magnesia, zirconia, carbides, oxides,
fused materials, and various combinations of additives and chemicals [3]. In addition to
the waste refractory materials, tons of impregnated frozen metals and slags are also
finding their way to the dumpsites. The most effective way to minimize waste is to
https://doi.org/10.1007/978-3-031-17425-4_27
176 A. Sadri et al.
reduce generating waste [4]. This can be achieved by extending the service life of the
refractory materials and delaying furnace relining and rebuilding.
To extend the campaign life of a furnace, there are three steps that must work
together. The first step is to improve furnace designs. For example, utilizing cooling
blocks to prolong refractory service life for the walls and the roof or by improving
refractory quality and construction. The second step involves technologies to improve
efficiencies by reducing energy consumption and waste. Modern control systems, auto-
mated furnace feed systems, and digital metal level measurement in the furnaces are
examples of such technologies. And the third step is to inspect, monitor, and gather data
on all equipment and use the information to design a predictive maintenance program.
In the following sections, we are discussing the extreme furnace failure reasons and
the cost of small- and large-scale furnace failures. Then, we discuss our approach to
maintenance management and provide details on two unique Hatch technologies for
furnace refractory lining inspection and monitoring and how the two technologies can
help increase the campaign life of a furnace and reduce relining waste production.
2 Furnace Run-Outs
Uncontrolled release of molten metal from a furnace can be divided into runaway taps,
leaks, and run-outs. A furnace run-out is one of the most serious disruptions for a smelt-
ing operation. A run-out is a safety hazard that causes equipment and structural damage
and a loss of production at a minimum and can put the lives of employees at risk.
Furnace run-outs can be minor (leaks) where a small finger of molten metal passes
through gaps between refractory bricks and reaches the outside shell or very severe
when a large amount of molten material is released. A leak is considered a reality of
furnace operation and commonly occurs for various reasons but run-out, on the other
hand, or unique and devastating.
3 Causes of Run-Out
There are several known causes for metal run-outs from furnaces and process vessels:
1. Binding and Refractory. The most common cause of run-outs is the movement/
interaction within the structure (whether that be a fixed frame, shell, or adjustable
bindings) and the refractory system where metal leaks between refractory bricks or
between refractory and cooling blocks. This is particularly common when refractory
movement has occurred due to thermal cycling on a nonbound or poorly bound load.
The placement and repair of lining sections are also critical to ensure that no weak-
nesses and metal paths are formed during construction. Ensuring during repairs
(particularly in the taphole region of the furnace) that sufficient excavation occurs to
find a refractory surface that can be tied back and where possible, the repair is com-
pleted with properly bonded bricks with the minimum use of castable, ram, and
Predictive Furnace Refractory Maintenance Procedures to Extend Campaign Life… 177
mortar materials. Careful construction and subsequent heat-up are critical to form-
ing an impermeable structure.
2. Operation and Process. Another common cause of a run-out is operational and pro-
cess errors in which metal and slag levels exceed height limits or inventory limits.
Additionally, exceeding design parameters through metallurgical instability or poor
taphole practices exposes the furnace lining integrity and accelerates wear.
3. Water Ingress. Contact of water with furnace refractory materials can cause hydra-
tion. The release of water into the furnace can also result in run-outs. This release of
water can be sudden or gradual. The most common reason for water leakage into a
furnace is a faulty pipe fitting or damaged/worn-out cooling blocks. A sudden
release of water into the furnace results in water instantaneously turning into steam
as it encounters molten metal. This, in turn, produces a sudden volumetric change of
up to 1600 times, resulting in an explosion that causes damage to the lining and
furnace and consequently molten metal run-out. In comparison, a gradual release of
water can result in a hydration reaction of the refractory lining, causing a rapid
expansion of the lining. Weak, hydrated refractory and displaced lining may result
in binding failure and metal run-out. The failure of a fitting above the furnace roof
either in electrodes or water-cooled roof panels is the greatest risk for refractory
hydration. Other risks of water contact with molten material are most substantial in
the tap block area due to poor lancing practice or rapid degradation of the tap block
during an uncontrolled tapping event per above.
4. Cooling Failure. The loss of cooling can result in severe and rapid wear of refractory
lining and binding and eventually causes metal leaks. More commonly, inadequate
cooling results in accelerated wear of the refractory and loss of refractory structural
and chemical properties. As a result, the entire lining is weakened and more suscep-
tible to temperature and heat load excursions. In high heat flux operations with very
active cooling components in the wall, the loss of cooling is a critical failure and will
result in almost instant damage to the refractory and structure.
4 Cost of a Run-Out
A furnace run-out impacts the smelting and production at three levels:

1. Health and Safety. The highest cost of a run-out is injury and death of the personnel.
This type of tragedy affects co-workers, families, and communities. Aside from
human loss, a plant now must deal with the loss of experienced and trained plant
operators and personnel. There will also be associated insurance costs paid to the
victims’ families related to cases of injury and death that can cost millions of
dollars.
• “On November 8th, 2001, Corus blast furnace no. 5 at Port Talbot, South Wales
at UK exploded. Three employees were killed, and 12 employees and contractors
sustained severe injuries. Many more suffered minor injuries and shock” [5].
• “In February 2008, six workers were killed, and five were seriously injured as a
result of an Assmang’s Cato Ridge furnace explosion in South Africa” [6].
178 A. Sadri et al.
• Run-outs often result in large-scale fires and the destruction of additional prop-
erty. These fires contribute to toxic gas emissions that exceed the environmental
limits and create unhygienic and unsafe air masses for both the plant and the
community. In cases where fugitive emissions from the molten material itself
contain harmful components, plant evacuations, and high-risk situations develop.
2. Equipment and Structural Damage. A run-out can cause excessive damage to the
furnace and surrounding components and results in an aftermath clean-up/repair
and, at worst, a furnace rebuild. In more severe cases, the structural integrity of a
furnace could be compromised. For instance, the repair and partial rebuild due to
furnace run-out at Lonmin No. 1 platinum furnace in June 2009 cost about US $4
million [7]. While the extensive run-outs at Onca Puma ferro-nickel furnaces 1 and
2 in May and June 2012 caused long periods of shutdown and the rebuild of one
furnace cost about US $188 million [8]. The financial incentive to rebuild the other
furnace has yet to be proven, and significant corporate write-downs resulted from
the failures.
3. Production Loss. The loss of production due to furnace shutdown is the greatest
financial loss from a process vessel run-out. The extended furnace downtime from a
run-out directly affects plant production resulting in extensive financial losses. The
time frame for a rebuild or repair depends on numerous factors, including access to
rebuild materials such as refractory bricks, the extent of the damage to electrical
cabling and components, availability of specialist staff to affect the reline overall
planning, and access.
The loss of production from a furnace or process vessel is given by the following
Eq. (1):
% Metal Tapped Metal or Matte Produced

Metal loss = × × Days Emergency Shutdow
Tapped Matte or Metal day (1)
the revenue lost is given by Eq. (2):
Metal Market Value

Revenue loss Metal loss (2)
Volume
Considering a market value of platinum to be $900/oz at the time of run-out, and

that the run-out of a platinum furnace has caused the production to decline by
150,000–200,000 oz [9], the revenue loss due to the run-out is estimated to be
between $135 and $180 million.
In most cases, part of the production loss can be recovered after the repairs and
furnace ramp-ups by increasing production for a period.
4. Creating Toxic Waste. Most of the refractory, concrete, and metal parts after repairs
and rebuilding of the furnace and affected area end up in the landfill site. In the case
of refractories, the chemical components are reasonably stable for a long time if they
are properly stored. However, in landfills, eventually, they can break down and cause
environmental issues. For example, Chromium, one of the components used in pyro-
metallurgical furnaces, is extremely toxic, such as graphite, tar and pitch resin, chro-
mite ore, and various silicates among many others. For example, refractory from a
copper furnace contains various sulfides, arsenic, and selenium among many others
[2]. Hence, the waste from a run-out must be deposited in special “hazardous landfill
sites.” The hazardous landfill sites must be monitored for a long time resulting in
additional long-term costs.
5 Maintenance Strategy
Smelting/melting furnaces and vessels have life cycles like any other equipment. After
commissioning the vessel, the main task is to maintain production, control the costs,
and ensure safety. In a metallurgical plant, smelting furnaces are the indisputable key
equipment that controls production. An effective maintenance program is essential to
keep a furnace running continuously. An ongoing “inspection, monitoring and audit”
program that involves internal and third-party experts would keep the maintenance pro-
gram “effective” and “proactive.” The aim of the maintenance program is to identify
potential risks and carry out just-in-time mitigation or preventive measures to avoid any
equipment damage or furnace standing time. It is believed that through a systematic
inspection, monitoring, and audit program, the optimum times for furnace maintenance
can be decided and consequently prolong the campaign life of the furnace (see Fig. 1)
without jeopardizing the production rate and product quality. The approach to a lining
is not different from any other high capital cost piece of process equipment.
A successful maintenance program depends on the implementation of the following
steps:
• Identification of the key risk areas
• Identifying the inspection and monitoring tools and techniques
• Reliable and relevant data
• Continuous or constant data gathering
• Analysis, comparison, and correlation
• Predictive and proactive actions
Based on the recommendations from the vessel designers and operators, the key
areas of the vessel must be identified for performance and integrity inspection and
monitoring. The acceptable variance from the original performance of an asset must be
recognized based on the maturity of the equipment. For example, the most crucial and
vulnerable areas for a smelting furnace are the tapholes and tapping areas. Tapholes are
changed and relined based on the loss of refractory. The refractory wear is due to the
number of tappings and flow of the molten metal, refractory quality, temperature, depth
of drill penetration on opening, closure material usage rates, and lancing, and it is
recognized by the flow rate of the metal and change in thermocouple temperatures
inserted around the tapholes.
After identifying the key areas, the methodologies, techniques, and instrumentation
that are required to inspect and monitor that area should be identified. The most favor-
able inspection and condition monitoring equipment are the systems that are minimally
intrusive and nondestructive in nature. In some cases, there are several techniques and
180 A. Sadri et al.
Fig. 1. Furnace life cycle.
instrumentation that must be used to be able to characterize the changes in a particular

area of the vessel. For example, in the case of the tapholes, thermocouples, and the fiber
optic temperature probe, combined with modeling and acoustics systems, along with a
database collecting key tap data such as depth of channel drilled, the volume of closure
material used, and tap times could be used to identify the wear and deterioration of the
surrounding refractories [10].
The next step is to make sure that the data is reliable. In most cases, calibrations,
pre-inspection modeling, and calculations are the key to good and reliable data. When
considering data reliability, one must be aware of the sources of error, and the percent-
age of error must be established. For example, refractories are not created equally with
the same quality. There will always be a variation in refractory quality in terms of size,
density, elasticity, stress wave speed, and strength. For calibrations, an average value
must be established.
Once the accuracy of the collected data is established, the frequency of the data col-
lection must be identified. Based on the intensity of the operation, some equipment
requires “continuous” monitoring. For example, tapholes must be monitored continu-
ously. In most cases, the area is frequently inspected at regular intervals. This is known
as “constant” monitoring. On the other hand, furnace sidewalls and endwalls are
inspected at regular intervals to determine refractory wear and thickness.
Data analysis and interpretation are other important parts of the process. A sound
interpretation requires a deep understanding of the vessel. An interpretation of the
inspection and monitoring data performed by a multidisciplinary team who has knowl-
edge in the design, process, and operation of the vessel produces the best results. This
interpretation and analysis could be part of an ongoing audit program. The same multi-
disciplinary team is capable of interpreting measurements, trends, statistics, and occa-
sional anomalies to predict and prevent equipment failure.
6 Proactive and Predictive Maintenance Strategy
Rarely does any equipment or vessel operate and reach its designed life without repairs,
upgrades, and overhauls. Maintenance activities keep an asset in working condition.
There are three main maintenance strategies: reactive, planned, and predictive.
Reactive maintenance is based on the “run the assets to fail” philosophy. In this
strategy, no steps are taken to assure a vessel reaches its designed life. Reactive main-
tenance is the predominant maintenance strategy and is practiced 55% of the time [11].
The actual numbers vary between different industries, and in some industries, reactive
maintenance is the only maintenance strategy. For example, a major automotive manu-
facturing company in the United States has a maintenance staff of 15,000–18,000. The
results show that 85–90% of their utilization is around crisis handling or reactive main-
tenance [12]. The majority of refractory systems are managed in this manner. The lining
is installed, and lining management is a function that only occurs after the first failure.
The first sign of failure or actual failure then initiates the capital cycle of replacement
at the next available period determined by resource availability and the business cycle.
Planned or preventive maintenance is a strategy where the areas of a vessel or equip-
ment are replaced and repaired based on the performance indicators and a fixed time
schedule. This type of thinking started in the 1960s and is more effective than reactive
maintenance. Nevertheless, data show that less than 31% of the plants [11] are using
preventive maintenance. In terms of refractory management, this is the typical tap block
replacement cycle of 3 months or every 36,000 MW/h. Ladle management is often a
good example as a ladle is expected to last 300 taps and is replaced in whole after that
period but in many cases comes up short or could last even longer. It provides for better
cost control in that the majority of the expenditure is planned and spent to schedule;
however, the overall costs could be much lower.
The most modern and effective maintenance strategy that has been the target strat-
egy for most plants is predictive maintenance, where inspection and monitoring data
(not fix schedule) dictate the repairs and replacement of equipment or parts of a vessel.
Predictive maintenance is also known as condition monitoring. Successful predictive
maintenance requires good data collection in addition to good engineering judgment.
Less than 12% of plants [11] use preventive maintenance strategies. For furnace lining
management, this means understanding the state of your lining and having good, reli-
able, accurate data to base your understanding on. In practical terms, the lining is now
actively managed and proactively repaired prior to failure, thereby reducing overall
cost, and increasing the productivity of the asset. Modern furnaces and lining designs
require predictive maintenance for optimal asset life and utilization.
As maintenance strategies evolve and become more complicated, from reactive to
predictive, they require more complex tools, equipment, and technologies. Figure 2
shows an evolution of maintenance strategies over time.
182 A. Sadri et al.
Fig. 2. Maintenance strategy over time.
7 Role of Data in Condition-Based Maintenance: Data Is King
Condition-based maintenance is the most evolved way of monitoring furnace compo-

nents. Condition-based maintenance is the practice of monitoring the condition of com-
ponents to determine when a component should be replaced prior to failure. Predictive
maintenance is an extension of condition-based maintenance: any changes in compo-
nents’ conditions over time can be discovered and make possible a prediction about
when the unit will break down. Condition-based monitoring should be employed when
the cost of monitoring the component is lower than the cost of component failure. This
strategy leads to significant reductions in the cost of labor and the amount of downtime
because components can be replaced and repaired. For conductive linings or actively
cooled walls, a simple change in operating parameters is sometimes enough to re-
establish the refractory protection if the operator is aware of deterioration.
Data is the key component of the condition-based monitoring strategy and the most
important item for success. Data has a diverse definition, but in our application, data is
explained as information output by one or several sensing devices that includes both
useful and irrelevant or redundant information and must be processed to be meaningful.
If the relevant data is used correctly in conjunction with other factors, it can be a leading
element for decision-makings to extend the campaign life of a furnace. Nevertheless,
data is not the only tool for decision-making and even having the best data does not
guarantee correct and optimal decisions making.
To collect data, technology and instruments must be used. To apply the condition-
based monitoring strategy on a furnace or on furnace components, one must install
instruments permanently or intermittently, or alternatively, the area must be inspected
periodically. The data gathered from instrumentation will be analyzed and processed.
Some data management and decision support systems can collect and integrate all
operational and condition monitoring data to provide a comprehensive assessment of
furnace health and status. Automated algorithms, analytical tools, notifications, and
exception tracking ensure data are utilized to optimize operational performance and
maintenance policies.
8 Nondestructive Testing (NDT) and Monitoring
The utilization of NDT equipment and technologies to collect reliable and continuous
data is part of the predictive and proactive maintenance strategy. The NDT methodol-
ogy is decided based on the form of deterioration expected or the monitoring program
required. The inspection of a new or newly relined process vessel is a baseline for safety
and the future performance of the vessel through its expected life cycle. Using the NDT
methodologies, a rate evaluation process will take place to determine the inspection
strategy or remaining service life of a vessel. NDT technologies are feeding data to a
preventive and proactive maintenance strategy to improve process vessel utilization and
profitability. As part of a proactive maintenance strategy, gathered data must be ana-
lyzed and correlated to provide trends and pinpoint problem areas. A multidisciplinary
expert with good engineering judgment must evaluate the results and make decisions.
For example, if the NDT results demonstrate the possibility of a metal run-out in an
area of a furnace, good engineering judgment must be followed by a decisive action:
shut down the operation and start the overhaul and repairs. Any prevention of metal
run-outs or untimely shutdowns will save money for smelting plants. A competent
maintenance strategy should be considered as a money-making center, not a cost center,
by providing more continuous and less interrupted operation. Figure 3 shows how using
the Acousto Ultrasonic-Echo (AU-E) NDT technique can help a predictive maintenance
strategy increase profitability of the plant.
By using online or intermittent AU-E NDT solutions, the operators and managers are
able to obtain a unique and accurate picture of the status of the vessel. Based on the
status and the design of the vessel, operators can make informed decisions regarding the
appropriate action needed. The actions will vary greatly from a small change in process
metallurgy or set point to the immediate shutdown of the facility. More likely is the
ability to proactively plan the repairs to coincide with market downturns, yearly main-
tenance on mechanical components, high input pricing environments, or other advanta-
geous periods. Additionally, if the operator has a longer planning period and information
about the nature of the repair period, the associated costs and overall viability are
greatly increased. Gone are the days of having multiple contractors on-site and opening
the shell to establish what needs to be done. This approach always results in poor repairs
or longer downtimes. The operator now has time to plan, resource and execute the work
with a high level of confidence. Subsequent AU-E NDT work provides near real-time
feedback on the effectiveness of the repair and the value of the original decision. AU-E
NDT solutions are, however, only one part of a complex puzzle – even if they represent
the most commonly overlooked piece.
184 A. Sadri et al.
Fig. 3. Increasing profitability by implementing AU-E NDT inspection into predictive maintenance
strategy.
9 Decision-Making Tools
Decision-making is complex, crucial, and in most cases, does not follow any particular
rule. A combination of the following tools is used to make decisions:
• Intuition – The gut feeling is hugely used in decision-making. It is highly unscien-
tific but usually relies on the experience and reliability of the individuals that are
making the decision. Relying on intuition is clearly risky and exposes the organiza-
tion and business to additional risks associated with staff retention.
• Data – A favorite tool among scientists and engineers. There are no limits in data
gathering, the higher the diversity of the sources, the better. The quality of data and
how it is interpreted is a crucial part of decision-making.
• Information – This is the details exchanged among the individuals that are involved
with various aspects of the furnace, which help give a different tool for decision-
making. Collecting and presenting this information in a comprehensive manner is
critical for good planning and action.
• Knowledge – Knowledge reflects the education and analytical tools that individuals
use to make decisions.
• Risk – Managing risk is the key goal of decision-making. Risk is not necessarily
negative or positive. Risk is the acknowledgment of the probability of a system fail-
ure. Knowing the risk, decisions could be made to reduce the probability of risk and
increase the possibility of success.
• Money – Money is another key factor that has a crucial role in decision-making.
Budget restrictions are the key decision motivators, and the need to plan cash flow
accurately in terms of size and timing is critical in low-margin environments.
• Time – Time restrictions and scheduling are key decision-making tools. Sometimes
decisions are made only based on the timing, for example, replacements of equip-
ment on a furnace due to schedule shutdowns.
Unfortunately, having all the decision-making tools does not resolve in making the
best decisions. For example, history is full of bad decisions that were made based on
best data or best intuition or longest and deepest experience and knowledge. However,
if the above-mentioned tools are used in combination, it may have the management to
make the best decision possible.
10 Where and What to Monitor?
In terms of process vessels and particularly furnaces, we divide them into three areas:
roof, walls, taphole regions, hearth, and furnace structure.
Roof Refractory thickness and quality, in addition to build-up or skull formation, are
determined by periodical NDT inspection of the refractory lining. For intense base
metal furnaces, copper cooling blocks are also utilized, in addition to the refractory lin-
ing to prolong the service life of the roof. Of primary concern here is ensuring sufficient
refractory protection is in place to allow for effective cooling without water leaks.
Sidewalls and Endwalls Refractory thickness and quality of the sidewalls and end-
walls are monitored by periodical inspection of the lining. One common method is the
use of thermocouples and thermal modeling. Experience has shown that thermocouples
are very location-specific and are unreliable if isolated following brick movements. A
newer methodology is the utilization of stress wave NDT and particularly the AU-E
system for determining the remaining refractory thickness. For the furnace walls, the
refractory shows to be in three conditions: good brick, impregnated brick (by molten
metal), and build-up. Gaps between bricklayers caused by movement can also be
detected.
Modern high-power base metal furnaces utilize copper coolers for high-intensity
smelting operations. Water provides efficient cooling that allows furnaces to operate at
higher power, temperature, and for a longer time. Blast furnaces use a combination of
copper and cast-iron cooling blocks to keep the walls cool. Usually, the refractory lin-
ing of castable refractory is placed in front of the cooling blocks, as a buffer to protect
the coolers from direct contact with molten materials. Once the protective refractory
lining is worn out, the cooling creates a freeze lining or build-up in front of the cooling
blocks. Water circuit instrumentation gives a very good indication of the average condi-
tion of a cooled area of the furnace and whether it is in trouble. If water temperatures
go up by a factor of 2, it is very different if this wear is spread over the entire circuit path
186 A. Sadri et al.
versus only at one section of one cooler. Thermocouples if installed in exactly the right
location would give much more precise information about where the problems might
be, but it is not practical to install them at a density necessary to detect localized issues.
Fiber optics have been used on the hot face of the cooling blocks to monitor refractory
and build-up wear, in front of the cooling blocks. Also, AU-E can provide additional
information on the localized scale that indicates the severity of the problem.
Tapholes Inspection and monitoring of tapholes are essential and crucial. There are
several techniques for monitoring the inner tapping block innerwear, such as Taphole
Acoustic Monitoring (TAM) system, which uses acoustic noise caused by the flow of
molten metal to recognize refractory wear inside the tapping block. Also, AU-E is used
frequently to determine the remaining thickness of the refractory for the tapholes. Fiber
optics have been used on the hot face of the tapping blocks to determine temperatures
and subsequently remaining refractory. The combination of these systems allows the
operator to receive real-time information about the health of the taphole area.
Hearth The thickness and quality of the refractory lining are inspected and monitored
by the NDT AU-E system or alternatively by thermal modeling. The main area of con-
cern is the skew bricks and hearth bounding. Floating, impregnation, or infiltration of
the hearth are common problems and can be detected by AU-E NDT techniques.
Furnace Structure The furnace’s structural integrity is monitored by observing the
physical changes to the overall furnace structure. For example, the height, length, and
width of the furnace are periodically measured and monitored. An alternative technique
is an application of the Furnace Integrity Monitoring System (FIMS) which utilizes an
acoustic emission (AE) to monitor noises generated due to physical changes in the fur-
nace so it can triangulate and identify the noise source. The source of noise could
potentially be caused due to a lining failure or gradual expansion or contraction of the
furnace structure.
11 Furnace Monitoring and NDT Inspection
The following sections illustrate the use of continuous monitoring and periodic inspec-
tion to determine the condition of the refractory lining.
Case 1 – The Use of FIMS for Real-Time Continuous Monitoring of Furnace
Lining
Crack initiation/propagation, refractory movement, and wear in the furnace lining gen-
erate elastic waves, which is a rapid release of energy. By installing acoustic emission
sensors on the shell of the furnace, the acoustic wavefronts propagated from the cracks,
joint movement, and wear can be detected. These signals can then be used for source
location similar to earthquake detection and location. FIMS was developed in 2008 and
has been patented for use in various vessels for continuous real-time monitoring of the
refractory lining.
At a small circular smelting furnace, we installed sensors on the most vulnerable
area, that is at the skew brick level and at the taphole regions. The AE sensor layout was
carefully designed (Fig. 4). The FIMS detected AE activities were located on the fur-
nace model in a pseudo-real time based on Eq. (3) for an array of i number of sensors.
x xi y yi z zi cti
2 2 2 2
(3)
The location of the sensors is (x1, y1, z1), (x2, y2, z2), …, (xi, yi, zi). For an array of i
sensors, i unique nonlinear equations can be formed. t0 is the travel time required for the
P-wave to reach the sensor which is closest to the source and Δti is the time difference
between the P-wave arriving between the closest sensor and the i-th sensor, such that:
ti = t0 + Δti. Then the source location can be determined by solving for the four unknowns
x, y, z, and t0 using four or more measured Δti values.
Figure 5 shows the AE activities with time, in which the period of high AE activities
can be isolated for detailed investigation. The AE activities during the high activity
periods can also be visualized on the furnace model. If the AE activities were clustered
in small areas, those areas were identified as damaged zones. For large vessels, local-
izing damage zones would be useful for limiting the maintenance time and cost.
The regular furnace lining campaign life of this furnace ranged from 3 weeks to
~3 months. This provides an opportunity for the physical measurement of the remain-
ing lining thickness at various locations of the furnace at the end of the first campaign.
A correlation between the cumulative AE energy/activity with the amount of refractory
wear was made at various locations (Fig. 5). In general, the higher the volume of refrac-
tory loss the more cumulative AE energy and events were recorded (Fig. 5); however,
the rate of wear would be different for different vessels and for different sections of the
Fig. 4. Cumulative AE events with time (top) and AE location at two different time slots showing
significant AE increase (bottom) (red dots indicate AE events; green dots are AE sensor locations).
188 A. Sadri et al.
Fig. 5. Correlation relationship between refractory wear and AE energy level at the bottom row of a
furnace.
same vessel. Thus, for more accurate results, the vessels would be divided into zones
for analysis.
Three empirical correlations were developed for the various zones of the subject
furnace. The furnace was divided into 39 zones (i.e., 3 levels by 13 horizontal sections),
and their corresponding empirical correlation was used for the calculation of the
remaining refractory. A developed view of the furnace showing 39 zones can be used
for plotting heat maps of the remaining refractory thickness (Fig. 6). For larger vessels,
more zones can be identified, and also the resolution of the grids can be changed based
on when required. The software was later refined, and the second-generation FIMS
software can illustrate the AE activity in 3D and color code (Fig. 7).
It should be noted that if the furnace operation and raw materials remained similar,
the same empirical equations can be used for the analysis. However, when the operation
process and raw material change significantly, then it is recommended that a new cali-
bration study should be carried out to refine the empirical relationship.
In the application of AE monitoring on typical furnaces with longer campaign lives,
typically we can use the AU-E NDT technique to detect the remaining refractory for
calibration purposes. Once this calibration is done, the cumulative AE events will be
plotted and illustrated as the remaining refractory thickness at the corresponding zones.
Case 2 – The Use of AU-E Inspection for Periodic Monitoring of Furnace Linings
AU-E is a patent NDT technique that can be used for periodic monitoring of vessel lin-
ings. For a stable furnace operation, AU-E inspection is recommended for annual
inspection to determine the trend of wear. AU-E uses transient stress wave propagation
for the analysis. The stress wave is generated by a signal generator, a mechanical impac-
tor, a vibrator, or a broadband frequency stress wave pulse. The signal is generated on
the shell of the vessel. The compressional waves travel through the media (e.g., shell
and refractory) and are reflected at media boundaries. The reflected signals are received
Fig. 6. Remaining refractory thickness at the 39 zones of a developed furnace.
Fig. 7. Illustration of second-generation FIMS with 3D color-coded AE activity displays and client-
specified zoning matrix for remaining refractory thickness display.
by a transducer. The reflections can be caused by the boundary between the molten
metal/build-up material, altered/unaltered refractory, cracks, gaps, and joints (Fig. 8).
Thus, AU-E is useful for detecting remaining refractory, boundaries of altered/unal-
tered refractory, and locating cracks/gaps.
190 A. Sadri et al.
Fig. 8. A typical AU-E signal transmitting–receiving through furnace wall.
One of our inspections involved the inspection of several identical electrical fur-
naces. However, the refractory was purchased at different times and from a different
brand. This indicated that the refractory property might vary. To investigate the possible
variation of the refractory property, several sample bricks were used for calibration and
each batch, and the results indicated that there was only ~1% variation between the
refractory batches.
The AU-E data analysis and interpretation are the key to determining correct thick-
nesses and detecting flaws and build-ups. The collected signals are first filtered to
remove any noise, then used to create an image of the sidewalls of the inspected vessel.
The frequency-domain data is converted into a pseudo-thickness domain dataset. The
thickness plots are colorized, in which, high-amplitude reflections take on a red color,
and low-amplitude reflections take on a blue color. Multiple data points were then
shown according to their position within the furnace, and an outline of the furnace is
plotted alongside the data to complete the colorized dataset (Fig. 9).
The right side of Fig. 9 indicated the remaining refractory thickness (in blue), the
build-up extent (in green), and the location of cracks/gaps (in red). The anomaly map
illustrated that the furnace wall consisted of a significant number of anomalies. When
doing a more detailed analysis, it was found that most of the gaps were found between
11 and 13 inches measured from the cold face of the refractory. This could be related to
refractory interfaces (Fig. 10).
When using AU-E inspection, it is recommended that an annual inspection be done
in order to develop the wear trend of the furnace (Fig. 11). When it is close to the end
of the campaign life, the AU-E inspection should be increased to biannually or
quarterly.
Fig. 9. Visualization of AU-E results.
Fig. 10. Location of anomalies of the inspected furnace wall.
Fig. 11. Wear trend of refractory lining.

192 A. Sadri et al.
12 Conclusion
Throughout the campaign life of a furnace, successful refractory maintenance proce-

dures will ultimately reduce relining costs, furnace downtime, waste disposal volume,
and carbon footprint of the smelter. A reliable maintenance program requires the iden-
tification of risk areas, the use of proper monitoring tools and techniques, data collec-
tion and analysis, and the development of predictive and proactive maintenance actions.
As an example, currently available continuous monitoring techniques, FIMS, and peri-
odic inspection techniques, AU-E would allow the achievement of predictive refractory
conditions and hence, the development of proactive measures.
Acknowledgments The authors would like to thank their friend and colleague Mr. Terrence (Terry)
Koehler for his review of the paper and his suggestion and contributions.
References
1. Bennett JP, Kwong KS. Refractory recycling-concept to reality. Environ Issues Waste Manag
Technol VII Ceram Trans. 2002;132:3–15.
2. Donald JR, Walsh MK, Burnham CD, Lauzon PH. Dealing with spent refractory materials. https://
p2infohouse.org/ref/31/30773.pdf.
3. Angara Raghavendra PK. Recovery of materials from recycling of spent furnace linings. Masters
Theses 4631. 2008. https://scholarsmine.mst.edu/masters_theses/4631.
4. Smith JD, Peaslee KD, Barnes AS, Fang H. An economic, logistic, and technological approach to
refractory recycling. In Proceedings of UNITECR’97. p. 17–25; 1997.
5. HSE Report: the explosion of No. 5 Blast Furnace, Corus UK Ltd, Port Talbot. 8 November 2001.
http://www.hse.gov.uk/pubns/web34.pdf. published by the Health and Safety Executive 09/08.
6. Webb M. Assmang furnace explosion death toll climbs. 27 February 2008. www.miningweekly.
com/article/assmang-furnace-explosion-death-toll-climbs-2008-02-27.
7. Onstad E. Furnace shutdown costs world No. 3 platinum miner $4 million. 24 June 2009. http://
www.mineweb.com/mineweb/content/en/mineweb-platinum-group-metals?oid=85427&sn=Det
ail.
8. Northern Miner, Daily News. Vale takes hefty write-downs. 21 December 2012. http://www.north-
ernminer.com/issuesV2/VerifyLogin.aspx.
9. de Bruyn C. Angloplat shuts Polokwane smelter. 5th November 2008. www.miningweekly.com/
article/angloplat-shuts-polokwane-smelter-2008-11-05.
10. Braun W, Gebski P, MacRosty R, Pula R, Sadri A, Ying W. Tap-hole monitoring technologies.
In: Furnace tapping conference 2014, April 27–28. The Southern African Institute of Mining and
Metallurgy, Johannesburg, South Africa. p. 169–182.
11. Sullivan GP, Pugh R, Melendez AP, Hunt WD. Operations & maintenance best practices, a guide to
achieving operational efficiency. Release 3.0., prepared by Pacific Northwest National Laboratory
for the Federal Energy Management Program U.S. Department of Energy. p. 5.15.5; 2010.
12. Fitch JC. Proactive maintenance can yield more than a 10-fold savings over conventional predic-
tive/preventive maintenance programs. Viewed May 2012. n.d. http://www.oilfiltrationsystems.
com/media/pdfs/case_studies/proactive.pdf.
Green Hydrogen for Steel Production: A Case Study
Gareth Richardson1, Peter Appleby1, Maria De Campos2, and Sina Mostaghel2 (*)
1
Atkins, Bristol, UK
SNC-Lavalin, Toronto, ON, Canada
2
sina.mostaghel@snclavalin.com
Abstract. Decarbonization of the steel sector is currently receiving significant attention

in many countries, mostly focusing on the direct reduction of iron using hydrogen.
However, unless green hydrogen from renewable energy sources becomes available in
sufficient quantities and at an economical cost, the overall decarbonization of steelmaking
remains a challenge. This paper covers the learnings from a recent feasibility study com-
pleted for the conversion of a direct reduction iron (DRI) and associated electric arc fur-
nace (EAF) plant to a green steel plant. The feasibility study centered on the transition
from hydrogen, produced by natural gas combustion, to a green hydrogen supply system,
based on electrolysis with a power supply mix of renewables and nuclear energy. Four
areas of interest are laid out in this paper: (a) electricity source and its impact on the level-
ized cost of hydrogen, (b) electrolyser plant types and suitability for integration, (c) value
stacking through electrical load management and benefits for plants with electric arc
furnaces, and (d) impact of high-purity hydrogen on the DRI plant and the necessary
adjustments to be made. The identified benefits include producing minimized carbon
dioxide during the production of the steel, possible increased DRI capacity, mitigation of
EAF impact on grids, and tuneable carbon content of the DRI.
Keywords: Green hydrogen · Steelmaking · Carbon footprint · CO2 emission
1 Introduction
Currently, the majority of the global steel production is via one of the two process path-
ways; the integrated Blast Furnace – Basic Oxygen Furnace (BF-BOF) or the Direct
Reduced Iron – Electric Arc Furnace (DRI-EAF). DRI-EAF plants can utilize coal in
the rotary kiln process route or Natural Gas (NG) in the shaft or fluidized bed process
routes. The NG-DRI pathway has relatively low-specific carbon emissions of <1.4
tCO2/tsteel compared to the BF pathway at ~1.9 tCO2/tsteel [1]. There are two main tech-
nology suppliers of NG-DRI process route; namely, MIDREX and ENERGION with
~60% and 12.4% market shares in 2020, respectively [2]. In both technologies, the NG
can be reformed separately using Steam Methane Reforming (SMR) or via an inte-
grated reforming process. This case study focuses on the replacement of an SMR by a
water electrolyser, feeding an Energion DRI plant. Figure 1 shows a block flow diagram
(BFD) of the process in which NG is reformed to syngas in an SMR unit; the gas is
composed primarily of CO (~15 Vol%) and H2 (~73 Vol%), providing both reductants
and carbon for carburization of the iron.
https://doi.org/10.1007/978-3-031-17425-4_28
194 G. Richardson et al.
Heat Water
Iron Ore Recovery Condenser
Compressor
Water
Steam
Heater CO2 Removal SMR
Natural Gas
Water
Condenser
Oxygen
Natural Gas
DRI
Fig. 1. BFD of the original process, NG reformed in a SMR unit.
Heat Water
Iron Ore Recovery Condenser
Compressor
Sea Water
Reverse
Osmosis
Heater
CO 2 Removal
Ion Exchange
H2 Buffer
Vessels
Separators
Natural Gas Oxygen
DRI
Electrolyser
Fig. 2. BFD of the modified flowsheet, water electrolysis process.
The reformer syngas is then blended with a recycle stream from the reactor contain-
ing CO2, then passed through heat recovery, and fired in the furnace using tail gas and
NG to reach a temperature in excess of 900 °C. To increase the temperature to above
1000 °C for improved reaction kinetics, O2 is added to combust some of the methane
and hydrogen from the SMR.
The aim of the conducted feasibility study was to replace the SMR with hydrogen,
produced via water electrolysis, changing the block flow diagram to what is shown in
Fig. 2.
Green Hydrogen for Steel Production: A Case Study 195
2 Electrolyser Types and Suitability for Integration with DRI
2.1 Electrolyser Types
Apart from several types of electrolysers under investigation and at research and devel-
opment stage, currently, three types are mainly used at industrial-scale; namely, alka-
line, proton exchange membrane (PEM), and solid oxide cell (SOC).
Alkaline electrolysis is the most-established water electrolysis technology that has
been used for industrial-scale applications since 1920 and is commercially available
from multiple vendors. Large-scale plants have been installed up to a capacity of
3000 kg H2 h – or ~162 MW(e) input; however, currently, the typical scale of the green
hydrogen projects is in the range of 1–10 MW(e).
Proton exchange membrane (PEM) technology, also known as polymer electrolyte
membrane, is a more recent method than alkaline electrolysis. PEM electrolysis is a
mature technology that is commercially available from multiple vendors. However, it is
yet to be established at the same scale as alkaline electrolysis, the current capacity is
20 MW.
Unlike low-temperature electrolysers (Alkaline or PEM), SOC technology is a high-
temperature process, which is fed by steam. The required input steam conditions can
vary significantly between units that are provided by different manufacturers, from
120 °C to above 700 °C; however, 150 °C is the most commonly used temperature. Heat
is then recovered from the outlet cell flows and a final electric trim heater is used to
reach the operating temperature of 500–1000 °C. While SOC systems are commercially
available at MW(e) scale, operation at this capacity in an industrial setting is still under
development.
A summary of the key technical parameters for each electrolyser type is shown in
Table 1. The information is based on recent projects and does not represent any single
original equipment manufacturer (OEM).
The key outcomes of the comparison in Table 1 are that alkaline and PEM electroly-
sers are close in performance and LCOH. PEM provides the greatest flexibility in terms
of ramp rate with a comparable specific energy to that of alkaline. The specific electri-
Table 1. Key parameters of electrolyser types.

LCOHa in
Specific 2021
Specific Stack Discharge Ramp CAPEX @30$/
Type system energy operating life pressure rate 2021 MWhe
Units kWhe/kg Hours Bar s USD/kW USD/kg
Alkaline 55–60 80,000 1–30 ~30 500–800 2.34
PEM 55–58 80,000 1–40 <2 600–900 2.37
SOC 38–41 40,000 1–2 900–1200 1.94
(Integrated)
SOC (Elec. 44–46 2.18
only)
LCOH stands for levelized cost of clean hydrogen production
a
cal consumption of SOC is ~30% lower, when steam can be supplied. SOC can be
operated with an electric heater in the feed which adds the latent heat of vaporization,
leading to an increase in specific energy consumption, which brings the LCOH more in
line with alkaline and PEM. It should be noted that the lowest cost electrolysers on the
market are currently from China, reaching values as low as 300–400 USD/kW which
can lower LCOH to 2.14 USD/kg.
2.2 Integration of Electrolysers
There are many factors to consider when integrating an electrolyser into a brownfield
site, including utility requirements, footprint, heat integration, and reliability.
Utilities There are five key utilities to be taken into account:
• Steam – The loss of steam produced by the SMR will require consideration and will
likely require the replacement of several pump/compressor turbine drives with elec-
tric motors.
• Nitrogen – Required for vent stack, the safe shutdown, and purging of the electroly-
ser plant, which can be supplied from the site or a new liquid N2 storage and evapo-
ration system.
• Compressed air – The majority of the automated valves will be air operated due to
the flammability risk of hydrogen.
• Cooling water – As the electrolysers are not 100% efficient at low temperatures,
significant amounts of heat may be discharged, typically 20–30% of the electrical
load, depending on the amount of degradation of the electrolyser stack. Low-
temperature electrolysers typically run at between 50 and 90 °C, with a cooling
water inlet of ~20–30 °C. High-temperature electrolysers, SOC, have much lower
system cooling requirements due to the higher efficiency and heat recovery in the
system which intrinsically requires heat input.
• Feed water – Deionized and demineralized water purity grade is required for elec-
trolyser feedwater; therefore, water treatment is dependent on the source of water. In
this case, a seawater Reverse Osmosis (RO) plant is utilized with the second pass of
the RO plant feeding a mixed bed ion exchange (MBIX) or electrodeionization
(EDI) plant to achieve the required demineralized grade water.
Footprint The footprint of an electrolyser facility is primarily dependent on the tech-
nology selected. Table 2 shows the indicative sizes for different technologies based on
Table 2. Indicative required size for different technologies.

Power required Power density Footprint
Technology MWDC m2/MWDC Hectares
Alkaline 647 101 6.5
PEM 647 69 4.5
SOC (Steam) 459 100 4.6
SOC (Elec) 541 101 5.5
the complete plant, single level, and a capacity of 100,000 tonnes per annum (TPA) of
hydrogen.
A number of electrolyser plants are being designed to be housed in multistory build-
ings, allowing the footprint to be significantly reduced, e.g., by 25–30% for a two-story
design.
Heat Integration Heat integration is only of benefit for SOC-based electrolyser
plants. Significant modifications are required to a brownfield DRI plant for full integra-
tion, allowing the highest electrical efficiency through the utilization of high-grade
heat. As DRI plants are typically integrated with EAFs, it is possible to recover waste
heat from the EAF off-gas system. The batch operation of the EAF requires a thermal
buffer system. If there is more than one EAF on-site, then a single thermal store can be
used to reduce the CAPEX and increase steam availability to the continuous level that
is required for SOC. A steam generator is then used to create a continuous supply of
saturated steam at a temperature suitable for the SOC system, typically saturated steam
at 5 bar is required. As such, a single 150-tonne EAF could supply ~51 tonne/h of
steam, depending on EAF operation. Atkins experience indicates that a system CAPEX
of ~$20–25 million per EAF can be assumed for a waste heat boiler and accumulator
system. A SOC system typically has a steam to hydrogen ratio of 9–12 kgSteam/kg H2
with the ratio depending on the SOC system design. Assuming a ratio of 9 kgSteam/kgH2
then a total hydrogen production from SOC of ~5700 kg/h can be realized from a single
EAF. The hydrogen requirement for DRI production is between 47 and 68 kgH2/
tonneDRI, assuming a value of 57.5 kg H2/tonneDRI which is just above the stoichiometric
amount required for hematite of 54 kg H2/tonneDRI [3], it is equivalent to 100 tonne/h of
steel or 0.88 million TPA.
Reliability A DRI plant requires a high level of reliability from the hydrogen supply.
Electrolysers are inherently modular and the cell stacks can range in size from a few
kW to multi MW. Hence, a hydrogen plant will be formed of 100’s of cell stacks.
Manufacturers are increasing the size of the stacks to reduce CAPEX, with some of the
largest single stacks being above 2 MW.
The cell stacks degrade over time with degradation rates of between 0.5% and 2%
per year, leading to increased specific energy consumption. As such, electrolyser plants
can be designed with a small amount of additional capacity to allow the isolation of
either individual cell stacks or modules without impacting the production rate. Large-
scale plants will typically utilize combined balances of plants supplying the electroly-
ser modules. The redundancy of cooling tower cells, pumps and valves, compressors,
and transformers can be sufficient to allow for continuous hydrogen production in the
event of a failure of one of these components. The highly modular nature of the elec-
trolyser facilities and limited common modes of failure should yield an improvement in
reliability over the traditional SMR.
2.3 Effect of Electricity Cost on LCOH
The specific energy consumption and the specific CAPEX are the two critical parameters
in determining LCOH. As shown in Fig. 3, specific energy consumption is most impact-
ful at electrical costs above 20 USD/MWh. CAPEX becomes more important when
electrical prices are significantly lower. It should be noted that 100% utilization of the
electrolysers is considered in developing this sensitivity analysis and no possible benefits
from the utilization of oxygen or providing grid services have been taken into account.
2.4 Effect of Capacity Factor
A key contributor to the cost of hydrogen production is the availability of input electri-
cal power. Should a hydrogen production plant be fed purely by renewable power,
assuming no electrical storage, it must be significantly oversized and hydrogen storage
utilized to accommodate the low capacity factor. Table 3 shows an assessment of the
effect of the power supply capacity factor on LCOH with three scenarios.
The effect of capacity factor on overall LCOH, without inclusion of hydrogen stor-
age costs, is illustrated in Fig. 4. It is evident that even when considering the lower
electrical cost for Solar PV ($15/MWh) and Solar PV/Wind ($22/MWh) scenarios over
the 100% capacity scenario ($28/MWh), significant savings can be obtained from max-
imizing capacity factor of the plant.
Fig. 3. LCOH of Technologies vs. Cost of Electricity.
Table 3. Effect of power supply capacity on LCOH.

Capacity factor
Case (%) Required plant nameplate capacity MW for 100 kTPA H2
Solar PV 33 303
Solar PV and wind 68 147
Renewables and nuclear 100 100
Fig. 4. Effect of capacity factor on LCOH.
3 Value Stacking Through Electrical Load Management

and Benefits for Plants with Electric Arc Furnaces
Steel facilities operating EAF will have significant fluctuation in electricity demand,
resulting from the intermittent operation of the EAFs. Electrolysers can alter produc-
tion rates to offset these sudden fluctuations in steel facility power demand. Where
electrolysers are used to provide grid support, additional electrolyser capacity is
required. This allows the electrolysers to cope with the additional production necessary
to offset the largest steel plant drop in power consumption. To maintain continuous
hydrogen supply to the steel plant, hydrogen storage is required.
Assessment of the benefits of electrolysers for a steel plant case with 100 MW base-
load electricity requirement and 100 MW variable load, resulting from EAF intermit-
tent operation, is examined below, Table 4. Three cases were considered, each with the
requirement for 100 kTPA hydrogen fed to the steel plant. For these scenarios, the total
EAF cycle time has been assumed to be 50 minutes with a 40-minute EAF operation
time and 10 minutes shut down, prior to the start of the next cycle. The electricity price
has been assumed to be $35/MWh.
4 Impact of Using Near 100% Green Hydrogen on the DRI

Plant and the Necessary Adjustments to Be Made
4.1 Hydrogen Feed and Its Reaction Effects
DRI plants run with H2 in the feed from 50 to 80 Vol%, depending on the vendor and
plant set-up. Atkins experience is that DRI plants with non-integrated reformers have
the highest levels of hydrogen in feed, typically >70 Vol%. The metallization degree of
the DRI has been found to be equivalent to the NG-DRI process route [4]. The thermo-
dynamics and kinetics behaviors of iron oxide reduction with hydrogen have been
examined by others [5], showing that hydrogen reduction has much faster reaction
kinetics. At the operating temperatures of the reactor under study, both CO and H2 gases
Table 4. Benefits of the electrolyser for a steel plant with 100 MW base load and 100 MW variable
load in three different scenarios.
Grid support – no Grid support − 5%
Base overload overload
Name plate electrolyser capacity (MW) 647 718 689
H2 storage capacity (t) – 5.1 5.1
Overall additional cost over 20 years (MM$) – 43.9 28.2
have the same reduction potential. However, H2 has a much higher effective diffusion
rate of 5.835 × 104 m2/s at 900 °C versus carbon monoxides diffusion rate of
0.124 × 104 m2/s. The diffusion rate difference is driven by the smaller molecular size
of H2 and the lower viscosity this allows reduction to occur quicker and at lower
temperatures.
It is expected that the improved reactor kinetics from using green hydrogen will
make it possible to increase reactor throughput. In the case study under investigation,
an increase of approximately 10% may be achievable.
4.2 Effect on Plant Equipment
The composition of the recycle gas from the reactor will change. Producing a composi-
tion at the outlet of hydrogen, water, and some CO2 depends on any natural gas injec-
tion in the reactor cone. The change in composition results in a much lower molecular
weight gas at the compressor inlet leading to an increase in recycle compressor energy
demand. As the compressor will typically be turbine driven from steam generated in the
natural gas reformer then this will also need to be changed to an electrical motor drive.
The reduced level or possible elimination of CO2 in the recycle gas will mean the
CO2 absorption system will either not be required or will have a reduced load. If there
is a suitable location to sequester CO2 then it is possible to have an increased carbon
content in the DRI through injection of natural gas in the cone and reducing gas while
still being able to achieve very low levels of CO2 emissions.
Hydrogen temperature from the electrolysers will be dependent on the operating
temperature, compression needs, and the amount of water that can be allowed in the
feed to the reactor. The typical operating temperature is >1000 °C; however, with the
improved reaction kinetics from using a high proportion of hydrogen, the operating
temperature can be slightly lower ~900 °C. Reducing gas temperature is therefore sig-
nificantly above low-temperature electrolysis outlet temperatures which are <100 °C. It
is possible to reuse the recycled gas furnace that currently uses natural gas to boost the
temperature. However, to mitigate the emissions, this would need conversion to burn
some of the recycled hydrogen from the reactor or the installation of a new electric
resistance heater. The use of SOC technology for all or part of the hydrogen demand
can reduce or eliminate the need for a preheater there are ongoing studies at the
University of California Irvine funded by the US Department of Energy [6] in this area
of research.
5 Concluding Remarks
Electrolysers can be effectively integrated into existing brownfield steel plants. The
cost of electricity has a significant effect on project economics along with the capacity
factor of the electricity supply so both should be considered carefully.
Three major electrolyser types that are currently available in the market are alkaline,
PEM, and SOC. A mixture of PEM and alkaline offers the benefit of PEM quick
response times and corresponding grid integration along with alkaline technology
maturity. Later, this decade SOEC is likely to take a larger market share as the technol-
ogy matures and efficiency benefits can be realized.
Higher hydrogen concentrations, provided by electrolysis, inside the DRI reactor
improve reaction kinetics and allow the plant to operate at lower temperatures. These
improved kinetics could also yield 10% additional DRI production.
When integrating an electrolyser plant into a brownfield steel facility, the key issues
that need to be considered are reduction in steam production, EAF heat recovery
options, and utility requirements.
Atkins has successfully completed a feasibility study for conversion of a world-scale
DRI production facility to operate using hydrogen produced from electrolysis and is
now working through concept development. As the findings in this paper show, Atkins
has excellent insight into the subject and is helping lead the way in the decarbonization
of key industries.
References
1. EPRS. Carbon-free steel production. [Online]. Available: https://www.europarl.europa.eu/

thinktank/en/document/EPRS_STU(2021)690008.
2. Midrex. 2020 World Direct Reduction Statistics. [Online]. Available: https://www.midrex.com/
wp-content/uploads/Midrex-STATSbookprint-2020.Final_pdf.
3. Patisson F, Mirgaux O. Hydrogen ironmaking: how it works. Metals. 2020;10(7):922. https://doi.
org/10.3390/met10070922.
4. Pauluzzi D, et al. CFD study of an energiron reactor fed with different concentrations of hydrogen.
[Online]. Available: https://www.energiron.com/wp-content/uploads/2021/09/MARCH-2021_
AIST-IRON-STEEL-TECHNOLOGY.pdf.
5. Zuo HB, et al. Reduction kinetics of iron oxide pellets with H2 and CO mixtures. [Online].
Available: https://www.researchgate.net/publication/282499310_Reduction_kinetics_of_iron_
oxide_pellets_with_H2_and_CO_mixtures.
6. DOE. CX-101959: Solid Oxide Electrolysis Cells (SOEC) integrated with Direct Reduced Iron
(DRI) plants for producing green steel. [Online]. https://www.energy.gov/nepa/downloads/
cx-101959-solid-oxide-electrolysis-cells-soec-integrated-direct-reduced-iron-dri.
Stabilizing Chromium in Ferrochromium Slag
for Utilization as Aggregate Material
Tahmeed Bin Tasnim1 and Leili Tafaghodi2 (*)
BC Research Inc., Richmond, BC, Canada

1
2
Department of Materials Science and Engineering, McMaster University, Hamilton, ON,
Canada
tafaghodi@mcmaster.ca
Abstract. The Canadian economy relies largely on natural resources, yet processing of
these resources and/or associated by-products is well known to negatively affect the envi-
ronment. Reusing slag, a common waste stream of pyrometallurgical extraction pro-
cesses, as construction materials serves to reduce the impact on the natural resources
currently used in construction industry such as sand, gravel, and crushed stone. Every
year 12–16 million tonnes of ferrochromium slag is produced, a majority of which is
dumped in landfills. The mechanical properties of this slag make it a potential material to
be used as an inexpensive construction material, the utilization of which can reduce the
use of natural resources. Despite its potential use, ferrochromium slag is treated as a waste
due to environmental and health concerns regarding the leaching of its heavy metal con-
tent, the most concerning of which is carcinogenic chromium (VI). Research has shown
that the spinel phase in ferrochromium slag stabilizes chromium by trapping it in the
spinel structure and preventing its leaching to the environment. This study examined slag
samples of the MgO-Al2O3-SiO2-CaO-FeO-Cr2O3 synthesized at 1650 °C. The slag was
synthesized to mimic the possible composition of the Ring of Fire’s ferrochromium slag.
Samples were heat treated before quenching. Leaching tests on heat-treated slag samples
showed that samples held at 1400 °C have the lowest chromium leachability. Higher Cr
leaching is observed from samples as the Al2O3 content increases. An increase in basicity
in the range of 0.3–0.7 increases the amount of Cr released from the samples during
leaching experiments.
Keywords: Ferrochromium · Slag · Reutilization
1 Introduction
Ferrochrome production involves the production of slag as the by-product at a rate of

1.1–1.5 times that of the metal itself. This results in a global ferrochromium slag pro-
duction of around 12–16 million tonnes annually. Although most of this slag is dis-
carded as waste in landfills, it is an attractive construction material due to its suitable
mechanical properties. Ferrochromium producers from Finland, Sweden, and Russia
have utilized slag in road making and building materials [1]. However, the use of fer-
rochromium slag as a construction material is still in its infancy mainly due to
https://doi.org/10.1007/978-3-031-17425-4_29
204 T. B. Tasnim and L. Tafaghodi
environmental concerns about the content and leachability of heavy metals, especially
chromium. Environmentally safe utilization of this slag would offset the cost of produc-
tion and reduce the impact on the natural resources used as these aggregates.
The leaching of carcinogenic Cr(VI) from ferrochromium slag is a major environ-
mental concern for its application in the construction industry. The slag composition is
expected to have a strong effect on the leachability of Cr from the slag.
This study examines the chemical composition of a synthetic ferrochromium slag
for its utilization as construction materials. The main concern with the utilization of
ferrochromium slag as a construction material is the leachability of Cr and its oxidation
to carcinogenic and environmentally hazardous Cr(VI). Canadian and Ontario criteria
require a maximum 2.77 mg/L of total chromium in aqueous waste leachate using
TCLP (Toxicity Characteristic Leaching Procedure) leaching method [2].
2 Methodology
The chemical composition of the synthesized samples with variations in Al2O3 content
and basicity (CaO/SiO2) is shown in Table 1. Reagent-grade oxides were mixed in an
agate mortar before being pressed inside a MgO crucible. The sample was then hung
inside a vertical tube furnace using Molybdenum (Mo) wire. Following melting, the
temperature was reduced to 1400 °C and held for 12 h. After the holding time, the
sample is dropped in a bucket of cold tap water for quenching, by cutting the Mo wire.
Quenched slag samples were extracted, milled, and sieved to <106 μm for leaching and
chemical analysis.
Previous studies have examined the leachability of Cr from EAF using various
leaching tests such as DIN 38414-S4 [3, 4]. However, in the present study, the leaching
experiments were conducted using AV002.1 availability test [5] on slag samples with a
particle size of <106 μm using the extraction liquid of 50 mM EDTA. The pH of the
solution was adjusted to 7.5 before leaching by using 2 N nitric acid (HNO3) solution
and 1 N potassium hydroxide (KOH) solution. One gram of <106 μm slag with 100 ml
Table 1. Chemical composition (in wt%) of synthetic slag.

Series Basicity (CaO/SiO2) Al2O3 MgO SiO2 FeO CaO Cr2O3
A1/B3 0.5 20.00 25.00 30.00 4.00 15.00 6.00
A2 0.5 15.00 25.00 33.33 4.00 16.67 6.00
A3 0.5 12.00 25.00 35.33 4.00 17.67 6.00
A4 0.5 9.00 25.00 37.33 4.00 18.67 6.00
A5 0.5 6.00 25.00 39.33 4.00 19.67 6.00
A6 0.5 3.00 25.00 41.33 4.00 20.67 6.00
A7 0.5 0.00 25.00 43.33 4.00 21.67 6.00
B1 0.3 20.00 25.00 34.61 4.00 10.39 6.00
B2 0.4 20.00 25.00 32.14 4.00 12.86 6.00
B4 0.6 20.00 25.00 28.13 4.00 16.88 6.00
B5 0.7 20.00 25.00 26.47 4.00 18.53 6.00
Stabilizing Chromium in Ferrochromium Slag for Utilization as Aggregate Material 205
of EDTA extraction liquid was placed in 150 ml high-density polypropylene (HDPP)

containers. The extraction containers were tightly sealed and set to rotate in an end-
over-end fashion for 48 h at 35 rpm. The leachate was then filtered to remove the solid
residue. The leachate was analyzed with ICP-AES [6].
Slag samples A1-A7 were prepared with Al2O3 content varying from 0% to 20%, to
study the effect of Al2O3 content on the Cr leachability of FeCr slag. The Cr concentra-
tion of the leachate obtained from slag samples A1–A7 with particle size <106 μm is
shown in Fig. 1. The Cr concentration increases from 0.10 to 0.18 ppm as the Al2O3
content increased from 0% to 20%. The increase in Al2O3 content of the slag promotes
Cr leaching out of it. It is proposed that Al3+ ions replace the Cr3+ ions in the stable
spinel phase that resists Cr dissolution in water. The Cr present in silicate phases
(Åkermanite, olivine, quartz) of the matrix would not be able to completely resist water
dissolution and thus leach out of the slag.
Slag samples B1-B5 with basicity ranging from 0.3 to 0.7 were made to determine
its effect on Cr leachability of FeCr slag. A noticeable Cr concentration difference in
the leachates is observed by increasing basicity. The leachate of slag B1 with basicity
of 0.3 has 0.15 ppm Cr. The Cr concentration increases to 0.25 ppm for the leachate of
sample B5 with the basicity of 0.7 (Fig. 2).
The Cr content of the leachate of all the slag samples examined in this work are
below the Canadian and Ontario criteria of maximum 2.77 mg/L of Cr in waste leachate
using TCLP test. The Cr content of the leachates is also lower than the non-potable
ground water limit of 0.81 ppm total Cr but higher than Ontario’s total Cr content stan-
dard limit of 0.05 ppm for drinking water [7].
0.20
Cr Concentration (ppm)
0.18
0.16
0.14
0.12
0.10
0.08
0 3 6 9 12 15 20
Alumina Content (wt%)
Fig. 1. Cr concentration of leachate as a function of Al2O3 content of slag using AV002.1 availability
test.
206 T. B. Tasnim and L. Tafaghodi
0.30
Cr Concentration (ppm)
0.25
0.20
0.15
0.10
0.3 0.4 0.5 0.6 0.7
Basicity
Fig. 2. Cr concentration of the leachate vs. basicity of the slag using AV002.1 availability test.
Acknowledgments The authors would like to thank Dr. Dogan Paktunc for his valuable insight. We
would also like to acknowledge NRCan’s financial support.
References
1. Ndlovu S, Simate GS, Matinde E. Waste production and utilization in the metal extraction industry.
CRC Press, Taylor and Francis; 2017.
2. Ontario Environmental Protection Act R.R.O 1990, Regulation 347, General Waste Management.
p. 65–145.
3. Kuhn M, Mudersbach D, Baena Liberato JM, De Angelis V, Capodilupo C, de Fries U. Chrome
immobilisation in EAF slags from high-alloy steelmaking: development of a slag treatment pro-
cesses. In: European Commission, technical steel research, steel-making processes; 2006.
4. Romera Q, Capodilupo, Behmenburg, Kuhn. Decreasing the scorification of chrome. In: European
Commission, technical steel research, primary steel-making; 2000.
5. Kosson DS, van der Sloot HA, Sanchez F, Garrabrants AC. An integrated framework for eval-
uating leaching in waste management and utilization of secondary materials. Environ Eng Sci.
2002;19(3):159–204.
6. Bin Tasnim, T., Tafaghodi Khajavi, L. Chromium Stabilization in Ferrochromium Slag for its
Utilization as Aggregate Material. J. Sustain. Metall. 2022;8:1041–1052.
7. Soil, Ground Water and Sediment Standards for use under Part XV.1 of the Environmental
Protection Act. Ontario Ministry of Environment, Canada, 15 April 2011.
Iron Segregation Roasting as a Potential
Decarbonization Pathway for Ironmaking
P. Kerr1 (*), Q. Liu2, and T. H. Etsell2

1
MINEPROMET, Quebec City, QC, Canada
pkerr@ualberta.ca
2
University of Alberta, Chemical and Materials Engineering, Edmonton, AB, Canada
Abstract. Ironmaking accounts for 80% of steelmaking greenhouse gas (GHG) emis-
sions, while steelmaking accounts for 8% of global GHG emissions. These significant
contributions to GHG emissions illustrate the value of decarbonizing ironmaking through
the implementation of iron segregation roasting where applicable.
Keywords: Iron segregation · Ironmaking · Direct reduced iron (DRI)

Decarbonization · Segregation roasting
1 Introduction
Ironmaking accounts for 80% of steelmaking greenhouse gas (GHG) emissions, while
steelmaking accounts for 8% of global GHG emissions. These significant contributions
to GHG emissions illustrate the value of decarbonizing ironmaking through the imple-
mentation of iron segregation roasting where applicable.
Iron segregation is ideally used with difficult-to-beneficiate iron ores where the ore
is roasted at an elevated temperature with a carbonaceous reductant and a chloride addi-
tive in a closed reactor to allow for reducing conditions and contain the volatile chlo-
rides. Although iron segregation roasting need not be limited to complex iron ores,
oxide minerals of aluminum, titanium, and chromium are complex in nature and are
frequently associated with ferrous/ferric oxides, vanadium, niobium and tantalum
oxides, and rare-earth-bearing oxides [1]. Iron segregation roasting provides a potential
upgrading solution for these complex ores as it allows for the ferrous and ferric oxides
to be separated and upgraded to metallic iron thereby simplifying subsequent process-
ing of the desired elements while producing a widely used value-added product. Powder
metallurgy samples generated in this manner have high purity, ideal physical properties
for powder metallurgical processing due to their cubic structure while they are resistant
to oxidation at atmospheric conditions and are, in turn, not pyrophoric [2]. The iron
product that does not meet the requirements for powder metallurgy can be subsequently
briquetted and transported to an electric arc steelmaking facility or fed into an iron blast
furnace as a scrap iron substitute.
https://doi.org/10.1007/978-3-031-17425-4_30
208 P. Kerr et al.
The chemistry of segregation is essentially the same for iron as for copper and nickel
where a halide salt is hydrolyzed by steam in the presence of gangue (usually a silicate)
producing substantial HCl exothermically at an elevated temperature, but significantly
below the melting point of the metal (~950 °C for iron):
SiO2 s CaCl 2 s H 2 O v CaSiO3 s 2HCl g (1)
The HCl then reacts endothermically and relatively slowly with the metal oxide in
turn mobilizing iron for transport:
FeO s 2HCl g FeCl 2 g H 2 O g (2)
Hydrogen is formed endothermically by the carbonaceous reductant:
C s H 2 O v CO g H 2 g (3)
The metal chloride is then reduced by hydrogen on or near the carbonaceous reduc-
tant. In turn, segregating it from the metal oxide’s original position and liberating HCl
to react with additional metal oxide completes the cycle and resulting in a catalytic
behavior as a less than stoichiometric amount of chloride:
FeCl 2 g H 2 g Fe s 2HCl g (4)
The process is followed by some form of physical separation to liberate the metallic
iron from the gangue.
Iron segregation roasting lends itself well to decarbonization as the reactor type is
ideally an indirect fired kiln or screw conveyor furnace, in order to contain the volatile
chlorides, which act as a transport catalyst within the reactor [3–11]. Efforts to develop
a direct-fired process have been studied and they have been met with limited success
[12]. In turn, an electrically heated reactor can take advantage of a renewable electrical
energy source.
An additional benefit that comes from the transport catalyst phenomena is that com-
minution energy requirements can be significantly reduced by avoiding the necessity of
fine grinding as the iron is transported away from the gangue material. In turn, capital
expenditures can also be reduced in this case, as less grinding equipment is needed.
Unlike coke required to maintain bed stability and a reducing atmosphere in tradi-
tional blast-furnace ironmaking, iron segregation roasting can use different types of
renewable biochar, as the bed is not fluidized.
At the same time, hydrogen injection at the end of the process can also be used to
increase metallization and further reduce the amount of carbon reductant necessary for
ironmaking. Although the exact ratios possible have not been optimized, the concept
has been proven [4].
While iron segregation roasting has only been tested in the laboratory and up to the
small-pilot scale, it holds great potential for difficult to beneficiate ores, especially in
the context of decarbonization in ironmaking.
Iron Segregation Roasting as a Potential Decarbonization Pathway for Ironmaking 209
Acknowledgments We thank Dr. Edward Bertram for providing literature and guidance regarding
grain enlargement. This work was supported by grants from the Natural Sciences and Engineering
Research Council of Canada (NSERC).
References
1. Sanchez-Segado S, Makanyire T, Escudero-Castejon L, Hara Y, Jha A. Reclamation of reactive

metal oxides from complex minerals using alkali roasting and leaching – an improved approach to
process engineering. Green Chem. 2014;17:2059–80.
2. Williams CE, Henderson AS. Method of producing crystalline iron by the hydrogen reduction
of ferrous chloride. U.S. Patent No. 2,418,148A. Washington, DC: U.S. Patent and Trademark
Office; 1947.
3. Stewart DF, Pollard LJ. Process of beneficiating titaniferous ores in the presence of hydrogen
chloride. U.S. Patent No. 3,854,929. Washington, DC: U.S. Patent and Trademark Office; 1974.
4. Bertram E, Harris P, Myroniuk DW. Beneficiation process for oxidized iron ore. U.S. Patent No.
4,246,025. Washington, DC: U.S. Patent and Trademark Office; 1981.
5. Fensom DJ. Processes of iron segregation. U.S. Patent No. 4,295,878. Washington, DC: U.S. Patent
and Trademark Office; 1981.
6. Reeves JW, Zander BH, Ericson AS. Method for beneficiating titanium-bearing material contain-
ing iron. U.S. Patent No. 5,660,805. Washington, DC: U.S. Patent and Trademark Office; 1997.
7. Al-Ajeel AA, Mahdy SN. Beneficiation of Al-Hussainyiat low grade iron ore by segregation roast-
ing. Iraqi Bull Geol Min. 2006;2(2):103–13.
8. Kerr P, Liu Q, Etsell TH. A study of the reaction mechanisms of the iron segregation process,
IMPC2016, Quebec City, QC. Paper No. 185.519-26; 2016.
9. Kerr P, Liu Q, Etsell TH. Beneficiation of the Clear Hills ironstone, COM2015, Toronto, ON. Paper
No. 8936; 2015.
10. Kerr P, Liu Q, Etsell TH. Iron segregation roasting processes. In: Davis B, et al., editors. Extraction
2018. The minerals, metals & materials series. Cham: Springer; 2018. p. 1029–37.
11. Bhattacharya M. Reaction mechanism and thermodynamics of segregation roasting of iron oxide.
Int J Miner Process Extr Metall. 2016;1(5):64–9.
12. Hernandez V, Peake K, Dalvi A, Brown R, Olurin J, O’Farrell T, et al. Process development of a
new DRI technology for oolitic iron ores. In: The iron and steel technology conference and exposi-
tion (AISTech). Pittsburgh, PA; 2013. p. 519–26.
Using Hydrogen as a Reductant in Fire Refining
at Aurubis Hamburg’s “Down-town” Smelter
Torben Edens1 (*) and Johann Steindor1

1
Aurubis AG, Hamburg, Germany
t.edens@aurubis.com
Abstract. The deoxidation of copper or “poling” is the final step in the pyrometallurgical
process of primary and secondary copper extraction. The deoxidation generates CO2
since gaseous and liquid hydrocarbons (natural gas) are the common choice for reducing
agents. Previous energy and cost-saving measures increased the efficiency of the process,
but the inherent production of CO2 cannot be avoided. The use of hydrogen as a reductant
may decarbonize a core process of copper production, which is a desirable target for the
metal industry in general and Aurubis in particular.
The primary smelter in Hamburg runs two anode furnaces with a capacity of 270 t per
batch each. From September to December 2021, the anode furnaces were provisionally
supplied with hydrogen and 14 batches were poled using hydrogen. These experiments
were designed to determine in full-scale operational tests the properties of the poling with
hydrogen in terms of efficiency and process control.
The average efficiency of poling with hydrogen was higher than poling with natural
gas. The adjustment of the volume flows and the temperature of the process were always
controllable. The endpoint of the poling with hydrogen could be determined similarly to
the poling with natural gas based on the temperature curve.
The tests showed the importance of nitrogen addition for stable jetting conditions
where clogging of the tuyeres was a phenomenon observed when pure hydrogen was
used. An important observation from the tests was that the efficiency of the reduction was
not affected by the addition of nitrogen.
Keywords: Hydrogen · Decarbonization · CO2 · Copper · Anode furnace · Fire

refining · Nitrogen
1 Introduction
1.1 Poling
The deoxidation of copper or “poling” is the final step in the pyrometallurgical process
of primary and secondary copper extraction. The deoxidation generates CO2 since gas-
eous and liquid hydrocarbons are the common choice for reducing agents. Aurubis uses
natural gas and liquefied petroleum gas (LPG) as reducing agents in its anode furnaces.
Previous saving measures increased the efficiency of the process, but the inherent pro-
duction of CO2 cannot be avoided.
https://doi.org/10.1007/978-3-031-17425-4_31
212 T. Edens and J. Steindor
1.2 Literature Survey
There is some experimental data and industrial tests that deal with hydrogen poling. Air
Liquide [1, 2] and Linde [3] employed hydrogen for the deoxidation of copper on a
larger scale. Air Liquide used porous plugs in crucibles and tundish. Linde tested poling
with hydrogen in a reverberatory furnace. Unfortunately, the tests were carried out 50
and 20 years ago, respectively. None of the participants were available for questions.
The experimental conditions of the different tests from the literature are listed in
Table 1.
Foulard and Galey [1, 2] gave neither information on the kinetics of the reduction
nor detailed information on the procedure. But the data revealed a high efficiency of the
reduction process (98%). Yamauchi and Sano [4–6] investigated the deoxidation with
hydrogen concerning kinetic, stirring, and crucible surface effects. They found a com-
plicated reaction, which included the mass transfer of hydrogen and oxygen. The deoxi-
dation rate was enhanced by stirring at the free surface as well as at the inert crucible
wall. The porosity of the walls enhanced the nucleation of steam bubbles and therefore
increased the reaction in the melt. Monzen et al. [7] found that the deoxidation is depen-
dent on hydrogen partial pressure but independent of the temperature. Fukunaka et al.
[8] found that the gas film mass transfer controlled the reduction for high oxygen
concentrations, whereas the liquid film mass transfer was limited at low oxygen con-
centrations. The control regime from gas to liquid mass transfer changed at 1000 ppm
oxygen in the melt, which is in good agreement to the results of Riveros et al. [13, 14]
with natural gas. Iwamura et al. [9] deoxidized copper with various gas mixtures con-
taining hydrogen, carbon monoxide, argon, and nitrogen. They found that high hydro-
gen concentration did not accelerate the deoxidation at low oxygen concentrations.
Parra and Capocchi [10] observed gas mass transfer limit for oxygen concentrations
between 4000 and 2000 ppm in copper too. Likewise, Marin [11] validated the gas
phase mass transfer as the limiting step. Furthermore, Marin derived a low activation
energy for the reduction with hydrogen. Klaffenbach et al. [12] measured an efficiency
of about 90% for the reduction.
Table 1. Laboratory and industrial tests on deoxidation with hydrogen.

Author Cu mass Initial oxygen H2 2nd gas Gas flow Flow regime
t % % Nm³/h
[1, 2] 0.02–2.5 1.18 60–100 N2 ? Bubbling, porous plug
[4–6] 0.000135 0.2–0.4 10–50 Ar 0.009–0.03 Crucible, free surface
[7] 0.0005 0.15–0.2 5–100 Ar 0.006–0.03 Bubbling, lance
[8] 0.5⋅10−6 0.036–1.9 0.2–4 Ar 0.72 Levitated droplet
[9] 0.04 0.001 5–50 Ar, N2 ? Bubbling, lance
[10] 0.007 0.4 15, 33 N2 0.012–0.054 Bubbling, lance
[11] 0.14⋅10−6 1 15–100 N2 0.035–0.050 Surface crucible
[12] 0.016 0.59 100 None 0.18 Bubbling, lance
[3] 145 0.9 53–81 N2 200–350 Jetting, lance
This paper 275 0.9–1.2 60–97 N2 1400–2000 Jetting, tuyere
Using Hydrogen as a Reductant in Fire Refining at Aurubis Hamburg’s “Down-town… 213
The tests of Linde with MKM in Hettstedt described by Fasshauer et al. [3] come
closest to the technique used at Aurubis. The size of the batches was reasonably large.
Furthermore, the nozzle diameter as well as the flow regime was equal to the anode
furnaces in Hamburg. Very good efficiencies were achieved. However, the efficiencies
were often above 100% because the burners were operated strongly sub-
stoichiometrically. This and the long reducing time of 6 h made the results uncertain. A
startling finding of Fasshauer et al. was the wear-off of the lances at high hydrogen
concentrations. There is neither a chemical nor an obvious physical effect that can
explain the wear-offs, but the high concentrations reported are interesting since the gas
phase mass transfer is limited and addition of nitrogen means additional technical effort
and cooling of the melt.
Most of the test work does not fit well to the design of the anode furnaces employed
by Aurubis. In most cases, the size and the flow regime were not comparable to the
industrial application with submerged and jetting tuyeres. The only suitable test, on
closer inspection, shows details that do not allow the transfer to our furnaces. Therefore,
it was necessary to perform our own tests to obtain data for a possible new design of the
anode furnaces.
1.3 Chemistry of Gaseous Reduction
The chemical reaction is quite simple:
CH 4 4 O CO2 2H 2 O (1)

where [O] denotes dissolved oxygen. Nevertheless, the reaction is not straightforward;
first, the methane is cracked to hydrogen and soot (2) at high temperature within the
melt:
CH 4 C H 2 (2)
H 2 O H 2 O (3)

C O CO (4)

CO O CO2 (5)

The evolving gases react further, and a gas mixture is formed
C + H 2 O  CO + H 2 (6)
CO2 + H 2  CO + H 2 O (7)
C + CO2  2CO (8)
The hydrogen easily reacts with the dissolved oxygen (3). The soot reacts slower to
carbon monoxide since the liquid–solid reaction is slower by nature (4). Additionally,
the soot may react to carbon monoxide with the water that is produced by the hydrogen
reaction (Eqs. 6 and 7). The carbon monoxide (5) has only half of the reaction rate and
mass transfer of hydrogen [14]. It can therefore be assumed that the hydrogen under-
goes the reduction very quickly and the carbon only comes into play afterwards, which
fits well with the operational experience of the strong soot emissions of poling with
natural gas and other hydrocarbons.
The calculation of the efficiency has to consider the generation of water and carbon
dioxide as final products (1):
N O
CH4 (9)
4 N CH 4

The reduction with hydrogen consists of only one reaction (3). The efficiency is sup-
posed to be higher since hydrogen needs no time for cracking and no parallel reactions
(6 and 7) that consume hydrogen are present. The efficiency of the reduction with
hydrogen is as follows:
N O
H2 (10)
N H 2

2 Experimental
2.1 Safety
The use of hydrogen in a large-scale operation was new for Aurubis Hamburg and
implied many questions. Therefore, HAZID and HAZOP studies were used to identify
the risks. The studies were chaired by an external consultant and attended by smelter
operations, production engineering, R & D, environmental department, fire department,
hydrogen supplier, technical inspection association, and external engineering. The stud-
ies produced several technical safety measures and routines as, risk assessment, operat-
ing instruction, explosion protection document, operating manual, checklists, and
handover rules between operations and R & D.
2.2 Layout
The necessary hydrogen amount was derived from the present natural gas consumption
for poling.
The gas control and mixing station should be able to provide all volume flows and
mixtures between the smallest and largest quantity (0–1000 Nm3/h and tuyere) and the
concentration ranges between 0% and 100% for each gas. The outlet pressure is a func-
tion of the tuyere pressure drop, which is determined by tuyere diameter and flow rate.
The volume flow should be selected so large that the pressure is at least 7 bar for 8 mm
and 5 bar for 16 mm tuyeres, so that the jetting is ensured [15, 16].
2.3 Gas Supply
Since this was a trial with a very short time period, a mobile supply was used. A hydro-
gen high-pressure trailer served as storage location for the required quantity of hydrogen
for one test. Two extraction panels reduced the hydrogen pressure from 200 to 16 bar. A
mobile cold gasifier with a vacuum-insulated storage tank and the required air evapora-
tors served as a storage location for the liquid nitrogen. The evaporation of the liquid
nitrogen generated the pressure of 18 bar. The hydrogen trailer was exchanged after
each test, whereas the liquid nitrogen storage tank was refilled after two or three tests.
2.4 Gas Control and Mixing Station
The gas mixture for the tests was produced in a gas control and mixing station. The sta-
tion received the two gases. The hydrogen as well as the nitrogen line were equipped
with a manual and two automatic shut-off valves. The shut-off valves stopped the gas
flow when the pressure was too high or too low. The gas flow was automatically stopped
when the electric power or control air failed. In addition, the shut-off valves could also
be closed by two emergency-stop switches located in the smelter. The hydrogen line
was equipped with a gas safety train, which automatically controlled the tightness off
the shut-off valves at each start of the hydrogen supply. The control station was planned
under the rules of EN 746-2 [17]. The volume flow and composition of the gas mixture
were manually controlled for each of the two lines. The measured values of pressure,
temperature, and volume flow were transmitted to the process control system. The val-
ues were displayed on the touch panel of the control station too.
Hoses connected the gas trailers, the gas control station, and the piping to the anode
furnace. Hoses with quick connect couplings joined the piping with the tuyeres. The
tuyeres were equipped with a single connector, which secured that either hydrogen or
natural gas was connected to the tuyere.
2.5 Oxygen Measurement
There was some uncertainty about the oxygen measurement. It was known that the
analysis of solidified copper samples is uncertain since cuprous oxide depreciates at
higher concentrations depending on the temperature. Therefore, we [12] investigated
the accuracy of the oxygen measurement. The evaluation of the initial oxygen content
included the comparison of different measurements and sampling methods, respec-
tively. The comparison included simultaneous measurements with vacuum samplers,
dip samplers, and coin samples that are analyzed by AES and an electrochemical sen-
sor. The solidified samples tend to report lower oxygen contents at elevated concentra-
tions and to report higher oxygen contents at low concentrations compared to the
electrochemical sensor.
2.6 Performed Tests
The tests were conducted in normal operation. All batches could be completed without
major disturbances to the smelter schedule. However, the prolonged poling time of 190
instead 150 minutes puts pressure on the schedule. To convert from natural gas to
hydrogen, the supply hoses to the nozzles had to be changed and a leakage test carried
out. The procedure took about 20 minutes. The average batch sizes for tests and stan-
dard operation (September 16 to December 14) were 269 and 274 t per batch, respec-
tively. In total, 14 tests were performed (Table 2). The mixtures covered the concentration
range from 60% to 100% hydrogen. The gas mixture was altered in the second half of
four tests to increase the temperature of the melt. The tuyere diameter was altered
between 8 and 16 mm. Test 11 was scheduled during a converter duct cleaning. The
cleaning allowed to operate with reduced flow rate as more time is available between
standard operation batches. This test should clarify if the reduction is depended on the
volume flow. Unfortunately, other repairs postponed the test start. Nevertheless, the first
half of the poling stage was completed with hydrogen.
3 Results
For comparison, the specific depletion of oxygen in the melt and the specific consump-
tion of natural gas or hydrogen per batch were presented. This is necessary as each
batch had a different initial oxygen contents, sizes, and flow rates which would compli-
cate the evaluation.
Table 2. Performed tests. During four of the experiments, the hydrogen concentration of the gas
mixture was changed in the 2nd half to control the temperature of the copper.
Test H2/N2 Cu Tuy. Tstart Tend [O]start [O]end η Comment
%/% t mm °C °C g/t g/t %
1 70/30 269 16 1182 1239 >10,041 160 94
2 70/30 (90/10) 245 8 1160 1258 9835 1393 62 Burner λ > 1
3 70/30 255 8 1172 1256 >10,041 101 83
(85/15)
4 85/15 308 11 1190 1245 8741 92 75 Burner clogged
(100/0)
5 85/15 275 8 1191 1269 8571 609 68 Tuy. fails
6 85/15 260 11 1156 1248 9552 277 97
7 75/25 275 11 1230 1269 8987 59 90
(70/30)
8 80/20 240 11 1187 1264 11,340 656 78
9 85/15 291 11 1161 1243 8707 638 91
10 100/0 239 11 1181 1263 10,011 1022 83 Clogging of tuyeres
11 60/40 275 8 1209 1239 10,890 5115 93 ½ flow rate
12 60/40 273 16 1178 1230 9125 627 94
13 60/40 285 16 1183 1235 8133 444 92
14 70/30 272 11 1207 1248 9089 160 94
During the tests, the oxygen content of the liquid copper was sampled with an elec-
trochemical probe every 45–60 minutes. Therefore, each test produced 4–5 data points
from high to low oxygen levels. The sequence of data points provides an overview of
the course of the reduction. The oxygen depletion follows a linear trend until the poling
is terminated. This is shown in Figs. 1 and 2 for natural gas and hydrogen, respectively.
Figure 2 includes all tests of Table 2 from 60% to 100% hydrogen (the natural gas was
always undiluted.) The linear trend for all tests is an indication that the reduction is
independent of concentration (cf. Figures 1, 2, and 3) or nozzle diameter.
3.1 Test Runs
Test 1 finished the batch in time (156 min) with 94% efficiency. Unfortunately, exces-
sive splashing and vibrations prevent any further tests in this load range (2000 Nm3/h
gas mixture). The processing time would have been even shorter if the initial volume
flow had been maintained. However, even the 6% lowered flow rate caused vibrations
that were too severe. For the following tests, a maximum flow rate of 1500 Nm3/h gas
mixture was the limit.
Tests 2 and 5 were excluded from the reduction efficiency evaluation since the test
conditions were unregular and the reduction process was disturbed. Nevertheless, the
discarded tests were interesting. Particularly, test 2 drew an interesting finding.
Commonly, the air ratio λ of the burner is not a critical issue. Operators frequently
Fig. 1. Specific oxygen depletion and natural gas consumption of 14 batches that were poled with
CH4, the blue points show the individual measurements, the blue solid line indicates the average refer-
ring to 69% efficiency (9), the dotted lines referring to 60% and 85% efficiency.
Fig. 2. Specific oxygen depletion and hydrogen consumption of 14 batches that were poled with H2,
the blue points show the individual measurements, the blue solid line indicates the average referring to
90% efficiency (10), the dotted lines referring to 75% and 100% efficiency.
Fig. 3. Efficiency of the reduction with hydrogen and hydrogen concentration in the gas mixture. The
figures at the left side of the points indicate the number of measurements that were used for the calcula-
tion of the average.
chose higher air ratios when the blister copper is too cold. During the 2nd test, the
burner was therefore set very lean, which resulted in a poor reduction. Therefore, a
stoichiometric burner setting was determined for the following tests. Test 5 is difficult
to interpret because tuyere malfunctions and clogging occurred together.
The finding of test 4 is more obvious: The high bath level of large batches promotes
the clogging of the burner and exhaust.
In normal operation, the rise of the temperature indicates the end point of the reduc-
tion with natural gas. The temperature rise corresponds with 330 ppm oxygen in copper.
Most of the hydrogen tests were terminated when a sufficiently low oxygen concentra-
tion was reached in order to maintain the smelter schedule. The tests 1, 3, 4, 6, and 14
were terminated when the temperature in the after burning chamber started to rise. The
final oxygen content was 160 ppm oxygen when poling with hydrogen, which is lower.
Test 10 produced a series of tuyere clogging events when poling with pure
hydrogen.
Test 11 employed only a halved volume flow but achieved a satisfying efficiency.
3.2 Efficiency
As described above, Figs. 1 and 2 show the decrease in oxygen content in copper during
poling with natural gas and with hydrogen, respectively. The chemical efficiency of
natural gas is between 55% and 85%. When averaged over all the tests, an efficiency of
69% is obtained. In the case of poling with hydrogen, an average efficiency of 90% was
determined. The values of all tests vary between 75% and 100% when the data are dis-
carded with faulty conditions, such as burner and nozzle malfunctions or incorrect
burner settings.
Figure 3 shows the average efficiency of the reduction with hydrogen for different
concentrations of hydrogen in the gas mixture. The points show the average efficiency
of all measurements at the different concentrations. Again, measurements with tuyere
malfunctions or odd operating conditions were discarded. The number of measure-
ments is displayed on the left side of the points. The efficiency is independent of N2
dilution in the tested concentration range between 60 and 85% nitrogen. Low efficien-
cies above 90% H2 in gas are caused by unstable operation, e.g., clogging and leakages
that resulted in low number of valid measurements. But one can assume that the effi-
ciency is also higher in undisturbed operation.
3.3 Tuyere Clogging
Figure 4 gives insights into the clogging of a tuyere. The trends show how the tuyere
was clogged by a stepwise blocking of the tuyere. These events frequently occurred in
test 10 and made it almost impossible to complete the test regularly. The blue arrows
indicate single clogging events resulting in higher pressure and lower flow. The green
arrows indicate manual control overrides that temporarily increased the flow rate. The
clogging progresses unstoppably even though the jet is under-expanded and the volume
flow is high.
Fig. 4. Section of flow rate and pressure curve during test 10. Blue line is the H2 flow rate and the grey
line is the pressure. Blue arrows: clogging. Green arrows: manual override.
3.4 Temperature Rise
The blister copper loses temperature during ladle transport, slag skimming, and desul-
furization. During poling, the temperature rises, which is a wanted effect. The tempera-
ture increase during the poling process ensures the correct casting temperature since the
flow of copper via the launders to the casting wheel is also associated with temperature
losses.
Figure 5 shows the heat of reaction of the reduction of cuprous oxide at 1200 °C in
consideration of the heat up of hydrogen and nitrogen. The convective and radiative heat
losses were neglected. At 45% hydrogen, no heat is released by the reaction and external
heat has to be supplied by the burner. Figure 6 shows the temperature increase of the
copper during poling for different hydrogen concentrations. The data are not on a straight
line since additional heating of the burner provides an additional rise in temperature and
some repairs and idle times lowered the temperature. However, the nitrogen concentra-
tion of the gas mixture may be used for proper adjustment of the temperature.
3.5 Anode Quality
The surface of the anodes had no blisters or bumps, and no additional rejected anodes
were recorded during the tests with hydrogen compared to standard operation. Blisters
and bumps are caused by dissolved gases (hydrogen and sulfur dioxide) when the cop-
per solidifies in the anode molds. If too much hydrogen is dissolved in liquid copper,
the anode will have blisters.
Fig. 5. Heat of reaction of 1 Nm3 of a H2/N2 gas mixture. Data were obtained from HSC10 for the
reaction of Cu2O at 1200 °C and the heating of the gases from 10 to 1200 °C.
Fig. 6. Temperature rises during poling. The increase was influenced by the heat of reaction, the heat-
ing burner, and idle times.
4 Discussion
4.1 Data Scattering
The maximal efficiency of 100% for hydrogen or 85% for natural gas is often but not
always achieved (cf. Figures 1 and 2). The oxygen measurement in the Cu melt by an
electrochemical probe shows a deviation of up to 8% in subsequent measurements. If
copper slag is not completely skimmed, the residual cuprous oxide will consume addi-
tional hydrogen. Five tons of slag may contain 280 kg of oxygen that will alter the
efficiency from 100% to 90%. If the furnace retains some copper from previous batches,
the amount of blister will be underrated and therefore the amount of oxygen, too. A
retained mass of 15 t will drop the efficiency by 5% if the initial oxygen content is
10,000 ppm. The measurement of the volume flow had been confirmed to be less than
3% by comparison of trailer pressure and flowmeter. It is still not clear if this error is on
the side of the pressure gauge of the trailer or the flowmeter.
4.2 Process Time for Reduction
One of the test batches (test 1) conducted with a high flow rate of hydrogen was deoxi-
dized in 156 minutes, a comparable time to standard operation. These high flow rates of
1000 Nm3/h per tuyere could not be maintained, let alone increased because the furnace
vibrated (rolling around the longitudinal axis) too much and the splashing clogged
burner and exhaust. The present flow rate is 300 Nm3/h per tuyere with natural gas. The
splashing and vibration made it necessary to reduce the flow rate by 25%. Therefore,
the poling with hydrogen lasted on average 40 minutes longer than the usual 150 min-
utes per batch recorded in normal operation. So, the speed of reduction is dependent on
the flow rate of hydrogen.
4.3 Chemical Efficiency
The tests showed that the efficiency is independent of hydrogen concentration, volume
flow, and tuyere diameter. The present literature, which dealt with comparable small
amounts of copper and often with low hydrogen or oxygen concentrations, described a
mass transfer limit of the reduction.
In industrial operations, the mass transfer limit does not seem to slow down the pro-
cess. Furthermore, higher stirring rates or flow rates by additional nitrogen produced
not a higher efficiency (cf. tests 9 and 13). Despite only half the volume flow, test 11
achieved just as good a result as test 1 in terms of efficiency. The tuyere diameter is
naturally important for the necessary head pressure of the tuyere to ensure jetting con-
ditions and the wanted flow rate. But no influence on the efficiency could be observed.
The reduction is probably more dependent on the mass transfer in the plume and not on
the short jet that breaks down bubbles instantly.
4.4 Clogging of Tuyeres
In contrast to Fasshauer et al. [3], we found no burned-off tuyeres but clogging when
poling with hydrogen. Overheating is unlikely. Under jetting conditions, the cooling of
the lance with hydrogen is higher than that with nitrogen. Heat capacity by volume is
more or less the same for hydrogen and nitrogen but the speed of sound of hydrogen is
approximately four times higher. Therefore, hydrogen cools more than nitrogen. The
crucial injection of jets with high hydrogen concentrations originates in the insufficient
jetting of gases with low densities. Not even a highly under-expanded jet (e.g., M′ > 2)
that accelerates at the tuyere tip [18, 19] might be enough. This effect was described by
Ozawa and Mori [15]. They found that the density ratio of gas and liquid has to be con-
sidered for jetting conditions:
g
M 0.009 (11)
g

where M′ is the nominal Mach number at standard conditions and ρg and ρl are the gas
and liquid densities. Interestingly, natural gas complies with the criterion for jetting in
liquid copper with 0.011, whereas hydrogen fails with 0.004. The addition of nitrogen
to hydrogen increases the density of the gas and for a gas mixture with 70% hydrogen
and 30% nitrogen, the value is 0.009.
The modified Froude number Fr′ is another way of describing an unexpanded sonic
jet [18]:
g u2
Fr (12)
g l g d

where u is the gas velocity at the tip of the tuyere and d is the tuyere diameter, u exceeds
the real speed of sound and corresponds with the nominal Mach number M′.
Figure 7 depicts the flow regimes of submerged jets [18]. The transition regime
divides the bubbling and the steady jetting regime. The transition regime is broader for
low-density ratios, which corresponds with the findings of Ozawa and Mori [15]. For
comparison, Fig. 7 shows the operating range of steel converters (basic oxygen process)
and Peirce–Smith converters. The solid square indicates the operating condition of the
anode furnace with natural gas poling. The circles show the conditions for hydrogen–
nitrogen mixtures with 100% and 70% hydrogen, respectively.
The application of the Ozawa–Mori criterion and the modified Froude number Fr′
explains the complicated run of test 10 with 100% hydrogen. A lot of clogging events
slowed the batch down and the inserts and the tuyeres had to be exchanged. Pure hydro-
gen jets are in the transition between bubbling and jetting regime (cf. Figure 7).
Although test 4 used 100% hydrogen at the end of the run, no clogging occurred. This
is a hint that bigger tuyeres ensure better jetting conditions as described by Kapusta
[16]. May be test 5 had a disadvantageous combination of a small tuyere and an insuf-
ficient density ratio, besides a malfunction of the tuyere pipes.
Fig. 7. Jet behavior diagram – Flow regime boundaries and Q-BOP are extracted from Hoefele and
Brimacombe [18]. AF-NG refers to the anode furnace with natural gas. AF-H2 and AF-H2N2 refer to
poling with pure hydrogen and a 70/30 H2/N2 mixture. PSC is the range of Peirce–Smith converters
from matte to copper. Q-BOP refers to bottom-blowing steel converters.
The criterion explains too why Fasshauer et al. [3] observed the wear-off of lances.
When the lance was clogged and the gas flow stopped, the lance heated up and melted
without the cooling gas flow.
Figure 8 shows the minimum inlet pressure for a given diameter that is necessary for
stable jetting. The lines of equal Mach number were calculated from [12, 16] according
to the relations with the Fanno flow. The solid line indicates the recommended Mach
number of 1.25 [16]. The dashed line shows the Mach number of 1. The standard
operation with natural gas is indicated by a solid square. The hollow dots show the
parameters, which were used during the hydrogen tests. The different pressures result
from varying flow rates. Apart from pure hydrogen, the computation of the minimum
inlet pressure was very helpful for the choice of tuyere diameters and minimum inlet
pressures for stable jetting operation.
4.5 Air Ratio
Test 2 revealed an operational difference between poling with hydrogen or natural gas.
In present operation, the air ratio of the burner is set up to two when the batch is too
cold. The excess air of the burner burns the unreacted natural gas products in the free-
board of the furnace. The incomplete conversion and therefore lower efficiency of the
natural gas generates sufficient amounts of soot and carbon monoxide. When we tried
the same in test 2, the copper was re-oxidized since the higher efficiency of the
Fig. 8. Pressure vs. tuyere diameter. The solid line for Mach 1.25 indicates the onset of stable jetting
[16]. The dots show the parameters of the hydrogen tests and the operation with natural gas.
hydrogen released no fuel for the post-combustion in the freeboard. Therefore, the air
ratio was fixed to one for all following tests.
This may also explain the below-average efficiency of test 4. Due to the bath level,
the burner clogged very quickly and could no longer contribute to a neutral furnace
atmosphere in the course of the test. The ingress air entering through the mouth oxi-
dized the copper on the bath surface.
5 Conclusions
The test campaign successfully demonstrated the poling of blister copper on an indus-
trial scale with real production conditions. The upstream Peirce–Smith converters pro-
duced blister copper in the usual batch size, composition, temperature, and oxygen
content. The chemical efficiency of hydrogen is close to 100%, which makes a neutral
furnace atmosphere mandatory (burner air ratio λ = 1).
Jetting properties of hydrogen are poor. For jetting with pure hydrogen in liquid cop-
per, high pressures are necessary to achieve high nominal Mach numbers. But high
temperatures and dusty environments are not favorable for the sealing of high-pressure
tuyeres. Therefore, nitrogen is a good additive to grant stable jetting. Strong dilution is
not necessary, since stable conditions are already achieved from 70% hydrogen. A higher
dilution would further not only improve the jetting behavior but also cool the bath more.
The present furnace design limits the flow rate by splashing and vibrations. While
vibrations could be mitigated by an improved gear, splashing obstructs the operation by
required exhaust and burner cleaning. The furnace needs a new layout that is not so
sensitive to splashing. To avoid excessive splashing, gas should be distributed to more
than two tuyeres. Exhaust and burner should be arranged so that they are not in the area
of influence of the ejection.
Hydrogen can be used as reductant in fire refining with no metallurgical drawbacks
or unmitigable safety concerns. After the trend to higher oxygen enrichments in 1990s,
which replaced preheating of the flash smelter blast, hydrogen will decarbonize the last
stage of pyrometallurgical copper extraction.
Acknowledgments Many people have contributed to the present paper. We would like to thank
Christian Mäther who cared for technical equipment and the safety audits in a very professional and
sound way.
We enjoyed the hospitality and skilled support of the operators Ibrahim Altincinar, Durak Anil,
Kasim Aydin, Christoph Baluniak, Mario Block, Tuncay Ciftci, Jens Horn, Achim Johns, Emrah
Kalankaldi, Dilaver Klamann, Kai Krüger, Ralf Leuchtenberger, Oliver Meins, Alexander Merski,
Bernd Storm, Alexander Wlassow, and Erdel Yavuz.
Torben is very grateful for the tireless and reliable support of the intern Dominik Rundshagen. He
was a great help during performance and evaluation of tests and put the right questions.
Thanks also to Phillip Mackey who showed me the way to the Froude numbers.
The City of Hamburg contracted Aurubis to perform the tests.
Notation
d tuyere diameter mm
g gravitational constant m/s2
M Mach number
M´ nominal Mach number at standard conditions
M molar mass mol
m mass kg or t
P pressure Pa
R gas constant J/mol K
Ri gas constant of i J/kg K
T temperature K or °C
t time minutes
u velocity m/s
V volume Nm3 aka standard cubic meter aka scm at P = 101,325 Pa and
T = 273 K
Subscripts:
exp expanded conditions, does not reflect the true conditions, but a
hypothetical situation where the gas jet can expand freely. It
therefore indicates how strongly the jet is under-expanded
g gas
inlet conditions at tuyere inlet
l liquid or molten metal
outlet conditions at tuyere inlet
Greek
η efficiency %
λ air ratio indicates lean, rich, or stoichiometric burner operation
ρ density kg/m3
Abbreviations
AES Atomic emission spectroscopy

AF Anode furnace for the fire refining of blister copper
NG Natural gas, mostly methane
PSC Peirce–Smith converter
Q-BOP Bottom-blown basic oxygen process – Steel converter
References
1. Foulard J, Galey J. Procédé pour la désoxydation et le dégazage du cuivre et de ses alliages. French
Patent FR1.552.137. Paris: Ministère de l’industrie, Service de la Propriété Industrielle; 1969.
2. Foulard J, Galey J. Désoxydation et dégaze du cuivre par insufflations d’hydrogène et d’azote. In:
Deuxième congrès international l’hydrogène dans les métaux. Pergamon Press; 1977.
3. Fasshauer K, Steffner F, Dauterstedt HJ, Albrecht M, Wernicke E. Verfahren zum Polen von
Kupfer mit Wasserstoff und Stickstoff. European Patent EP 0 992 597 B1. Munich: European
Patent Office; 1999.
4. Yamauchi C, Sano K. Kinetic studies on the deoxidation of molten copper by hydrogen
gas (in Japanese). J Jpn Inst Metals Mater. 1969;33(11):1249–55. https://doi.org/10.2320/
jinstmet1952.33.11_1249.
5. Yamauchi C, Sano K. Effect of stirring on the velocity of the deoxidation reaction of molten cop-
per by hydrogen gas (in Japanese). J Jpn Inst Metals Mater. 1969;33(11):1255–61. https://doi.
org/10.2320/jinstmet1952.33.11_1255.
6. Yamauchi C, Sano K. Effect of materials of crucibles on the velocity of the deoxidation reaction
of molten copper by hydrogen gas (in Japanese). J Jpn Inst Metals Mater. 1969;33(11):1262–7.
https://doi.org/10.2320/jinstmet1952.33.11_1262.
7. Monzen K, Yaoshima M, Nitta K, Kōzuka Z. Deoxidation mechanism of liquid copper by hydro-
gen (in Japanese). J Min Metall Inst Jpn. 1971;87(1006):1059–64. https://doi.org/10.2473/
shigentosozai1953.87.1006_1059.
8. Fukunaka Y, Tamura K, Taguchi N, Asaki Z. Deoxidation rate of copper droplet levitated in Ar-H2
gas stream. Metal Mater Trans B. 1991;22:631–9. https://doi.org/10.1007/BF02679018.
9. Iwamura T, Koya T, Sukekawa I, Hagiwara H, Asao H, Shigematsu T, Sato H. Method for manu-
facturing oxygen-free copper. US Patent 5,037,471. 1991.
10. Parra De Lazzari C, Capocchi JDT. Copper deoxidation by bubbling hydrogen/nitrogen mixtures
through the melt. In: EPD Congress 1997: proceedings of sessions and symposia sponsored by the
Extraction and Processing Division, Orlando, Florida, USA, 9–13 February 1997. The Minerals,
Metals & Materials Society; 1997. p. 107–18.
11. Marin T. Rate of reduction of liquid copper with H2(g)/N2(g) mixtures. Revista Minerales Edición
264 Instituto de Ingenieros de Minas de Chile. 2008. p. 39–46.
12. Klaffenbach E, Winkler S, Edens T. Usage of hydrogen in Pyrometallurgical copper refining pro-
cess. 80th Meeting of GDMB Copper Committee. 2021.
13. Riveros G, Warczok A, Voisin L, Marin T. Factors affecting the rate of copper reduction during
copper refining. In: International symposium, metallurgical and materials processing: principles
and technologies. 2003. p. 431–46.
14. Diaz L, Reyes F, Ingalls-Cruz A, Mendez C. Mass transfer model for the de-oxidation of molten
copper. JOM. 2017;69:980–6. https://doi.org/10.1007/s11837-017-2356-0.
15. Ozawa Y, Mori K. Effect of physical properties of gas and liquid on bubbling-jetting phenomena
in gas injection into liquid. Trans Iron Steel Inst Jpn. 1986;26(4):291–7. https://doi.org/10.2355/
isijinternational1966.26.291.
16. Kapusta JPT. Submerged gas jet penetration: a study of bubbling versus jetting and side ver-
sus bottom blowing in copper bath smelting. JOM. 2017;69:970–9. https://doi.org/10.1007/
s11837-017-2336-4.
17. DIN Deutsches Institut für Normung. EN 746-2 Industrielle Thermoprozessanlagen — Teil 2:
Sicherheitsanforderungen an Feuerungen und Brennstoffführungssysteme. 2010.
18. Hoefele EO, Brimacombe JK. Flow regimes in submerged gas injection. Metall Mater Trans
B. 1979;10:631–48. https://doi.org/10.1007/BF02662566.
19. Zucker RD, Biblarz O. Fundamentals of gas dynamics. Hoboken: John Wiley & Sons; 2002.
p. 241–61.
Part III Electrochemical Degradation
of Multi-Component Materials
Corrosion of Mg Alloy AM60B Micro-Alloyed with As
Joey Kish1 (*), Dong Fu1, Beth McNally1, Darren Feenstra1, Amjad Javaid2,
and Kumar Sadayappan2
1
McMaster University, Hamilton, ON, Canada
kishjr@mcmaster.ca
2
CanmetMATERIALS, Hamilton, ON, Canada
Abstract. The effect of micro-alloying AM60B (Mg-6Al-0.3Mn) with As (well-known

cathodic hydrogen gas (H2) evolution poison) on the reducing corrosion in 3.5 wt.% NaCl
(aq) was determined. Custom permanent mold casting AM60B material, with and with-
out, 0.23 wt.% micro-alloyed As was made for this purpose. Microstructure characteriza-
tion was carried out using light optical microscopy and scanning electron microscopy
with associated X-ray energy-dispersive spectroscopy. Corrosion rates were measured
using mass loss and volumetric hydrogen gas evolution measurements. Potentiodynamic
polarization measurements were made to reveal the global anode and cathode kinetics.
Micro-alloying with As is somewhat beneficial in reducing corrosion, but not to the extent
reported for unalloyed Mg. More importantly, corrosion involves toxic arsine gas (AsH3)
evolution, in addition to H2, consistent with thermodynamic predictions. Alternative
micro-alloying routes are proposed for improved corrosion control via micro-alloying
with non-toxic cathode poisons.
Keywords: Magnesium · Alloying · Corrosion
1 Introduction
Automotive applications of Mg alloy die-castings (powertrain, structural, and bracket)

have shown a growing trend over the past two decades [1]. Common alloys used to
manufacture die-casting for structural applications include AM50 (Mg-5Al-0.3Mn) and
AM60 (Mg-6Al-0.3Mn), which exhibit sufficient strength and ductility required for
crashworthiness (energy absorption) [2]. While Mg alloys provide substantial weight
reductions compared to steel, structural Mg alloy component use has been hindered by
the poor corrosion resistance (especially to galvanic corrosion) when exposed to an
aqueous salt solution [3].
Magnesium (Mg) corrosion in aqueous electrolytes involves the conversion of Mg to
Mg(OH)2 according to the simplified reaction: Mg + 2H2O → Mg(OH)2 + H2. Elementary
steps of this process include the following: (i) anodic dissolution of Mg to Mg2+, (ii)
cathodic reduction of H2O to H2, and (iii) precipitation of Mg(OH)2 from a super-
saturated, locally alkaline solution near the cathode sites [3]. Corrosion is “activated” at
the propagating front for reasons still being debated in the literature [4]. Intrinsic con-
trol of the cathode reaction (H2 evolution) rate by chemical means has been demon-
https://doi.org/10.1007/978-3-031-17425-4_32
232 J. Kish et al.
strated for Mg, with arsenic (As) being particularly effective as hydrogen (H)
recombination poison [5, 6]. A significant reduction in corrosion of Mg was achieved
by micro-alloying with sparingly soluble As (~0.37 wt.%) in spite of the formation of a
secondary Mg3As2 intermetallic phase [6]. One objective (technical) of this work was
to determine if micro-alloying As has a similar beneficial effect on the reducing corro-
sion of AM60B (Mg-6Al-0.3Mn) when immersed in NaCl (aq). Another objective
(health and safety) of this work was to determine whether corrosion is accompanied by
the release of arsine gas (AsH3). Placing the corrosion potentials of published Mg-As
alloys onto the Pourbaix diagram for As-H2O, such release can be expected.
2 Experimental
Custom permanent mold casting AM60B material, with and without, 0.23 wt.% micro-
alloyed As was prepared at CanmetMATERIALS. Table 1 lists the chemical composi-
tion. The elevated Al content in the AM60B + As material is a consequence of using a
master Al-As alloy to produce the material. Microstructure characterization was carried
out using light optical microscopy and scanning electron microscopy (SEM) equipped
with X-ray energy-dispersive spectroscopy (EDS). Corrosion rates were measured
using mass loss and volumetric hydrogen gas (H2) measurements over 24 h immersion.
A commercial gas detector tube system (Gastec) was used to confirm whether AsH3 gas
was released during corrosion of AM60B + As after by sampling the gas evolved after
96 h immersion. Potentiodynamic polarization measurements were made after 10 min
of conditioning at the open circuit potential, to reveal global anode and cathode kinet-
ics. In both cases, the cast material was tested at least three times to ensure reproduc-
ibility. The reported error bars represent a 95% confidence interval for the replicated
dataset. All measurements were made using a 3.5 wt.% NaCl (aq) electrolyte.
3 Results
The microstructure of the cast material AM60B consists of primary α-Mg grains with
discontinuous distribution of eutectic β-Mg17Al12 + Al-enriched α-Mg phase, as well as
a random distribution of micro-sized porosity (Fig. 1a.) Not shown is the expected ran-
dom distribution of smaller AlxMny intermetallic particles [7]. As Fig. 1b shows, micro-
alloying with As at 0.23 wt.% does not alter these major features. However, it does
introduce additional Si-As intermetallic particles on the micro-scale, as revealed by the
Table 1. Chemical composition of die-cast AM60B material.

Material Al As Mn Zn Si Cu Fe Ni
AM60B 5.7 0.002 0.34 0.001 0.017 <0.001 <0.001 0.002
AM60B + As 7.0 0.23 0.28 0.005 0.034 0.002 <0.001 0.002
Corrosion of Mg Alloy AM60B Micro-Alloyed with As 233
Fig. 1. Microstructure of (a) AM60B and (b) AM60B + As. Higher magnification view of
(c) second phase particle in AM60B + As and (d) asscoaited set of elemental EDS maps
SEM image and associated EDS maps shown in Fig. 1c, d, respectively. No evidence of
Mg3As2 intermetallic particles was found.
Figure 2a compares the volumetric H2 gas evolution with time over the 24 h immer-
sion period for both cast materials. AM60B + As shows somewhat higher average
values than AM60B at the beginning of immersion, yet somewhat lower average val-
ues toward the end of immersion. However, the kinetics are judged to be indifferent
when considering the experimental error. The observation of lower corrosion kinetics
exhibited by AM60B + As after the 24 h immersion period is corroborated by the mass
loss values, as compared in Fig. 2b. Here the difference is judged to be significant
when considering the experimental error. Although not shown, the gas evolved after
96 h immersion contained at least 10 ppm AsH3, as revealed by the calibrated gas
detection tube.
Figure 3 compares the replicate sets of potentiodynamic polarization response of the
two materials. Both materials exhibit a similar global anode and cathode polarization
response, with the anode (metal dissolution) kinetics showing lower polarizability than
the cathode (H2 evolution) at low overpotentials relative to the corrosion potential. Both
sets of curves a rather reproducible. When compared against each other, the replicate
set of AM60B + As measurements shows reduced anode and cathode kinetics, which is
consistent with a somewhat more positive corrosion potential.
234 J. Kish et al.
Fig. 2. Corrosion comparison in 0.1 M NaCl (aq): (a) volumetric H2 evolution and (b) mass loss
(measured after 96 h immersion).
Fig. 3. Replicate potentiodynamic polarization measurements in 3.5 wt.% NaCl (aq).
4 Discussion
Micro-alloying Mg with As is believed to stifle H2 evolution by poisoning the principal

cathodic sites. The proposed mechanism involves As(0) surface deposits or ad-atoms
effectively blocking H atom recombination on the Fe-rich impurity particle cathodic
sites [6]. The beneficial effect is manifested by a significant reduction in cathode kinet-
ics, which in turn yields a significantly more negative corrosion potential. Such features
Corrosion of Mg Alloy AM60B Micro-Alloyed with As 235
are not observed when comparing the polarization response of AM60B + As relative to
AM60B (Fig. 3). Thus, micro-alloying AM60B with As does not reduce corrosion to
the same extent as it does with unalloyed Mg. There are two major differences in the
microstructure between unalloyed Mg and AM60B that may account for the stifled
effect. One is the principle cathodic site. Such sites in AM60B include both β-Mg17Al12
and AlMn intermetallic secondary phase particles: the latter being the more effective
cathode [8–10]. Another is the As-containing secondary intermetallic phase that forms.
An As-Si intermetallic phase forms in AM60B, whereas an Mg-As intermetallic phase
forms in unalloyed Mg. The controlling factor is currently unclear. Even if micro-
alloying AM60B with As was effective in reducing corrosion, this approach would not
be practical given the evolution of the toxic AsH3 gas as a corrosion product. Fortunately,
the beneficial effect on reducing global cathode (H2 evolution) kinetics is likely not
limited to As, as other elements possess the key characteristic of a low M-H (metal-
hydrogen) bond strength, which translates into a low exchange current density (io) for
H2 evolution [11]. Elements such as Ge, In, Sb, and Sc have been shown to be effective
cathode deactivators when added to the Mg matrix [12, 13].
5 Conclusions
(i) Micro-alloying AM60B with As (well-known cathodic H2 evolution poison) does

not reduce corrosion to the same extent as it does with unalloyed Mg. Reasons for
this are likely the difference in principle cathodic site and/or the composition of the
secondary phase intermetallic phase that forms.
(ii) Even if micro-alloying with As was effective in reducing corrosion of AM60B, this
approach would not be practical given the evolution of toxic AsH3 gas as a corro-
sion product. Other non-toxic gas-forming elements such as In, Sb, and Sc may be
more attractive cathode poisons when added to the Mg matrix.
Acknowledgments Financial support was provided by the Natural Science and Engineering Research
Council of Canada (NSERC) in the form of an Engage grant with Alliance Magnesium as the industry
partner.
References
1. Abbott TB. Magnesium: industrial and research developments over the last 15 years. Corrosion.
2015;71:120–7. https://doi.org/10.5006/1474.
2. Luo AA. Magnesium: current and potential automotive applications. JOM. 2002;54:42–8. https://
doi.org/10.1007/BF02701073.
3. Esmaily M, Svensson JE, Fajardo S, Birbilis N, Frankel GS, Virtanen S, Arrabal R, Thomas
S, Johansson LG. Fundamentals and advances in magnesium alloy corrosion. Prog Mater Sci.
2017;89:92–193. https://doi.org/10.1016/j.pmatsci.2017.04.011.
4. Thomas S, Medhekar NV, Frankel GS, Birbilis N. Corrosion mechanism and hydrogen evo-
lution on Mg. Curr Opin Solid State Mater Sci. 2015;19:85–94. https://doi.org/10.1016/j.
cossms.2014.09.005.
236 J. Kish et al.
5. Eaves D, Williams G, McMurray HN. Inhibition of self-corrosion in magnesium by poison-

ing hydrogen recombination on iron impurities. Electrochim Acta. 2012;79:1–7. https://doi.
org/10.1016/j.electacta.2012.05.148.
6. Birbilis N, Williams G, Gusieva K, Samaniego A, Gibson MA, McMurray HN. Poisoning the
corrosion of magnesium. Electrochem Commun. 2013;34:295–8. https://doi.org/10.1016/j.
elecom.2013.07.021.
7. Barbagallo S, Laukli HI, Lohne O, Cerri E. Divorced eutectic in a HPDC magnesium–aluminum
alloy. J Alloys Compd. 2004;378:226–32. https://doi.org/10.1016/j.jallcom.2003.11.174.
8. Jönsson M, Thierry D, LeBozec N. The influence of microstructure on the corrosion behaviour of
AZ91D studied by scanning Kelvin probe force microscopy and scanning kelvin probe. Corros Sci.
2006;48:1193–208. https://doi.org/10.1016/j.corsci.2005.05.008.
9. Andreatta F, Apachitei I, Kodentsov AA, Dzwonczyk J, Duszczyk J. Volta potential of second
phase particles in extruded AZ80 magnesium alloy. Electrochim Acta. 2006;51:3551–7. https://
doi.org/10.1016/j.electacta.2005.10.010.
10. Danaie M, Asmussen RM, Jakupi P, Shoesmith DW, Botton GA. The cathodic behaviour of AlMn
precipitates during atmospheric and saline aqueous corrosion of a sand-cast AM50 alloy. Corros
Sci. 2014;83:299–309. https://doi.org/10.1016/j.corsci.2014.02.030.
11. Trasatti S. Electrocatalysis of hydrogen evolution: progress in cathode activation. In:
Gerischer H, Tobias C, editors. Advances in electrochemical science & engineering, vol. 2.
New York: VCH Publishers Inc.; 1992. p. 1–86. https://onlinelibrary.wiley.com/doi/pdf/10.100
2/9783527616763#page=8.
12. Gusieva K, Davies CHJ, Scully JR, Birbilis N. Corrosion of magnesium alloys: the role of alloying.
Int Mater Rev. 2015;60:169–94. https://doi.org/10.1179/1743280414Y.0000000046.
13. Gusieva K, Birbilis N, Hammond VH, Labukas JP, Placzankis BE. The influence of novel alloy-
ing additions on the performance of magnesium alloy AZ31B. Report No. ARL-MR-856, Army
Research Lab Aberdeen Proving Ground MD Weapons and Materials Research Directorate; 2013.
Studying Stress Corrosion Cracking Mechanisms
Using Novel Nanoscale Characterization
S. Y. Persaud1 (*)
1
Queen’s University, Kingston, ON, Canada
suraj.persaud@queensu.ca
Abstract. Stress corrosion cracking (SCC) is a concern for structural materials across
many industries. SCC requires the synergistic contribution of a susceptible material, cor-
rosive environment, and stress. There have been many proposals for viable mechanisms
to describe SCC; however, due to the complexity of the phenomenon, universal mecha-
nisms are not possible, and identifying and applying a particular SCC mechanism in a
known susceptible material-environment system has been challenging. This is due to the
need for understanding the interplay of a multitude of variables contributing to SCC from
the material science, mechanics, and chemistry/corrosion perspectives. In the past decade,
advanced electron microscopy and surface science methods have advanced significantly
to allow for multi-disciplinary nanoscale characterization and site-specific extraction of
regions of interest (e.g., crack tips). These novel characterization techniques have been
applied successfully to better elucidate SCC mechanisms of materials in certain aggres-
sive environments.
Keywords: Stress corrosion cracking · Electron microscopy · Surface science ·

Nuclear materials · Secondary side
1 Introduction
Stress corrosion cracking (SCC) is a concern for structural materials across many
industries. SCC requires the synergistic contribution of a susceptible material, corro-
sive environment, and stress. There have been many proposals for viable mechanisms
to describe SCC; however, due to the complexity of the phenomenon, universal mecha-
nisms are not possible, and identifying and applying a particular SCC mechanism in a
known susceptible material-environment system has been challenging. This is due to
the need for understanding the interplay of a multitude of variables contributing to SCC
from the material science, mechanics, and chemistry/corrosion perspectives. In the past
decade, advanced electron microscopy and surface science methods have advanced sig-
nificantly to allow for multi-disciplinary nanoscale characterization and site-specific
extraction of regions of interest (e.g., crack tips). These novel characterization tech-
niques have been applied successfully to better elucidate SCC mechanisms of materials
in certain aggressive environments.
https://doi.org/10.1007/978-3-031-17425-4_33
238 S. Y. Persaud
SCC of Ni- and Fe-based alloys in the high-temperature (~300 °C) aqueous environ-
ments present in nuclear power systems has been studied extensively. Specifically,
chemistries of interest have included: high-temperature pure water with dissolved
hydrogen [1, 2], environments with dissolved Pb [3–5] and/or S [6] species, and caustic
solutions [7]. The latter two environments are known to induce SCC in Fe-based alloys,
such as Alloy 800 (Fe-32Ni-21Cr), but only in laboratory experiments that simulate the
extreme end of plausible chemistries at heat transfer crevices present in the steam gen-
erator of nuclear plants. Thus, while in-service performance has been exemplary, proac-
tive research is being conducted to understand the mechanism of SCC of Alloy 800 in
extreme off-chemistry conditions. Nanoscale characterization has been used to evaluate
the mechanism of SCC for Alloy 800 in acidic sulfate-containing [6], Pb-contaminated
alkaline [3–5], and caustic environments [7, 8]. In these three environments, nanoscale
characterization has revealed that the mechanism of SCC of Alloy 800 is significantly
different.
This brief presentation will review the SCC of Alloy 800 in the mentioned nuclear-
relevant environments. Emphasis is placed on application of novel microscopy methods
to evaluate the SCC mechanism and provide a comparison across multiple environ-
ments of interest. Furthermore, recent work simulating SCC initiation and subsequent
characterization of Alloy 800 in boiling caustic solutions will be discussed.
2 Pb-Caustic SCC of Alloy 800
Alloy 800 is known to undergo SCC in Pb-caustic solutions in 330 °C aqueous environ-
ments at a pH330C > 8.5 [4]. Alloy 800 self-loaded samples were exposed to a 330 °C
solution containing 500 ppm dissolved Pb at a pH330C 9.5. Through-wall SCC was
observed. To better understand the mechanism, site-specific regions containing crack
tips were extracted for further characterization at the nanoscale in a transmission elec-
tron (TEM) microscope. Figure 1 shows the chemistry at a crack tip extracted from an
Alloy 800 sample after exposure to mentioned Pb-caustic solution.
Scanning TEM (STEM) EELS and EDX analysis are excellent complementary tech-
niques to understand nanoscale chemistry across a wide range of light (EELS) and
heavy (EDX) elements. For example, in Fig. 1b, EELS analysis reveals that the crack
tip extracted from the Alloy 800 sample shows a clear enrichment of Ni ahead of the
crack tip, depletion of Cr, and a gradient in Fe content. The latter, specifically, suggests
that Fe is being selectively dissolved from the region ahead of the crack tip (which is
thermodynamically favorable at this pH, near the hydrogen line). This dealloying pro-
cess can result in nanoporosity and formation of a brittle film. Fracture of the brittle film
can lead to propagation of SCC. An additional observation in Fig. 1b is the presence of
a Cr-rich oxide at the base of the crack tip. This oxide is normally protective, but impair-
ment has likely occurred to allow dealloying by selective Fe dissolution. The cause of
passivity impairment is revealed in the EDX elemental map shown in Fig. 1c; Pb is
observed to segregate to the oxide–metal interface, effectively occupying positions that
would normally allow for hydroxide anion adsorption as a precursor to Cr oxide forma-
tion. Thus, the combined EDX and EELS analysis suggests that the mechanism of SCC
Studying Stress Corrosion Cracking Mechanisms Using Novel Nanoscale Characterization 239
Fig. 1. STEM-HAADF (a), -EELS (b), and -EDX (c) imaging and elemental maps showing the
nanoscale chemistry of a crack tip in an Alloy 800 (Fe-32Ni-21Cr) sample formed after exposure to a
Pb-caustic solution at 330 °C (500 ppm dissolved Pb, pH330C 9.5). (Reprinted from Persaud et al. [4],
copyright (2018), with permission from Elsevier).
for Alloy 800 in a high-temperature Pb-caustic solution is related to Pb deposition

which impairs passivity, enabling dealloying at alkaline pH, which results in formation
of a brittle film which fractures during crack propagation.
3 Acid Sulfate SCC of Alloy 800
Alloy 800 is also known to undergo SCC in high-temperature acidic solutions with the
addition of sulfate species [6]. For example, a tensile test was carried out on Alloy
800 in a 280 °C aqueous solution containing 0.55 mol/kg SO42− and pH280C 3. Nanoscale
TEM analysis was carried out on crack tips, which also contained incipient pitting cor-
rosion. Figure 2 shows the scanning electron microscope (SEM) imaging (a) and
STEM-EDX analysis (b, c) of a crack extracted from an Alloy 800 sample after
exposure.
Figure 2 shows a distinct behavior of sulfur in SCC and pitting corrosion of Alloy
800 in high-temperature acid sulfate solutions. In Fig. 2b, a relatively corroded crack is
observed, in comparison to Fig. 1. In addition, sulfur has incorporated in the oxide film,
which also contains Cr (not shown). This mixed sulfide-oxide film impairs passivity
240 S. Y. Persaud
Fig. 2. (a) SEM image of the crack morphology in an Alloy 800 sample after exposure to 0.55 mol/kg
SO42− and pH280C 3 at 280 °C. (b, c) show subsequent nanoscale STEM-EDX analysis of a crack and
pit, respectively. (Reprinted from Persaud et al. [6], copyright (2018), with permission from Elsevier).
and likely leads to SCC via a slip dissolution-type mechanism. Furthermore, the sulfide
concentration at the surface of the metal can change. In Fig. 2c, in the incipient pit,
sulfur is observed to adsorb at the base of the pit, with a relative lack of oxygen.
Therefore, the surface coverage of sulfur seems to play a significant role in transitioning
Alloy 800 in an acidic solution between pitting and SCC degradation. When comparing
Figs. 1 and 2, it is apparent that, in the same material (and in high temperature water),
the mechanism of SCC can be completely different depending on the pH and impurity
present. Furthermore, using nanoscale characterization can allow for clearly defining
differences in the SCC mechanism and could allow for better failure analysis of com-
ponents, i.e., to identify the chemistry present which instigated SCC. Also, by under-
standing the SCC mechanism, precursors for SCC initiation can be postulated; in this
case, pitting corrosion in acid sulfate SCC and dealloying/nanoporosity in Pb-caustic
SCC of Alloy 800. The latter precursor is further evaluated in the next section.
4 SCC in Boiling Caustic Solutions
The preceding sections provided characterization of SCC propagation. However, while

this can provide insight, most SCC mechanisms are proposed in terms of SCC initia-
tion. Therefore, it would be useful to capture the point of crack initiation and apply
similar characterization methods, ideally building on the fundamental understanding of
SCC propagation. For example, analysis of SCC in Alloy 800 exposed to Pb-caustic
solutions, Fig. 1, suggested that dealloying and formation of nanoporosity had a signifi-
cant impact on SCC propagation. To evaluate the mechanism further, tests were con-
ducted in model environments to specifically probe the influence of a dealloyed,
nanoporous layer on fracture. This was conducted by exposing self-loaded Alloy 800
samples to a 135–140 °C 50 wt. % NaOH boiling caustic solution and carefully identi-
fying crack initiation. While very aggressive, this environment was selected to promote
only dealloying-induced SCC, without the complexities associated with Pb deposition
and formation of a passive film (Cr and Fe form soluble species). Figure 3 shows
Fig. 3. Nanoscale characterization of a TEM foil extracted from a sample containing caustic SCC: (a)
STEM image of the TEM foil; (b) high magnification image of nanoporous surface film; (c) EELS
elemental maps of the crack tip and nanoporous film, revealing enrichment of Ni; (d, e) TEM image
and EDX elemental maps of the red square region in (c) showing the nanoporous film at the crack tip;
and (f) concentration profile across the crack tip.
STEM-HAADF, -EELS, and -EDX analysis of crack initiated from an initially formed
porous film in an Alloy 800 sample.
A porous film has formed on the surface of the material, and Fig. 3a shows that the
film is Ni-rich, which continuous downward along the crack path. The Ni enrichment is
due to selective dissolution (dealloying) of Cr and Fe. Figure 3b shows that the crack
has clearly initiated in the porous film and propagated into the underlying pristine mate-
rial. Therefore, the dealloyed layer has directly induced SCC in the Alloy 800 speci-
men. Furthermore, the energy of the crack is sufficient for SCC initiation to cleave into
the underlying substrate material. EDX analysis in Fig. 3d–f reveals that a nanoporous
film is present at the crack tip of the material, which is likely the initial brittle film
which promotes SCC initiation. While this imaging and chemical analysis is beneficial,
a comprehensive analysis for SCC requires further analysis of plasticity (mechanics)
along the crack and at the initiation point. Dislocation analysis in the TEM and using
transmission Kikuchi diffraction (TKD) is shown in Fig. 4.
Figure 4b, c shows that dislocation density tends to increase at the crack tip, suggest-
ing that plastic blunting has occurred resulting in dislocation emission. Along the crack
path, in Fig. 4d, a relative lack of dislocations is observed. The latter suggests that the
crack cleaved into the material without stopping. In other words, the fast fracture initi-
ated in the dealloyed layer had sufficient energy and velocity to propagate some dis-
tance into the pristine alloy before plastically blunting. TKD analysis in Fig. 4f confirms
the same observation, an increase in dislocation density at the crack tip, with a relative
242 S. Y. Persaud
Fig. 4. (a) STEM-HAADF image of a TEM foil extracted from the cross-section of an Alloy
800 U-bend sample after 48 h of exposure to 50 wt.% NaOH at 135–140 °C. (b, c) show BF images of
the crack tip marked by the yellow dashed square in (a) with g vectors [200] and [−11–1], respectively.
(d) shows a BF image of the area behind the crack tip in (a) with g vector of [200], and (e) shows the
[011] zone axis diffraction corresponding to BF images in (b–d). (f) is a GND map of the area outlined
by the red dashed rectangle in (a).
lack of dislocations along the crack walls. Combining the chemical analysis and nano-
porosity observed in Fig. 3 with the dislocation analysis in Fig. 4 provides strong sup-
port for a film-induced cleavage mechanism of SCC [9]. This mechanism may also be
operating in the Pb-caustic cracking of Alloy 800 discussed previously, but is more
complex due to Pb deposition and formation of a Cr-rich passive film.
5 Conclusions
This brief review has shown that application of modern microscopy methods is benefi-
cial for characterizing SCC at the nanoscale and providing new insight into historical
SCC mechanisms. Pb-caustic SCC of Alloy 800 is proposed to occur due to dealloying
and deposition of Pb at the oxide–metal interface. In contrast, acid sulfate SCC of Alloy
800 may occur via a slip dissolution mechanism, furthered by impairment of the pas-
sive film by sulfur incorporation. By determining the corrosion phenomena instigating
Pb-caustic SCC, it is possible to design an experiment to study SCC initiation of Alloy
800 in a model environment – in this case, 50 wt. % NaOH solution. Nanoscale charac-
terization of SCC initiation in the latter environment allows for determining the mecha-
nism of dealloying-induced fracture, which may be relevant to the Pb-caustic
environment.
Acknowledgments The authors would like to acknowledge H. Gholamzadeh and A. Shaik for provid-
ing for Figs. 3 and 4, as well as valuable discussions. This research was funded by the Natural Sciences
and Engineering Research Council of Canada (NSERC) and a University Network of Excellence in
Nuclear Engineering (UNENE) Research Chair in Corrosion Control and Materials Performance.
References
1. Thomas LE, Bruemmer SM. High-resolution characterization of intergranular attack and stress
corrosion cracking of alloy 600 in high-temperature primary water. Corrosion. 2000;56:572–87.
2. Dugdale H, Armstrong DEJ, Tarleton E, Roberts SG, Lozano-Perez S. How oxidized grain bound-
aries fail. Acta Mater. 2013;61:4707–13.
3. Mazzei GB, Duff J, Burke MG, Scenini F, Meredith G, Horner T. Microstructural characterization
of alloy 690TT exposed to Pb-containing caustic solutions. In: 19th Int. Conf. Environ. Degrad.
Mater. Nucl. Power Syst. – Water React. Boston, USA: American Nuclear Society (ANS); 2019.
p. 1174–80.
4. Persaud SY, Smith JM, Judge CD, Long F, Korinek A, Capell BM, Wright MD. High resolution
characterization of Pb-SCC in alloy 800 exposed to 330 °C mildly caustic environments. Corros
Sci. 2018;140:333–48.
5. Persaud SY, Smith JM, Langelier B, Capell BM, Wright MD. An atom probe tomography study of
Pb-caustic SCC in alloy 800. Corros Sci. 2018;140:159–67.
6. Persaud SY, Long F, Korinek A, Smith JM. High resolution characterization of sulfur-assisted
degradation in Alloy 800. Corros Sci. 2018;140:122–33.
7. Bryk MA, Artymowicz D, Carcea AG, Newman RC. Nanoporosity formation in alloy 800 inves-
tigated with respect to stress-corrosion cracking in high-temperature caustic solutions. In: Proc.
36th Annu. Conf. Can. Nucl. Soc. Boston, USA: American Nuclear Society (ANS); 2016.
8. Persaud SY, Smith JM, Judge CD, Bryk M, Newman RC, Burke MG, de Curieres I, Capell BM,
Wright MD. Using modern microscopy to “fingerprint” secondary side Scc in Ni–Fe alloys.
Minerals, metals and materials series. American Nuclear Society (ANS), Boston, MA, USA. 2019.
9. Sieradzki K, Newman RC. Stress-corrosion cracking. J Phys Chem Solids. 1987;48(11):1101–13.
On-Line ICP-MS in Electrocatalysis Research:
Platinum Dissolution Studies
Valentín Briega-Martos1 (*) and Serhiy Cherevko1

1
Forschungszentrum Jülich GmbH, Helmholtz-Institute Erlangen-Nürnberg for Renewable
Energy (IEK-11), Erlangen, Germany
v.briega@fz-juelich.de
Abstract. The degradation of electrocatalysts as a consequence of their dissolution in

the operating conditions of fuel cells and electrolyzers remains one of the main challenges
in electrochemical energy conversion. In order to investigate and predict the kinetics of
dissolution of electrocatalysts in real conditions, the electrochemical on-line inductively
coupled mass spectrometry (ICP-MS) technique was developed. This approach is
employed in electrocatalysis research as a tool for enabling the detection of dissolved
elements in the electrolyte going down to the parts-per-trillion (ppt) range and providing
at the same time multielement analysis and a wide linear dynamic range of quantification.
In this way, on-line ICP-MS has been successfully applied for fundamental studies of
noble metal dissolution, and the previous works related to the particular case of platinum
are revised in the present chapter.
Keywords: Dissolution · Stability · Platinum · Inductively coupled plasma mass

spectrometry · Scanning flow cell · Fuel cell
1 On-Line Inductively Coupled Plasma Mass Spectrometry

in Electrochemistry
The degradation of electrocatalysts as a consequence of their dissolution in the operat-

ing conditions of fuel cells and electrolyzers remains one of the main challenges in
electrochemical energy conversion. In order to investigate and predict the kinetics of
dissolution of electrocatalysts in real conditions, the electrochemical on-line induc-
tively coupled mass spectrometry (ICP-MS) technique was developed [1]. This
approach is employed in electrocatalysis research as a tool for enabling the detection of
dissolved elements in the electrolyte going down to the parts-per-trillion (ppt) range
and providing at the same time multielement analysis and a wide linear dynamic range
of quantification. In this way, on-line ICP-MS has been successfully applied for funda-
mental studies of noble metal dissolution [2, 3] and for the investigation of several
electrocatalytic reactions such as oxygen and hydrogen evolution reactions (OER and
https://doi.org/10.1007/978-3-031-17425-4_34
246 V. Briega-Martos and S. Cherevko
HER) [4, 5], hydrogen oxidation reaction (HOR) [6], and oxygen reduction reaction
(ORR) [7]. Among the different ways for connecting the ICP-MS to an electrochemical
cell, the scanning flow cell coupled to the ICP-MS (SFC-ICP-MS), in which the outlet
of a V-shaped SFC is directly hyphenated to the sample introduction system of ICP-MS,
stands out for its ability to screen stability and high-throughput activity of gradient
composition libraries [8].
2 On-Line ICP-MS in Platinum Stability Studies
Fuel cell cars use polymer electrolyte membrane fuel cells (PEMFCs) with Pt nanopar-
ticles for ORR at the cathode, primarily because of their better long-term stability in
comparison to other electrocatalysts. However, even pure Pt catalysts degrade under
real-life conditions, and this degradation is mainly linked to electro-oxidation and dis-
solution processes. Polycrystalline platinum dissolution was studied in both acidic and
alkaline media by SFC-ICP-MS, and it was observed that the oxidation and subsequent
reduction of the surface lead to transient dissolution, which is higher by a factor of two
in base [9]. In addition, the results suggested that varying the pH does not modify the
mechanism of the OER on Pt, since its dissolution rate is the same in acidic and alkaline
solutions. More recently, investigations with Pt single-crystal electrodes have been car-
ried out, which offer the possibility of a more detailed understanding of these processes
at the atomic level [10]. These studies have shown clear trends in the onset potential for
dissolution and the amounts of dissolved Pt that can be correlated to the differences in
the surface structure energies and coordination of the different well-defined surfaces. In
this way, Pt(111) resulted to have a more positive onset potential for dissolution, and
the generalized observed trend in dissolution rates and dissolved amounts was
Pt(110) > Pt(100) > Pt(111).
The dissolution results for Pt single-crystal electrodes from on-line ICP-MS can be
combined with in situ surface X-ray diffraction (SXRD) measurements, and by follow-
ing this approach, it was proposed that the differences in the onset potential for anodic
dissolution on Pt(100) and Pt(111) have their origin in the different atomic structures of
the initial oxide [11]. On Pt(111), the extracted Pt atom lies directly above its original
site, and therefore the reversibility observed at low coverage is explained in terms of its
facile return to that site. However, on Pt(100), the extracted Pt atom moves laterally
away from its initial site and provokes the immediate extraction of a second atom, lead-
ing to the formation of a stripe structure. According to Density Functional Theory
(DFT) studies, this mechanism would produce unstable surface atoms at strip ends that
could dissolve both during the oxidation itself and during the subsequent oxide reduc-
tion, making the process irreversible from its onset. More recently, potentiostatic hold
experiments for Pt(100) and Pt(111) have been carried out. This program allows sepa-
rating the anodic and cathodic dissolution peaks and studying oxidation and dissolution
kinetics versus time. The fundamental knowledge obtained from model bulk electrodes
will contribute to delve into the understanding of restructuring and dissolution pro-
cesses of Pt, which ultimately will provide guidelines for minimizing Pt catalyst degra-
dation, both in half-cell aqueous electrolyte and fuel cell environment.
On-Line ICP-MS in Electrocatalysis Research: Platinum Dissolution Studies 247
Acknowledgments Funding is acknowledged by Deutsche Forschungsgemeinschaft (Grant no.

1763/5-1).
References
1. Kasian O, Geiger S, Mayrhofer KJJ, Cherevko S. Electrochemical on-line ICP-MS in electroca-

talysis research. Chem Rec. 2019;19:2130–42. https://doi.org/10.1002/tcr.201800162.
2. Schalenbach M, Kasian O, Ledendecker M, Speck FD, Mingers AM, Mayrhofer KJJ, Cherevko
S. The electrochemical dissolution of noble metals in alkaline media. Electrocatalysis.
2018;9:153–61. https://doi.org/10.1007/s12678-017-0438-y.
3. Stumm C, Grau S, Speck FD, Hilpert F, Briega-Martos V, Mayrhofer K, Cherevko S, Brummel
O, Libuda J. Reduction of oxide layers on Au(111): the interplay between reduction rate, dis-
solution, and restructuring. J Phys Chem C. 2021;125:22698–704. https://doi.org/10.1021/acs.
jpcc.1c03969.
4. Lončar A, Escalera-López D, Cherevko S, Hodnik N. Inter-relationships between oxygen evolu-
tion and iridium dissolution mechanisms. Angew Chem Int Ed. 2022;61:e202114437. https://doi.
org/10.1002/anie.202114437.
5. Cherevko S, Geiger S, Kasian O, Kulyk N, Grote J-P, Savan A, Shrestha BR, Merzlikin S, Breitbach
LA, Mayrhofer KJJ. Oxygen and hydrogen evolution reactions on Ru, RuO2, Ir, and IrO2 thin film
electrodes in acidic and alkaline electrolytes: a comparative study on activity and stability. Catal
Today. 2016;262:170–80. https://doi.org/10.1016/j.cattod.2015.08.014.
6. Davydova ES, Speck FD, Paul MTY, Dekel DR, Cherevko S. Stability limits of Ni-based hydrogen
oxidation electrocatalysts for anion exchange membrane fuel cells. ACS Catal. 2019;9:6837–45.
https://doi.org/10.1021/acscatal.9b01582.
7. Polani S, MacArthur KE, Klingenhof M, Wang X, Paciok P, Pan L, Feng Q, Kormányos A,
Cherevko S, Heggen M, Strasser S. Size and composition dependence of oxygen reduction reaction
catalytic activities of Mo-doped PtNi/C octahedral nanocrystals. ACS Catal. 2021;11:11407–15.
https://doi.org/10.1021/acscatal.1c01761.
8. Schuppert AK, Topalov AA, Savan A, Ludwig A, Mayrhofer KJJ. Composition-dependent oxygen
reduction activity and stability of Pt-Cu thin films. ChemElectroChem. 2014;1:358–61. https://doi.
org/10.1002/celc.201300078.
9. Cherevko S, Zeradjanin AR, Keeley GP, Mayrhofer KJJ. A comparative study on gold and plati-
num dissolution in acidic and alkaline media. J Electrochem Soc. 2014;161:H822–30. https://doi.
org/10.1149/2.0881412jes.
10. Sandbeck DJS, Brummel O, Mayrhofer KJJ, Libuda J, Katsounaros I, Cherevko S. Dissolution
of platinum single crystals in acidic medium. ChemPhysChem. 2019;20:2997–3003. https://doi.
org/10.1002/cphc.201900866.
11. Fuchs T, Drnec J, Calle-Vallejo F, Stubb N, Sandbeck DJS, Ruge M, Cherevko S, Harrington
DA, Magnussen OM. Structure dependency of the atomic-scale mechanisms of platinum electro-
oxidation and dissolution. Nat Catal. 2020;3:754–61. https://doi.org/10.1038/s41929-020-0497-y.
Part IV Energy and Environmental Materials
Adsorption of Naphthenic Acids from Oil Sand
Process-Affected Water (OSPW) Using Commercially
Viable Petcoke-Sourced Activated Carbon
Elmira Nazari1 (*), Tyler M. Roy1, Oliver K. L. Strong1, and Andrew J. Vreugdenhil1
1
Materials Science, Trent University, Peterborough, ON, Canada
elmiranazari@trentu.ca; avreugdenhil@trentu.ca
Abstract. Oil reserves present in Alberta, Canada, in the form of oil sands, are estimated
to contain 169 billion barrels of recoverable bitumen [1]. The extraction process of bitu-
men involves the use of large volumes of caustic hot water, that even with recycling
efforts in place, have continued to accumulate for decades in large settling basins, known
as tailings ponds [2]. As of a 2019 report given by the Alberta Energy Regulator, there is
an estimated 1302 Mm3 of fluid tailings being stored in tailings ponds [3]. The high toxic-
ity of this wastewater commonly known as oil sands process affected waters (OSPW) has
impeded its release resulting in a zero-discharge policy. Although many different con-
taminates have been identified to contribute to this toxicity, evidence indicates naphthenic
acids (NAs) are the main toxic component present in OSPW and are therefore of major
concern [4, 5].
Keywords: Oil Sands Process · Affected Water (OSPW) · Naphtenic acid (NA)
· Activated carbon · Adsorption · Contaminent
Oil reserves present in Alberta, Canada, in the form of oil sands, are estimated to con-
tain 169 billion barrels of recoverable bitumen [1]. The extraction process of bitumen
involves the use of large volumes of caustic hot water, that even with recycling efforts
in place, have continued to accumulate for decades in large settling basins, known as
tailings ponds [2]. As of a 2019 report given by the Alberta Energy Regulator, there is
an estimated 1302 Mm3 of fluid tailings being stored in tailings ponds [3]. The high
toxicity of this wastewater commonly known as oil sands process affected waters
(OSPW) has impeded its release resulting in a zero-discharge policy. Although many
different contaminates have been identified to contribute to this toxicity, evidence indi-
cates naphthenic acids (NAs) are the main toxic component present in OSPW and are
therefore of major concern [4, 5].
The term naphthenic acid represents a very broad distribution of carboxylic acid spe-
cies that are ubiquitous in the Athabasca region and the oil sands [1, 5]. NAs are a
diverse and complex class of compounds with potentially thousands of different com-
ponents, which makes their removal from OSPW challenging. A promising remediation
strategy that has gained more attention recently is the use of activated carbon adsor-
bents [6–8]. However, there is a need to better characterize the NA adsorption system
while using activated carbon. A few studies have focused on model NA adsorption
https://doi.org/10.1007/978-3-031-17425-4_35
252 E. Nazari et al.
Fig. 1. Kinetics of adsorption of seven NA compounds using (1) KOHAC, (2) COMAC, and (3) PAAC.
recently using activated carbon [9–13], but given the diversity of NA species within
OSPW, characterizing the adsorption of a broader range of model NA species is
warranted.
In this study, we characterized the adsorption of seven model NA species onto three
different activated carbons. Two of the ACs were produced in our laboratory with a
focus on commercial viability and waste recycling. They are a petroleum coke-sourced
activated carbon made in our lab (KOHAC) and a phosphoric acid activated waste-
wood-based AC (PAAC). These were compared with a commercial AC (COMAC).
Adsorption profiles of each model NA in terms of kinetics and isotherms were used to
characterize each activated carbon as discussed below.
Adsorption kinetics results are shown in Fig. 1 for adsorption on the three ACs. This
shows that for each of the ACs, the uptake of NAs is very dependent on the NA itself.
Diphenylacetic acid (DPA) and dicyclohexylacetic acid (DHA) clearly are most
adsorbed, while heptanoic acid (HA), cyclohexane acetic acid (CHA), and 2-methyl-
cyclohexane acetic acid (2-MCH) all had lower percent adsorption, and 1,4 cyclohenx-
anedicarboxylic acid (1,4-CHA) and succinic acid (SA) had very little adsorption.
Although these species represent a very small cross-section of the possible NA species,
what this demonstrates is that optimal performance for AC removal of adsorption will
require more species-specific adsorption evaluation than total NA adsorbed.
The rate of adsorption for each of the evaluated model NAs is also dependent on
the method of activation, surface functionality, and textural features of the activated
Adsorption of Naphthenic Acids from Oil Sand Process-Affected Water (OSPW) Using… 253
Table 1. KOHAC textural modification by multiple activation cycles.

Surface area Mesoporosity Yield
Activation cycles Total activation time (minutes) (m2/g) (%) (%)
1 30 1217 23.4 64
2 60 1167 43.1 48
3 90 1084 61.2 34
Table 2. Naphthenic acid adsorption half-life for multiple activation cycles.

DPA t1/2 CHA t1/2 HA t1/2
Activation cycles (min) (min) (min)
1 26.5 34.3 51.4
2 10.5 18.5 11.1
3 6.64 2.98 12.0
carbon. For these model NAs, the rate of adsorption is fastest on the COMAC, in
which adsorption equilibrium was achieved between 30 and 60 minutes. Equilibrium
on the KOHAC is achieved between 8 and 24 hours depending on the model NA. This
difference in kinetics may be a result of the differences in pore size distribution
between the COMAC and KOHAC. However, the PAAC which has the highest meso-
porosity also exhibits a slower approach to equilibrium of 2–8 hours. The influence of
particle size is considered a confounding factor in these evaluations.
The interplay of surface functionality as generated by method of activation and feed-
stock makes it challenging to evaluate only the effect of textural material properties.
However, the use of multiple activation cycles gives rise to a convenient method for
increasing mesoporosity without significantly altering the surface composition or
surface area while using the same feedstock and activation chemistry. This approach
was evaluated for a range of model NAs on singly, doubly, and triply activated KOHAC.
As shown in Table 1, the multiple heating cycles resulted in a systematic modifica-
tion of the textural properties of the KOHAC with an approximately 20% increase in
mesoporosity percent surface area per activation cycle.
The multiple activations demonstrated faster kinetics for each model NA as shown
in Table 2 for DPA, CHA, and HA. The adsorption kinetics were modeled using both
pseudo-first-order kinetics and multiple exponential kinetics. The half-life of adsorp-
tion is shown for the multiple exponent kinetics models. X-ray photoelectron spectros-
copy (not shown) demonstrates no significant change in surface functionality. This
increase in mesoporosity is expected to improve the internal diffusion of adsorbate
within the AC pores to facilitate adsorbate access to the microporous space.
Conveniently, with minimal chemical input, the adsorption kinetics for a range of
model NAs can be improved by thermal cycling allowing for tailoring of the textural
properties of the material and their resulting potential application. The evaluation of
species-specific NA removal from both synthetic and industrial water samples contain-
ing a broad mixture of NAs is in progress using electrospray ionization mass spectros-
copy (ESI) to provide further insights into the optimization of AC surface chemistry and
textural properties for oil sands process water remediation.
254 E. Nazari et al.
References
1. Brown LD, Ulrich AC. Oil sands naphthenic acids: a review of properties, measurement, and
treatment. Chemosphere. 2015;127:276–90. https://doi.org/10.1016/j.chemosphere.2015.02.003.
Elsevier Ltd.
2. Allen EW. Process water treatment in Canada’s oil sands industry: II. A review of emerging tech-
nologies. J Environ Eng Sci. 2008;7(5):499–524. https://doi.org/10.1139/S08-020. ICE Publishing.
3. A. Energy Regulator. State of fluid tailings management for mineable oil sands. 2019. [Online].
Available: www.aer.ca
4. Li C, Fu L, Stafford J, Belosevic M, Gamal El-Din M. The toxicity of oil sands process-affected
water (OSPW): a critical review. Sci Total Environ. 2017;601–602:1785–802. https://doi.
org/10.1016/j.scitotenv.2017.06.024. Elsevier B.V.
5. Clemente JS, Fedorak PM. A review of the occurrence, analyses, toxicity, and biodegradation
of naphthenic acids. Chemosphere. 2005;60(5):585–600. https://doi.org/10.1016/j.chemo-
sphere.2005.02.065. Elsevier Ltd.
6. Mohamed MH, Wilson LD, Headley JV, Peru KM. Novel materials for environmental remediation
of tailings pond waters containing naphthenic acids. Process Saf Environ Prot. 2008;86(4):237–43.
https://doi.org/10.1016/j.psep.2008.04.001.
7. Islam MS, Zhang Y, McPhedran KN, Liu Y, Gamal El-Din M. Mechanistic investigation of indus-
trial wastewater naphthenic acids removal using granular activated carbon (GAC) biofilm based
processes. Sci Total Environ. 2016;541:238–46. https://doi.org/10.1016/j.scitotenv.2015.09.091.
8. Iranmanesh S, Harding T, Abedi J, Seyedeyn-Azad F, Layzell DB. Adsorption of naphthenic acids
on high surface area activated carbons. J Enviro Sci Health A. 2014;49(8):913–22. https://doi.org/1
0.1080/10934529.2014.894790.
9. Martinez-Iglesias A, Niasar HS, Xu CC, Ray MB. Adsorption of model naphthenic acids in
water with granular activated carbon. Adsorpt Sci Technol. 2015;33(10):881–94. https://doi.
org/10.1260/0263-6174.33.10.881.
10. Niasar HS, Li H, Kasanneni TVR, Ray MB, Xu CC. Surface amination of activated carbon and
petroleum coke for the removal of naphthenic acids and treatment of oil sands process-affected
water (OSPW). Chem Eng J. 2016;293:189–99. https://doi.org/10.1016/j.cej.2016.02.062.
11. Kawano T, Kubota M, Onyango MS, Watanabe F, Matsuda H. Preparation of activated carbon
from petroleum coke by KOH chemical activation for adsorption heat pump. Appl Therm Eng.
2008;28(8–9):865–71. https://doi.org/10.1016/j.applthermaleng.2007.07.009.
12. Niasar HS, Li H, Das S, Kasanneni TVR, Ray MB, (Charles) Xu C. Preparation of activated petro-
leum coke for removal of naphthenic acids model compounds: Box-Behnken design optimiza-
tion of KOH activation process. J Environ Manag. 2018;211:63–72. https://doi.org/10.1016/j.
jenvman.2018.01.051.
13. Seyedy Niasar H, (Chunbao) Xu C, Ray M. Treatment of oil sands process-affected water using
activated and surface modified petroleum coke for organic compounds recovery. Chem Biochem
Eng. 2017;Doctor of. 190.
Crystallization of Nanoscale Garnet-Type Li-Ion
Conductors for All-Solid-State Batteries (ASSBs)
Senhao Wang1, Hsien-chieh Chiu1, and George P. Demopoulos1 (*)
1
Materials Engineering, McGill University, Montreal, QC, Canada
george.demopoulos@mcgill.ca
All-solid-state lithium-ion batteries (ASSLIBs) are considered to have substantial ben-

efits over conventional Li-ion batteries based on flammable organic solvent electro-
lytes, such as high energy density, superior safety, and excellent mechanical strength [1,
2]. The two types of battery architecture are compared in Fig. 1. As the core component
of ASSLIBs, a lot of solid-state Li-ion conductors (electrolytes) have been developed
and extensively explored in recent years, including crystalline Li1 + xAlxTi2 − x(PO4)3
(NASICON), Li7La3Zr2O12 (Garnet), Li3xLa2/3 − xTiO3 (Perovskite), Li10GeP2S12 (thio-
LISICON), amorphous LiPON, etc [3, 4]. Among them, Al-doped cubic-structured
garnet-type Li7La3Zr2O12 (Al-c-LLZO) solid-state Li-ion conductor has attracted the
most attention due to its relatively high lithium-ion conductivity (10−3 ~ 10−4 S cm−1 at
ambient temperature), wide electrochemical window up to 6 V (vs. Li / Li+), and excel-
lent chemical stability against lithium metal used as anode [5]. Despite of its good
electrochemical properties, the commercial application potential of Al-c-LLZO is
unfortunately hindered by its complex synthesis procedure and fabrication, which are
time and energy consuming but also its interfacing with the electrodes. Up to now, the
synthesis and processing of cubic-structured garnet conductors are based on sol-gel and
solid-state synthesis and sintering usually done at temperatures above 1100 °C, as well
as multiple cycles of sintering and milling [5]. This process results in micro-sized par-
ticles and potential Li-loss, which are unfavorable for further processing and electrode–
electrolyte assembly with high Li-ion conductivity [6]. Here, we tackle this problem
and firstly report a novel low-temperature synthesis-processing route to stabilize the
cubic phase of Li6.1Al0.3La3Zr2O12 (Al-c-LLZO), while limiting particle growth to
~100–200 nm. This synthesis route involves three steps: (1) co-precipitation, (2) hydro-
thermal aging, and (3) calcination. The cubic Al-c-LLZO phase is obtained at tempera-
tures as low as 600 °C by reactive transformation/calcination of aqueous co-precipitated
metal hydroxide intermediate of controlled crystal orientation. Through extensive
structural characterization of the samples collected after different synthesis steps, the
crystalline phase transition from M(OH)x (amorphous) to M(OH)x (crystalline) to LazZryOw and
finally to Li7-vAlv/3La3Zr2O12 (Al-c-LLZO) was observed and presented in Fig. 2. Based
on the XRD patterns, the crystallization of Al-c-LLZO phase is fully developed when
Senhao Wang is the presenting author of this chapter.
https://doi.org/10.1007/978-3-031-17425-4_36
256 S. Wang et al.
Fig. 1. The schematic illustration of the principle of conventional lithium-ion battery and all-solid-
state lithium-ion battery.
Fig. 2. X-ray diffraction patterns of low-T synthesized Al-c-LLZO samples collected after co-
precipitation process, hydrothermal aging, and calcination at various temperatures from 400 to 700 °C.
Crystallization of Nanoscale Garnet-Type Li-Ion Conductors for All-Solid-State… 257
Fig. 3. Al-c-LLZO analysis. (a) Morphology and (b) nanosized distribution of low-T synthesized
Al-c-LLZO nanoparticles.
1000
800
600
-Z'' (ohm)
21 ºC
400 30 ºC
40 ºC
50 ºC
200 60 ºC
70 ºC
80 ºC
0
0 200 400 600 800 1000
Z' (ohm)
Fig. 4. Li-ion conductivity measurement via EIS Nyquist plots of the low-T synthesized Al-c-LLZO
samples at various temperatures (frequency range: 0.1 Hz to 1 MHz, amplitude: 10 mV).
temperature reached 600 °C, and no further phase transition occurred when the tem-
perature was raised to 700 °C, which suggests that the pure Al-c-LLZO phase can be
obtained at 600 °C, the lowest calcination temperature compared to the published
results so far. Moreover, according to the XRD results, the size of the primary nanocrys-
tals of the low-T synthesized Al-c-LLZO was calculated through Scherrer equation to
be ~29 nm. The morphology and particle size distribution of the low-T synthesized
Al-c-LLZO particles are presented in Fig. 3. The mean size of the particles is 180 nm.
As for the Li-ion conductivity, the low-T synthesized Al-c-LLZO particles after one-
step pellet compaction and sintering were characterized by Electrochemical Impedance
Spectroscopy (EIS) at room temperature (21 °C) as well as from 30 to 80 °C with 10 °C
temperature intervals. The Nyquist plots are shown in Fig. 4. The Li-ion conductivity at
RT is approaching 10-3 S ∙ cm−1 and the calculated activation energy of Li-ion migra-
tion is 0.174 eV. Through this novel synthesis route, nanoscale Al-c-LLZO particles
258 S. Wang et al.
with high ionic conductivity are particularly attractive in promoting compatibility

between cathode materials and solid electrolytes for ASSLIBs. The mechanism of the
low-T synthesis route, as well as the results in detail, will be provided during the oral
presentation.
Acknowledgments This work is supported by NSERC, Hydro-Québec, and McGill’s Engineering

Doctoral Award program.
References
1. Janek J, Zeier WG. A solid future for battery development. Nat Energy. 2016;1(9):16141.
2. Masias A, Marcicki J, Paxton WA. Opportunities and challenges of lithium ion batteries in automo-
tive applications. ACS Energy Lett. 2021;6(2):621–30.
3. Zhang B, Tan R, Yang L, Zheng J, Zhang K, Mo S, Lin Z, Pan F. Mechanisms and properties of
ion-transport in inorganic solid electrolytes. Energy Stor Mater. 2018;10:139–59.
4. Murali A, Sakar M, Priya S, Vijayavarman V, Pandey S, Sai R, Katayama Y, Abdul Kader M,
Ramanujam K. Insights into the emerging alternative polymer-based electrolytes for all solid-state
lithium-ion batteries: a review. Mater Lett. 2022;313:131764.
5. Xu L, Li J, Deng W, et al. Garnet solid electrolyte for advanced all‐solid‐state Li batteries[J].
Advanced Energy Materials, 2021, 11(2): 2000648.
6. Balaish M, Gonzalez-Rosillo JC, Kim KJ, Zhu Y, Hood ZD, Rupp JLM. Processing thin but robust
electrolytes for solid-state batteries. Nat Energy. 2021;6(3):227–39.
Resin Technologies to Improve Battery Metal
Recovery
Jeffrey Sehn1 (*) and Mike Weatherill1

1
Purolite LLC, King of Prussia, PA, USA
jeffrey.sehn@purolite.com; mike.weatherill@purolite.com
Abstract. The emerging electric vehicle industry has resulted in an increased

demand for metals used in the production of batteries. Quebec, like a good number of
jurisdictions around the world, intends to ban gas cars by 2035. These new cars will
need batteries. And competing with these, cell phones, laptops, and other battery-
powered equipment will be competing for the same materials, notably nickel, cobalt,
and lithium. Every gram will count.
Keywords: Resin Ion Exchange Chelating Lithium Cobalt Nickel Brine Ore
Recycling
The emerging electric vehicle industry has resulted in an increased demand for metals
used in the production of batteries. Quebec, like a good number of jurisdictions around
the world, intends to ban gas cars by 2035. These new cars will need batteries. And
competing with these, cell phones, laptops, and other battery-powered equipment will
be competing for the same materials, notably nickel, cobalt, and lithium. Every gram
will count.
While significant quantities of cobalt, nickel, manganese, iron, and aluminum are in
different types of batteries, lithium is present in most types of batteries.
These other minerals are typically mined. Lithium has other options.
Three main sources of lithium are being considered in the industry. We consider
brines, ores, and recycled materials.
Lithium brines, these lithium-rich liquors are pumped out and concentrate the lith-
ium. Applications, where this may be seen, are in salt lakes, oil field operations, and
geothermal power plants. New methods are in development for direct lithium recovery.
Ion exchange resin is included in some of the methods being considered.
Mining of lithium-rich ore can be found in spodumene, lepidolite, and clays around
the world. In Quebec and the rest of Canada, we mainly have ore-mining operations.
The ore is turned into a liquor or leach which would then be purified.
Urban mining, or recycling of batteries in other words, would have all the desired
components that would need to be broken down, split into base materials, and purified
for reuse.
https://doi.org/10.1007/978-3-031-17425-4_37
260 J. Sehn and M. Weatherill
Processing is required to separate the raw materials for the batteries and to remove
impurities. Existing treatments only go so far. Ion exchange resin technology is a
method to upgrade target metals from low concentration to high concentration or to
remove trace impurities from a target metal. Similar methods exist to extract as much
high-value metals as possible from effluent.
Ion exchange resin are small polymer beads that exchange an ion on the bead with
an ion in a solution. Resin beads typically range between 300 and 1200 μm in diameter,
and each has millions of ion exchange sites. For scale, there are nearly 7,000,000 indi-
vidual beads in each liter of resin. Resin exchanges ions from the bead for target ions
from a solution. This process is mostly reversible through regeneration, where the target
ions come off in a concentrated stream and the resin is returned to its initial form.
Chelating resins have a porous structure that helps in capturing metal ions. They are
also produced in different functional groups. Two functional groups for this subject are
iminodiacetic or aminophosphonic. These resins are designed to remove metals from
solution with different selectivity depending on impurity concentration, solution com-
position, and pH.
Lithium from ores involves purification through impurity removal. The following is
a simplified flow diagram that can highlight where chelating resin can be useful.
Conventional techniques, such as precipitation and solvent extraction, are used to
remove bulk impurities. Resin can be used where higher purity is required or where it
can improve product value.
Resin Technologies to Improve Battery Metal Recovery 261
In this block flow diagram, the text in blue is where there is an ion exchange resin
solution.
Selective impurity removal within hydrometallurgical processes can have great
advantages in purifying lithium leach.
Fluoride is found in lithium leach in concentrations of around 50 mg/L. Fluoride can
affect certain electrolyte processes. Typical fluoride removal is done by precipitation.
Ion exchange resins can be used as a polishing step to attain less than 1 mg/L of fluoride
prior to the LiCO3 precipitation steps. Resins preloaded with aluminum have been suc-
cessful for this application. The use of an amino phosphonic chelating resin or Shallow
Shell Technology cation resin has proven effective.
Calcium and magnesium hardness in lithium leach is a common impurity to remove.
Hardness affects the electrolyte processes. Chelating resins are common methods to
remove calcium and magnesium from solutions. High-capacity products are available.
These products would be ideal for removing typical feed levels of calcium (500–
800 mg/L) and magnesium (10–100 mg/L) to process specifications of less than 5 mg/L
and 1 mg/L, respectively. Different products could be used for coarse impurity removal
and final polishing.
Boron is commonly in the lithium leach and can affect the electrolyte process. There
are specialty boron selective resins that can isolate the boron even when there is a low
concentration within a more concentrated background solution. 80–800 mg/L is in lithium
solution and these resins can reduce this to below 1 mg/L prior to LiCO3 precipitation.
Whatever the impurity in the ore leach, ion exchange resin can remove those impuri-
ties to low levels which would improve process efficiency and product value.
Recycling of batteries will be crucial for the sustainability of the industry. The
expected life of EV batteries is 8 years. Recycling is important not only for environ-
mental reasons but also for the recovery of raw materials to produce new batteries.
The typical recycling method involves collecting batteries, crushing and removing
the plastic and metal casing materials, and creating black mass. The black mass c ontains
everything else that makes up the battery. The black mass is turned into a leach and then
processed.
262 J. Sehn and M. Weatherill
Pregnant Leach Solution contains the battery metals at different concentrations. The
bulk separation and purification would typically be done with precipitation, sedimenta-
tion, and filtration. The filtrate would typically still have under 100 mg/L of each cobalt,
nickel, and manganese. With different resins, different selectivity exist. The right che-
lating resin could reduce these impurities down to under 5 mg/L.
Having these impurities all in the same effluent after elution is not acceptable when
trying to recover as much raw material as possible.
In cases where ion-exchange resins lack the appropriate selectivity for certain target
metals, but solvents do exist, such as the ability to separate zinc from cobalt and cobalt
from nickel sulfates, solvent-impregnated resins have been developed that combine the
selectivity of the solvents with the ease of engineering of ion-exchange resins.
Solvent-impregnated resin is a specialty product and there are currently two types
based on solvents used. One is selective for zinc over cobalt and nickel and the other is
selective for cobalt over nickel. This type of impurity removal is efficient down to ppt
levels. Solvent-impregnated resin has the advantage of the physical characteristics of
ion exchange resin as well as the chemical characteristics of the solvent, same selectiv-
ity ranges, and same pH dependence.
These two solvent-impregnated resins were tested in a column set up in our R&D
facilities and we have established some capacities for these. The two products tested are
Purolite MTX7010 and MTX8010.1
MTX7010 is a D2EHPA-impregnated resin and has the ability to remove small con-
centrations of zinc from a cobalt background. Using a synthetic liquor containing
50 g/L cobalt and 1 g/L zinc was passed through the resin bed. 20 BV (1 BV is the
equivalent volume of the resin in the column) of the solution is passed through the resin
at a flow rate of 2 BV/hour until zinc in the barren returns to the feed concentration, and
there was 4 BV before zinc in the barren exceeded 2 mg/L. There was no cobalt loaded
on the resin, 94% of the zinc was eluted with 3BV of 1 mol sulfuric acid.
1
Test Results for MTX7010 and MTX8010 were published in “Application of Solvent-Impregnated
Resins for Production of High-Purity Nickel and Cobalt Liquors” by Johanna van Deventer, William
McDevitt, Purolite, Presented at the Alta Metallurgical Conference, 19–25 May 2019, Perth, Australia.
Resin Technologies to Improve Battery Metal Recovery 263
MTX7010: Cobalt and zinc in barren during adsorption, wash, and elution steps
MTX8010 is a di(2,4,4-trimethylpentyl) phosphinic acid-impregnated resin and has
the ability to remove small concentrations of cobalt from a nickel background. Using a
synthetic liquor containing 80 g/L nickel and 1 g/L cobalt was passed through the resin
bed. 12 BV of solution is passed through the resin at a flow rate of 2 BV/h until cobalt
in the barren reached 0.5 g/L breakpoint, there was 7 BV before cobalt in the barren
began to increase.
MTX8010: Cobalt concentration during adsorption, wash, and elution steps

Additional batch testing was done with a 100:1 solution to resin ratio over 24 hours.
The solution was 120 g/L nickel and 1 g/L cobalt. After the time elapsed, an elution was
done, and 8 g/L cobalt and 1 g/L nickel were extracted from the resin.
Ion exchange resins are set to play an important role in production of high-purity
battery metal liquors.
Study of Adsorption Thermodynamics of Model
Naphthenic Acids for Determining Adsorption
Mechanism Using Commercially Viable Petroleum
Coke-Activated Carbon
Tyler M. Roy1 (*) and Andrew J. Vreugdenhil2
1
Environmental and Lifesciences, Trent University, Peterborough, ON, Canada
tylerroy@trentu.ca
2
Department of Chemistry, Trent University, Peterborough, ON, Canada
avreugdenhil@trentu.ca
Abstract. Naphthenic acids (NAs) are a broad classification of organic acid contami-
nants commonly found in processed waters from oil extractions in Alberta Canada and
thought to be predominantly responsible for the high toxicity impeding the release of oil
sands process affected waters (OSPW) [1]. The oil reserves, present in the form of oil
sands, in Alberta Canada are one of the largest in the world, estimated to contain 169 bil-
lion barrels of recoverable bitumen. The extraction process of bitumen involves the use of
large volumes of caustic hot water that even with recycling efforts in place, has continued
to accumulate for decades in large settling basins, known as tailing ponds [2]. As of a 2019
report given by the Alberta Energy Regulator, there is an estimated 1302 Mm3 of fluid
tailings being stored in tailing ponds [3]. There is a clear need to find an effective remedia-
tion strategy for OSPW, since oil extraction is only projected to continue to increase over
the coming decade. The use of a commercially viable, petcoke-sourced activated carbon
(PAC) is a promising option for removing the toxic components in OSPW.
Keywords: Naphthenic acids · Isotherms · Kinetics · Modeling · Adsorption ·

Activated carbon
Naphthenic acids (NAs) are a broad classification of organic acid contaminants com-
monly found in processed waters from oil extractions in Alberta Canada and thought to
be predominantly responsible for the high toxicity impeding the release of oil sands
process affected waters (OSPW) [1]. The oil reserves, present in the form of oil sands,
in Alberta Canada are one of the largest in the world, estimated to contain 169 billion
barrels of recoverable bitumen. The extraction process of bitumen involves the use of
large volumes of caustic hot water that even with recycling efforts in place, has contin-
ued to accumulate for decades in large settling basins, known as tailing ponds [2]. As of
a 2019 report given by the Alberta Energy Regulator, there is an estimated 1302 Mm3
of fluid tailings being stored in tailing ponds [3]. There is a clear need to find an effec-
tive remediation strategy for OSPW, since oil extraction is only projected to continue to
increase over the coming decade. The use of a commercially viable, petcoke-sourced
activated carbon (PAC) is a promising option for removing the toxic components in
OSPW.
https://doi.org/10.1007/978-3-031-17425-4_38
266 T. M. Roy and A. J. Vreugdenhil
The optimization of PAC has required a detailed characterization of the adsorption

system containing NAs, including an understanding of the adsorption mechanism.
Here, we present an investigation of three model NA species: diphenyl acetic acid,
cyclohexane acetic acid, and heptanoic acid. To elucidate the mechanism of adsorption,
a thermodynamic approach is planned, which involves the study of aqueous adsorption
isotherms across a temperature range to determine energies of adsorption for each
model NA. The van’t Hoff equation, relating the equilibrium constant Kc, to the thermo-
dynamic properties ∆S and ∆H for adsorption is as follows:
H 1 S
ln K c (1)
R T R
Kc is approximated through isotherm modeling of adsorption isotherms conducted at

various temperatures. To ensure Kc remains dimensionless, in accordance with Eq. (1),
a conversion factor highlighted in Tran et al. can be used [4]. A linear plot based on Eq.
(1) can then used to determine the thermodynamic properties of each model NA adsorp-
tion system. However, determining the duration of time necessary to reach adsorption
equilibrium is required for accurate isotherm experiments. An example of preliminary
adsorption kinetic results at different temperatures is shown in Fig. 1 for heptanoic acid,
which demonstrates variability in the time necessary to reach equilibrium. A notable
Fig. 1. Complete adsorption kinetics of heptanoic acid on PAC at different temperatures. Initial
adsorption kinetics are further shown up to 2 hours.
Study of Adsorption Thermodynamics of Model Naphthenic Acids for Determining… 267
increase in initial adsorption kinetics can be observed as temperature increases, which

is consistent with that seen in other aqueous adsorption systems [5].
In addition to the thermodynamic property approximations that will be made through
isotherm modeling, kinetic modeling of adsorption data will also be explored to further
characterize the adsorption of model NAs, which will involve the use of the phenome-
nological mass transfer models for internal (IMT) and external (EMT) diffusion to
determine the rate limiting step at various temperatures [6]. A detailed understanding of
how the adsorption of these contaminants occurs within PAC will help us further
improve the use of activated carbon for remediating OSPW.
References
1. Brown LD, Ulrich AC. Oil sands naphthenic acids: a review of properties, measure-
ment, and treatment. Chemosphere. Elsevier Ltd. 2015:276–90. https://doi.org/10.1016/j.
chemosphere.2015.02.003.
2. Allen EW. Process water treatment in Canada’s oil sands industry: II. A review of emerging tech-
nologies. J Environ Eng Sci. ICE Publishing. 2008;7:499–524. https://doi.org/10.1139/S08-020.
3. Alberta Energy Regulator, A. State of fluid tailings management for mineable oil sands. 2019.
https://static.aer.ca/prd/documents/reports/2020-State-Fluid-Tailings-Management-Mineable-Oil-
Sands.pdf.
4. Tran HN, You SJ, Hosseini-Bandegharaei A, Chao HP. Mistakes and inconsistencies regarding
adsorption of contaminants from aqueous solutions: a critical review. Water Res. Elsevier Ltd;
2017. p. 88–116. https://doi.org/10.1016/j.watres.2017.04.014.
5. Marczewski AW, Seczkowska M, Deryło-Marczewska A, Blachnio M. Adsorption equilibrium and
kinetics of selected phenoxyacid pesticides on activated carbon: effect of temperature. Adsorption.
2016;22(4–6):777–90. https://doi.org/10.1007/s10450-016-9774-0.
6. Wang J, Guo X. Adsorption kinetic models: physical meanings, applications, and solving methods.
J Hazard Mater. Elsevier B.V. 2020. https://doi.org/10.1016/j.jhazmat.2020.122156.
Enhanced Strength and Electrical Conductivity
of the Al-Mg-Si Conductor Alloys by Adding Ag
and Applying the Preaging
Siamak Nikzad Khangholi1 (*), Mousa Javidani1, Alexandre Maltais2,

and X. Grant Chen1
1
Department of Applied Science, University of Québec at Chicoutimi, Saguenay, QC, Canada
siamak.nikzad-khangholi1@uqac.ca
2
Arvida Research and Development Center, Rio Tinto Aluminum, Saguenay, QC, Canada
Abstract. The behaviors of strength and electrical conductivity of the Al-Mg-Si con-
ductor alloys were studied by adding Ag and applying the preaging. In the conven-
tional thermomechanical treatment (CTMT), consisting of solution treatment and
cold wiredrawing followed by aging, the Ag addition is led to a slightly higher
strength for the Al-Mg-Si conductor alloys. However, modifying the thermomechani-
cal treatment with preaging at 120 °C for 24 h prior to cold wiredrawing (MTMT120)
resulted in higher strengths for both base and Ag-added samples above the minimum
required electrical conductivity (52.5% IACS). Compared the base sample with the
Ag-added sample, it was shown that the Ag addition to the Al-Mg-Si alloys under
MTMT120 exhibited the relatively higher increment of the strength relative to
CTMT. The transmission electron microscopy (TEM) results indicated that
MTMT120 increased the precipitate number densities by more than three times com-
pared to their CTMT counterparts. The preaging at 120 °C for 24 h was found more
effective than the preaging at 180 °C for 6 h for both base and Ag-added samples in
terms of the strength above the electrical conductivity of 52.5% IACS, resulting from
the higher strain hardening in MTMT120 during the cold wiredrawing. The Ag-added
samples under MTMT120 exhibited the highest strength among all samples owing to
the greater number densities of the strengthening precipitates and the relatively
higher strain hardening.
Keywords: Al-Mg-Si conductor alloys · Electrical conductivity · Strength · Ag

addition · Modified thermomechanical treatment
1 Introduction
The demand for the aluminum conductor alloys in the high-voltage power transmission
line has increasingly grown owing to their low density, excellent ultimate tensile
strength (UTS), and high electrical conductivity (EC) in the recent decades [1, 2]. The
Al-Mg-Si conductor alloys (such as AA6101 and AA6201) outperform the other alumi-
num alloys in terms of the strength while fulfilling the minimum required EC (52.5%
IACS) [2, 3]. The precipitate strengthening, strain hardening, solute strengthening, and
https://doi.org/10.1007/978-3-031-17425-4_39
270 S. N. Khangholi et al.
grain boundary hardening are among the main strengthening mechanisms in aluminum
alloys. On the other hand, the solute, precipitates, and the crystal defects (dislocation
and grain boundaries) increase the scattering of conductive electrons, causing the elec-
trical resistivity in aluminum alloys. Therefore, the factors improving the strength dete-
riorate EC. Thus, enhancing strength and EC simultaneously is often incompatible in Al
conductor alloys [2].
It is reported that the Ag additions (~0.5 wt.%) to Al-Mg-Si alloys might lead to an
increase in the peak strength. The Ag additions promote the formation of the clusters in
the early stage of the aging, which stems from the formation of Mg-Ag clusters due to
their strong affinity. Subsequently, the Mg-Ag clusters along with Mg-Si clusters evolve
into the higher number density of the strengthening precipitates. Besides, GP zones in
the Ag-added alloys more easily transform into β″ phases owing to their similarity in
size and composition [4]. Therefore, the Ag additions result in a higher peak strength in
the Al-Mg-Si alloys [5, 6]. However, the effect of the Ag additions on the behavior of
the strength and EC has been less studied in the Al-Mg-Si conductor alloys.
The conventional thermomechanical treatment (CTMT) is generally used to produce
the Al-Mg-Si conductor alloys, consisting of solution treatment and quenching, cold
wiredrawing, and aging [1, 7]. Nevertheless, it is reported that applying preaging prior
to wiredrawing might result in a better strength and EC, known as a modified thermo-
mechanical treatment (MTMT). Most of the modified thermomechanical treatments
were performed with preaging at 180 °C for 2–6 h [8–11]. In our recent work [11],
MTMT (preaging at 180 °C for 6 h) resulted in a better trade-off between strength and
EC for the base and Ag-added alloys compared to their CTMT counterparts. Applying
the preaging at 180 °C might lead to the material with a higher strength. For instance,
applying the preaging at 180 °C for 6 h led an increase in the hardness from 69 HV to
107 HV for the Al-0.65 Mg-0.4Si alloy and from 71.4 HV to 117 HV for the
Al-0.65 Mg-0.4Si-0.5Ag alloy [11]. The high-strength stock could cause a difficulty for
the wiredrawing. Therefore, preaging at a lower temperature could be taken into account
to minimize the processing problems during the wiredrawing. Furthermore, it is
intended to acquire a better compromise between the strength and electrical conductiv-
ity with the preaging at a lower temperature relative to the preaging at 180 °C.
The present work aimed at studying the effect of Ag on the behavior of the strength
and electrical conductivity using the modified thermomechanical treatment. The pre-
cipitates were quantitatively studied using the transmission electron microscopy (TEM).
2 Experimental
Table 1 displays the chemical compositions of the studied alloys. After preparing the
molten metal with the given chemical compositions, the melt was poured into a pre-
heated permanent steel mold (250 ± 5 °C) to produce the ingots with dimensions of
30 mm × 40 mm × 80 mm. Then, the cast ingots were homogenized at 560 °C for 6 h
and hot rolled at 350–480 °C with an area reduction of 70%. The hot-rolled materials
were cut and machined into the square bars (7 mm × 7 mm). Afterward, the materials
were subjected to the different thermomechanical treatments. Details of these
Enhanced Strength and Electrical Conductivity of the Al-Mg-Si Conductor Alloys… 271
Table 1. The chemical compositions of the experimental alloys.

Alloys Mg Si Ag Fe Ti B Al
Base 0.67 0.43 Trace 0.09 0.016 0.0022 Bal.
Base + Ag 0.63 0.38 0.53 0.11 0.012 0.0030 Bal.
Fig. 1. A scheme of (a) conventional thermomechanical treatment (CTMT) [11] (b) modified thermo-
mechanical treatment with preaging at 120 °C for 24 h (MTMT120) (c) modified thermomechanical
treatment with preaging at 180 °C for 6 h (MTMT180) [11].
thermomechanical treatments are schematically shown in Fig. 1. The conventional ther-

momechanical treatment (CTMT) consists of solution heat treatment at 540 °C for 2 h,
quenching, cold wiredrawing (70% reduction), and aging at 180 °C for different aging
times (Fig. 1a). However, the thermomechanical treatment was modified with applying
the preaging at 120 °C for 24 h prior to the cold wiredrawing (MTMT120) (Fig. 1b).
Figure 1c shows the modified thermomechanical treatment in our previous work with
the preaging at 180 °C for 6 h (peak aging) prior to the cold wiredrawing (MTMT180)
for the comparison purpose [11]. The wire with a diameter of 4.37 mm was obtained
after the wiredrawing.
The mechanical properties were evaluated using the hardness and tensile tests. The
hardness values were measured at least eight times with a load of 25 g and a dwell time
of 20 s. The tensile tests were carried out three times for each condition according to
EN50183 standard. The eddy current technique was used to measure EC according to
ASTM1004.
The TEM specimens were ground and then electropolished using electrolyte consist-
ing of 30% HNO3 in methanol at a temperature between −20 and −25 °C. The precipi-
tates were characterized with TEM in zone axis of <001>Al. The quantitative analysis
were described in detailed in [10, 12].
3.1 Mechanical and Electrical Properties
The hardness values for the base and base + Ag samples just after preaging at 120 °C
for 24 h were 80 and 109 HV, respectively, giving the lower hardness values compared
to the preaging at 180 °C for 6 h for the base (107 HV) and base + Ag (117 HV) sam-
ples. Therefore, it indicates that the preaging at 120 °C for 24 h (under aging) is led to
softer materials compared to the preaging at 180 °C for 6 h (peak aging), which is
appropriate for the following cold wiredrawing.
The evolutions of the hardness and EC over the aging times are plotted after wire-
drawing during the postaging at 180 °C in Fig. 2. The CTMT samples reached the peak
hardness after 1 h (Fig. 2a), while the maximum hardness for the MTMT120 samples
was acquired after 0.5 h aging. All samples were subjected to overaging with increasing
the aging time up to 24 h. In CTMT, the base + Ag sample showed a slightly higher
peak hardness (112 HV) relative to the base sample (108 HV), indicating the low impact
of Ag addition on the strengthening. However, in MTMT120, the base + Ag sample
exhibited a relatively higher hardness (133 HV) compared to the base sample (122 HV),
indicating that MTMT120 maximized the Ag effect on the strengthening. It is worth
mentioning that all samples under MTMT120 outperformed those using CTMT in
terms of the peak hardness. Figure 2c, d displays the EC variation over the aging time
for the samples under CTMT and MTMT120 after wiredrawing during the postaging at
180 °C. The EC rapidly raised in the early steps of aging process (up to 4 h). Afterward,
it was gradually improved (up to 24 h) in all samples. The samples under CTMT showed
a similar EC behavior over the aging time relative to their counterparts under MTMT120.
In general, the Ag-added sample (base + Ag) showed a lower EC level compared to the
base sample both in CTMT and in MTMT120. According to the EN50183 standard, the
minimum required EC ought to be 52.5% IACS [3]. The base and base + Ag samples
exceeded the minimum required EC after 2 h and 4 h aging using CTMT, and they met
52.5% IACS after 3 h and 5 h aging under MTMT120, respectively.
The plot of the hardness vs. EC is shown in Fig. 3a for all samples at all aging times.
Taking into account the minimum required EC (the dashed black line in Fig. 3a), the
base + Ag sample under MTMT120 showed the highest hardness up to 53.5% IACS
among all samples. The MTMT120 base sample has the second highest hardness above
52.5% IACS. Under CTMT, the base + Ag sample shows a slightly higher hardness
compared to the base sample (Fig. 3a).
Figure 3b shows the ultimate tensile strength for the samples at EC of 53% IACS. By
adding Ag to the base sample under CTMT, the strength was slightly improved from
295 to 305 MPa. However, under MTMT120, the strength was enhanced from 329 to
Fig. 2. The microhardness evolution of the samples as a function of the aging time using (a) CTMT
(b) MTMT120. The EC variation of samples over the aging time using (c) CTMT (d) MTMT120.
Fig. 3. (a) The plot of the microhardness versus the electrical conductivity for all samples. (b) The
strength values for the samples at EC of ~53%IACS. The strength results of MTMT180 come from
[11].
349 MPa with the Ag addition (Fig. 3b). Compared the strengths in MTMT120 for this
study with those in MTMT180 from our previous work [11], it is apparent that the
strengths under MTMT120 are higher than those in MTMT180 for the base and
base + Ag samples, indicating that MTMT120 is more efficient than MTMT180 in
terms of strength above EC of 52.5% IACS. Therefore, the base + Ag5 sample under
MTMT120 showed the highest strength among all samples at 53% IACS.
3.2 Microstructure Characterization
Figure 4a–d displays the TEM micrographs for the samples reaching EC of 53%
IACS. Figure 4e, f shows the precipitate number densities and precipitate size. All TEM
images were acquired in <001> Al zone axis. It should be noted the precipitate number
densities and size for the samples under MTMT180 obtained from the previous work
[11] were brought to compare with those under MTMT120 and CTMT. The Ag addi-
tion to the base sample moderately increased the precipitate number density and
reduced the length under CTMT (Fig. 4e, f), resulting in the better strength in the
base + Ag sample relative to the base sample. The predominant precipitates are lath-like
precipitates in the CTMT sample, shown as A with the white arrow (Fig. 4a, b). In
CTMT, the dense dislocations act as the favorable nucleation sites for the lath-like pre-
cipitates [10, 13].
Modifying the thermomechanical treatment (MTMT120) increased the precipitate
number densities by more than three times for both base and base + Ag samples com-
pared to their CTMT counterparts. In addition to the lath-like precipitates (arrow A), the
MTMT120 samples revealed the fine precipitates shown as B with the white arrows in
Fig. 4c, d. The higher precipitate number densities in MTMT120 compared to CTMT
might be attributed to these numerous fine precipitates, resulting in a higher strength at
EC of 53% IACS (Fig. 3b). As aforementioned, the large lath-like precipitates (arrow
A) are formed along the dislocations. However, it is speculated that the fine precipitates
(arrow B) were evolved from the pre-existing precipitates formed during the preaging.
Therefore, it is revealed that the preaging maximized the precipitate strengthening,
which is consistent with Ref. [10]. The samples under MTMT180 exhibited similar
precipitate number densities relative to the samples under MTMT120. In all thermome-
chanical treatment, the Ag-added samples showed higher precipitate number densities
relative to the base samples. It is reported that Mg atoms have stronger interaction
energy with Ag atoms compared to Mg atoms with Si atoms, meaning that Mg-Ag
clusters could be formed in the early stage of the aging. Subsequently, theses Mg-Ag
clusters along with Mg-Si clusters act as nucleation sites for the precipitation and
evolve into the copious subsequent precipitates in the Ag-added samples [4, 6, 14]. The
average precipitate lengths in the samples in MTMT120 are smaller than their counter
parts under CTMT, more likely attributed to the presence of the finer precipitates in
MTMT120. It is worth mentioning that the precipitates in MTMT180 are shorter than
in CTMT and MTMT120. It was reported that the β” precipitates after 5 h aging at
180 °C were sheared during the wiredrawing, ended up with the significantly shorter
precipitates [10, 11].
It is worthwhile to state that the Ag additions more likely cause a negligible solute
strengthening and solute resistivity, as the atom radius for Al (1.43 A°) and Ag (1.44 A°)
are similar [15]. Therefore, the lower EC in the Ag-added samples (base + Ag) relative
to the base sample (Fig. 2c, d) either in CTMT or in MTMT120 at a given aging time
could be attributed to the higher electrical resistivity of the precipitates with a high
number density in the Ag-added samples.
Fig. 4. TEM image of the (a) base sample after 2 h aging under CTMT (b) base + Ag sample after 4 h
aging under CTMT (c) base sample after 3 h aging under MTMT120 (d) base + Ag sample after 5 h
aging under MTMT120; (e) Precipitate number densities and (f) the precipitate lengths for all samples
under CTMT, MTMT120, and MTMT180 [11]. A and B correspond to the cross-sectional projections
of the lath-like and fine precipitates in <002>Al zone axis, respectively.
3.3 Strain Hardening via Cold Wiredrawing
The increased hardness values during the cold wiredrawing are given in Fig. 5. The
increased hardness was acquired by subtracting the hardness value immediately after
wiredrawing from that before wiredrawing. Given that the aging at 180 °C causes a
slight softening recovery [16], the increased hardness in the cold wiredrawing could be
correlated to the strain-hardening contribution in the samples. The results in MTMT180
were brought for the comparison purpose [11] with CTMT and MTMT120. The
Fig. 5. The increased hardness by wiredrawing in the base and base + Ag samples under CTMT,
MTMT180 [11], and MTMT120.
increased hardness for the base samples under CTMT, MTMT180, and MTMT120
were 26, 10, and 25 HV, respectively, and the increased hardness for the base + Ag
samples under CTMT, MTMT180, and MTMT120 were 26, 13, and 22 HV. It indicates
that the strain-hardening contributions in MTMT180 are considerably lower than
CTMT for both base and base + Ag samples. It means that the solutes are more efficient
on the work hardening than the shearable precipitates owing to the drag effects of sol-
utes on the dislocation movement, which is consistent with Ref. [10, 17, 18]. However,
the strain hardenings under MTMT120 are slightly lower than CTMT. The higher
strengths in MTMT180 and MTMT120 relative to CTMT (Fig. 3b) could be elaborated
with the fact that the superiority in precipitate strengthening overcame the inferiority in
strain hardening in MTMT120 and MTMT180 relative to CTMT. The higher strain
hardening in MTMT120 compared to MTMT180 might be related to more solutes
remained in the matrix in MTMT120 because the samples under MTMT120 were in the
underaging condition. Given that the precipitate number densities are similar in
MTMT120 and MTMT180, the higher strength in MTMT120 could be ascribed to the
higher strain hardening in MTMT120 relative to MTMT180.
3.4 Comparison of Achievable Strength and EC with EN50183 Standard
Figure 6 gives a comparison of the strength/EC in this work with the values in EN50183
standard (AL2-AL6) for Al–Mg–Si conductor alloys above the minimum required
EC. It is apparent that the CTMT samples (CTMT base and CTMT base + Ag) exhib-
ited similar results as EN50183 standard (AL2-AL4). However, the base sample under
MTMT120 showed a higher strength compared to all CTMT samples and exceeded the
values of AL2 and AL4 in the standard. The Ag addition (base + Ag sample) further
increased the strength relative to the base sample under MTMT120. Furthermore, the
Fig. 6. Comparison between the strength/EC of the studied alloys and those in EN50183 standard
(AL2-AL6) for the wires with a diameter of 4.37 mm-.
Ag-added sample under MTMT180 showed a higher strength than those under
CTMT. The base + Ag samples under MTMT120 showed the highest strength among
all samples above EC of 52.5% IACS.
4 Conclusions
1. The Ag-added samples showed a slightly higher peak strength relative to the base
samples under CTMT.
2. The preaging in MTMT120 resulted in a higher strength for the base sample com-
pared to its counterpart under CTMT above the minimum required EC. Besides, the
Ag addition further increased the strength under MTMT120.
3. The preaging promoted the precipitate number densities in MTMT relative to in
CTMT. The Ag-added sample under MTMT120 showed the highest precipitate
number density among all samples, which is consistent with the mechanical strength
results.
4. MTMT120 and MTMT180 exhibited similar precipitate number densities for both
base and base + Ag samples. However, MTMT120 exhibited a higher strain harden-
ing than MTMT180. The base + Ag samples under MTMT120 showed the highest
strength among all samples above EC of 52.5% IACS.
5. The optimized strength and EC in the base and base + Ag samples under MTMT120 in
this work outperformed the values in the EN50183 standard.
Acknowledgments The authors gratefully acknowledge the financial support of the Natural Sciences
and Engineering Research Council of Canada (NSERC) under the Grant No. CRDPJ 514651-17 and
Rio Tinto Aluminum through the Research Chair in the Metallurgy of Aluminum Transformation at
University of Quebec at Chicoutimi.
References
1. Wang Y, Zhu L, Niu G, Mao J. Conductive Al alloys: the contradiction between strength and elec-
trical conductivity. Adv Eng Mater. 2021;23(5):1–22.
2. Khangholi SN, Javidani M, Maltais A, Chen XG. Review on recent progress in Al–Mg–Si 6xxx
conductor alloys. J Mater Res. 2022:1–22.
3. EN 50183 standard. Conductors for overhead lines, aluminium magnesium silicon alloy wires. BSI
Standards Limited. 2000.
4. Weng Y, Jia Z, Ding L, Muraishi S, Liu Q. Clustering behavior during natural aging and artificial
aging in Al-Mg-Si alloys with different Ag and Cu addition. Mater Sci Eng A. 2018;732:273–83.
5. Kim J, Daniel Marioara C, Holmestad R, Kobayashi E, Satom T. Effects of Cu and Ag addi-
tions on age-hardening behavior during multi-step aging in Al-Mg-Si alloys. Mater Sci Eng
A. 2013;560:154–62.
6. Weng Y, Jia Z, Ding L, Pan Y, Liu Y, Liu Q. Effect of Ag and Cu additions on natural aging and
precipitation hardening behavior in Al-Mg-Si alloys. J Alloys Compd. 2017;695:2444–52.
7. Karabay S. Modification of AA-6201 alloy for manufacturing of high conductivity and extra high
conductivity wires with property of high tensile stress after artificial aging heat treatment for all-
aluminium alloy conductors. Mater Des. 2006;27(10):821–32.
8. Lin G, Zhang Z, Wang H, Zhou K, Wei Y. Enhanced strength and electrical conductivity of Al–Mg–
Si alloy by thermo-mechanical treatment. Mater Sci Eng A. 2016;650:210–7.
9. Liu CH, Chen J, Lai YX, Zhu DH, Gu Y, Chen JH. Enhancing electrical conductivity and strength
in Al alloys by modification of conventional thermo-mechanical process. Mater Des. 2015;87:1–5.
10. Khangholi SN, Javidani M, Maltais A, Chen XG. Effects of natural aging and pre-aging on
the strength and electrical conductivity in Al-Mg-Si AA6201 conductor alloys. Mater Sci Eng
A. 2021;820:1–12.
11. Khangholi SN, Javidani M, Maltais A, Chen XG. Effect of Cu and Ag on the strength and electrical
conductivity in Al-Mg-Si cable using conventional and modified thermo-mechanical treatment. J
Alloys Compd. 2022.
12. Engler O, Marioara CD, Aruga Y, Kozuka M, Myhr OR. Effect of natural ageing or pre-ageing on
the evolution of precipitate structure and strength during age hardening of Al–Mg–Si alloy AA
6016. Mater Sci Eng A. 2019;759:520–9.
13. Teichmann K, Marioara CD, Andersen SJ, Pedersen KO, Gulbrandsen-Dahl S, Kolar M, Holmestad
R, Marthinsen K. HRTEM study of the effect of deformation on the early precipitation behaviour
in an AA6060 Al–Mg–Si alloy. Philos Mag. 2011;91(28):3744–54.
14. Weng Y, Jia Z, Ding L, Liu M, Wu X, Liu Q. Combined effect of pre-aging and Ag/Cu addi-
tion on the natural aging and bake hardening in Al-Mg-Si alloys. Prog Nat Sci: Mater Int.
2018;28(3):363–70.
15. Porter DA, Easterling KE. Phase transformations in metals and alloys. 3rd ed (Revised reprint).
Milton Park: Taylor & Francis; 1992.
16. Nikzad Khangholi S, Javidani M, Maltais A, Chen XG. Optimization of mechanical properties
and electrical conductivity in Al–Mg–Si 6201 alloys with different Mg/Si ratios. J Mater Res.
2020;35(20):2765–76.
17. Cheng LM, Poole WJ, Embury JD, Lloyd DJ. The influence of precipitation on the work-hardening
behavior of the aluminum alloys AA6111 and AA7030. Metall Mater Trans A. 2003;34(11):2473–81.
18. Li H, Qingzhong M, Wang Z, Miao F, Fang B, Song R, Zheng Z. Simultaneously enhancing the
tensile properties and intergranular corrosion resistance of Al–Mg–Si–Cu alloys by a thermo-
mechanical treatment. Mater Sci Eng A. 2014;617:165–74.
Economic Analysis of Electrolytic Treatment of Waste
Cyanide
Jacob Schmidt1 and Eduard Guerra1 (*)
1
Laurentian University, Sudbury, ON, Canada
eguerra@laurentian.ca
Abstract. An economic analysis of the viability of electrolytic oxidation of cyanide on

304 stainless-steel electrodes with parallel plate geometry for waste gold effluent versus
the SO2–air process is presented. Both total replacement of the INCO SO2–air process and
applying electrolysis prior to applying INCO SO2–air are studied. The analysis was per-
formed for an applied current density of 20 A/m2 in the presence of copper ions at pH 10,
which previous experiments establish as a likely in the mass transfer limiting regime for
cyanide oxidation and reasonably fit using a pseudo-first-order decay model. The optimal
electrolysis plant size is estimated through comparing the internal rate of return, IRR, over
20 year projects and payback period of required estimations of capital costs. Total replace-
ment of INCO SO2-air is found to be economically unviable, while partial replacement is
found to be potentially viable, despite with a large payback period.
Keywords: Waste treatment · Electrolysis · Electrodes · Stainless steel · Cyanide
1 Introduction
As of 2002, cyanidation was used for 90% of primary gold extraction, with physical
separation being used for the remaining 10% [1]. The waste tailings generated from
cyanide leaching were historically discharged into shallow tailing ponds where cyanide
concentration diminished over time by a combination of natural volatilization and oxi-
dation by reaction with dissolved oxygen from the air, catalyzed by ultraviolet light
from the sun. Major spills from these ponds have occurred in several instances, includ-
ing an incident in South Africa in 1994 leaving 17 dead, and an instance in New Zealand
in 1995 contaminating the water table [2]. As a result of these risks, Canadian regula-
tions require that mill effluent contain at most an average of 1 ppm of cyanide per
month and never exceed 2 ppm cyanide [3].
1.1 Chemical Cyanide Destruction
Waste cyanide-bearing effluent is typically treated with chemical reagents. The INCO
SO2/Air process is by far the most widely used for treatment of industrial gold cyanide
waste streams. The INCO SO2/Air process introduces sulfite into solution, either
directly through dissolution of sulfite salts or by introducing sulfur dioxide liquid which
https://doi.org/10.1007/978-3-031-17425-4_40
280 J. Schmidt and E. Guerra
reacts to form sulfite and acid [4]. Oxygen is then introduced to solution through forced
air dissolution. Sulfite and oxygen react with cyanide to form cyanate and sulfate.
The primary cost for this method is the SO2 source. In practice, SO2 is added to an
excess between 22% and 103% [5]. This reaction is carried out in mildly basic condi-
tions. Typically, lime is added in conjunction with SO2 to keep the pH at an appropriate
level. A small amount of copper ions in solution is necessary to achieve acceptable rates
of cyanide oxidation [6]. Copper is typically added to solution by dissolution of copper
sulfate.
1.2 Electrochemical Cyanide Destruction
Kinetics Electrolytic cyanide oxidation of cyanide in the types of dilute electrolytes

typical of gold effluents is known be mass transfer limited on stationary electrodes.
Furthermore, it has been shown that during electrolysis of dilute cyanide solutions, at
sufficient current density where oxygen gas evolution is used to enhance cyanide mass
transport to the electrode, cyanide is oxidized under a pseudo-1st-order regime [7, 8].
dCCN t
kCN CCN t (1)
dt
In this regard, it is likely that cyanide oxidation is mass transfer limited and that the
value of the rate constant is affected by the nature of oxygen gas bubble evolution on
the electrode surface and the concomitant micro-convective transport of cyanide to the
electrode surface.
Using electrochemical treatment in lieu of chemical treatment presents a potentially
viable tradeoff in paying for power costs rather than chemical reagents. In regions
where power is relatively cheap, electrochemical treatment of cyanide may be more
cost effective than chemical treatments like the INCO SO2/Air process. This work aims
to characterize the cost efficiency of using electrolysis to oxidize cyanide in lieu of SO2/
Air treatment. In doing so, a general methodology for evaluating the economics of
electrolyte waste treatment processes that can potentially be applied for other systems,
is described.
2 Experimental
Electrolytic cells were built using 400 mL beakers containing 350 mL of electrolyte and
topped with a PVC lid in which slots had been milled in order to hold the electrodes
parallel and at a uniform distance of 6.5 cm apart. The electrolyte contained 471 mg/L
CN−, prepared by dissolving 1.13 g of NaCN (ACS, Anachemia) in 10 mM pH 10 car-
bonate buffer solution. When copper was added to the electrolyte, it was done so at a
concentration of 10 mg/L Cu2+, by addition of 1 mL spike of a concentrated copper
sulfate solution. The anode and cathode were both manufactured from 4″ × 1/8″ 304
stainless-steel plate (McMaster-Carr); the anode was either bare or coated with MnO2.
Economic Analysis of Electrolytic Treatment of Waste Cyanide 281
The cathode area was 4 cm × 6.5 cm; the anodes had the same dimensions, but the
edges, back, and top of the anode were covered using electroplating tape (470, 3 M),
restricting the exposed area to 4 cm × 5 cm. This was done to keep the exposed area
consistent for all anodes, as the actual construction of the electrodes had some small
variation in dimension.
Two sets of galvanostatic experiments were performed: one which explored various
current densities over a fixed period, and another set that explored the effect of time at
a fixed current density. The motivation for first set of experiments was to ascertain
which applied currents would be most effective for cyanide destruction; the logic is that
once cyanide destruction reaches the mass transfer limiting rate, larger currents are a
waste of power, as those currents will only contribute to oxygen evolution instead of
cyanide destruction. The second set of experiments was undertaken to explore the
kinetics of cyanide oxidation in order to extract kinetic parameters that could be used to
design an industrial electrolytic cyanide oxidation process. The effect of dissolved cop-
per as a catalyst for cyanide oxidation was also explored.
A DC power source (Heathkit, IP-28) was used to apply a variety of current densities
between 1 and 6 mA/cm2. The currents were applied for either 6, 12, or 24 hours. To
keep the test conditions consistent and perform multiple experiments simultaneously,
the cells were arranged in series. Voltages across each the cells were recorded at a
sample rate of 50 Hz using two USB-6009 National Instruments DAQ cards using dif-
ferential analog inputs. To make agitation consistent across cells, up to six cells at a
time were arranged on a multiplate stirrer (Labline Multi Magnestir) to be moderately
agitated using PTFE-coated magnetic stir beads (Fig. 1).
The residual cyanide concentration was measured using titration with an AgNO3
standard according to the procedure described by Breuer et al. [9]. AgNO3 titrants were
Fig. 1. Four galvanostatic cells in series.

prepared by mixing 4.12 g of AgNO3 in 200 mL of ultrapure water (Milli-Q Direct 16

Water Purification System, Millipore).
2.1 Results
An applied current density of 2 mA/cm2 had the greatest effect on lowering residual
cyanide concentration after 24 hours. Based on these results, further experiments were
conducted at 2 mA/cm2. Furthermore, the addition of a copper to the electrolyte was
found to significantly enhance the kinetics of cyanide oxidation.
Experiments at a current density of 2 mA/cm2 were done for varying time lengths; 6,
12, or 24 hours. The electrolyte pH was measured before and after electrolysis and
showed no significant drift. The residual cyanide concentration was fitted to an expo-
nential decay model, Eq. (1), corresponding to first-order mass transfer-limited kinetics
(see Fig. 2). The resulting ‘k’ value, from a least-squared fit of the results, was found to
be 1.406 cm/hour, with corresponding average cell voltages and overall current effi-
ciency values of 4.43 V and 8.1%, respectively. Note that the calculated current effi-
ciency is on the basis of a two electron process of oxidation of CN− to CNO−.
A
kt
CCN t CCN

e
V
(2)
3 Flow Cell Design
The results of the galvanostatic experiments were used to inform the design an electro-
lytic cyanide destruction cell and evaluate its feasibility for industrial application.
Figure 3 shows a thin slice of a mass transfer-limited flow cell, where v is the volumet-
ric flow rate, and A∗ is the anodic area of the slice. Equation (4) shows the solution to
the resulting differential equation where w is the spacing between the cathode and
anode, V is the volume of the flow cell, A is the total anode area in the cell, L is the total
cell length, and τ is the mean residence time. It is useful to measure the flow cell size in
number of “unit cells,” N, of a specific area, A .
dC x kC x h
(3)
dx v
kA kA k
C L C e v
C e V
Ce w
(4)
kNA
C N C e v
(5)
Fig. 2. Ratio of residual cyanide concentration as a function of time from at a current density of 2 A
cm−2 in cells equipped with bare 304 stainless anodes in 10 mM carbonate pH 10 buffer solution con-
taining 10 mg L−1 copper.
Fig. 3. Thin section of a mass transfer-limited electrolytic flow cell.
3.1 Power Efficiency
The current efficiency of the cell is given by Eq. (6) where i is the current density, while
Eq. (7) follows from Eq. (5):
i
(6)
itot

kNA
i N nFr N nFkC N nFkC e v
(7)
Using the first-order kinetics model, the current efficiency can be established over the
range of the electrolysis. The average current efficiency can be found through the mean
value theorem. This is equivalent to the ratio of the total charge transfer across the
anode to charge transfer associated with cyanide oxidation. Immediately apparent is the
drop in average current efficiency, the longer the cell becomes. The current efficiency
will be greater at earlier stages of electrolysis due to greater transport of cyanide to the
anode surface.
kNA
1
L
1
N
nFk

Li tot 0
idx
Ni tot 0
idN
Ni tot
C 1 e v

(8)

nFk 
lim C
N 0 itot
lim 0
N
3.2 Reactor Sizing and Power Costs
The required size of the reactor, and associated capital and operating costs were
estimated.
Capital Costs for the Electrolytic Flow Cell Since cyanide electrolysis is not cur-
rently done in industry, data for capital costs do not exist. However, copper-
electrowinning cells feature similar characteristics to the electrolysis cell described
above, using similar materials of construction (steel-based electrodes), and operating at
similar temperatures, flow rates, and voltages. For these reasons, capital cost estima-
tions for a copper-electrowinning plant were used as the basis for capital cost of similar
sized cyanide electrolysis plant. The major differences are the current densities and
electrolyte pH. Copper electrowinning is done under acidic conditions and roughly 300
A/m2, while the cyanide electrolysis described herein is done at 20 A/m2 in alkaline
conditions. Additionally, a cyanide electrolysis plant may require more safety consider-
ations, in particular more aggressive ventilation to mitigate potential cyanide off gases.
The capital cost estimations should, therefore, be considered optimistic. The capital
cost estimation is given in Eq. (9) in USD [10] and is valid between 14 and 975 unit
cells with an area of 30 m2.
kNA
1
L
1
N
nFk

Li tot 0
idx
Ni tot 0
idN
Ni tot
C 1 e v

(9)

379000
0.971
A = 30 m 2
Operating Costs The primary operating costs of an electrowinning cell are the power
requirements, replacement of electrodes, and replenishment of reagents. In the case of
electrolysis of cyanide, there are no reagents to replenish, and electrodes would need to
be replaced at a considerably lower rate. The primary operating cost of concern is,
therefore, the power costs. For this report, a power cost of 0.085 USD/kWh will be
assumed, based on the average global adjustment price reported by the Independent
Electricity System Operator (IESO) of Ontario in 2019. The operating power cost can
then be described by Eq. (10) where P is the power cost factor and E is the voltage.
 NE
Power cost = Pitot A (10)
The average voltage required to drive the 20 A/m2 current in the cells described in Sect.
2 was 4.5 V. However, a portion of the voltage is unrelated to the chemical reactions and
simply due to electrolyte resistance.
E Eanode Ecathode IR (11)
The voltage caused by can be mitigated by minimizing the gap between electrodes.
The electrolyte resistance between two electrodes follows Eq. (12), where σ is the elec-
trolyte specific conductivity.
l
IR itot (12)

The specific conductivity for the cyanide-bearing carbonate electrolyte used is 0.0054 S/
cm [11]. Assuming that the pH 10 cyanide-bearing carbonate solution has a comparable
conductivity to that of the waste effluent stream in industry, which is primarily cyanide
and NaOH, at an electrode spacing of 6.5 cm and a current density of 0.002 A/cm2, the
IR voltage is estimated to be 2.4 V. Hence, the required voltage required to drive the
reactions, Eanode − Ecathode, is only ca. 2.1 V (4.5 V − 2.4 V). The overall voltage of the
cell can be significantly reduced by reducing the electrode gap. For example, an elec-
trode gap of 2 cm would be predicted to result in an overall cell voltage of 2.8 V, based
on a corresponding estimated IR drop of 0.7 V.
Flow Cell Size for Total Conversion To meet environmental regulation for residual
cyanide, the reaction needs to proceed until a final concentration of 1 ppm has been
reached, as specified by Canadian federal regulation [3]. The corresponding required
number of reactors, N, can be determined by Eq. (13).
v 1ppm
N ln (13)
kA C
v 1ppm
A ln
k C 
Table 1 shows the calculated required area and number of unit cells of 30 m2 for flow-
rates between 5000 and 40,000 m3/day using a stainless-steel anodes.
On inspection, these values for number of reactors and electrode area are infeasible.
They are orders of magnitude greater than those of similar processes in industry [10].
However, this result is partly a result of the instantaneous current efficiency dropping
over the course of electrolysis. It is possible that this technology might be cost efficient
if it were to be applied as a first stage of cyanide oxidation followed by a conventional
chemical oxidation process as a polishing stage.
3.3 Electrolysis and INCO SO2/Air in Series
If the flow cell does not sufficiently destroy the cyanide, the remainder must be dealt
with using conventional chemical methods. In 2010, INCO reported they spent between
1.22 USD and 4.5 USD per kilogram of cyanide destroyed using their process, primar-
ily due to chemical costs and varying concentrations of cyanide. For this analysis, val-
ues of 1.22 USD/kg as a best case and 2.86 USD/kg as a median case will be considered,
where Q is the variable reagent cost factor:

Reagent Cost vCCN Q, without electrolysis (14)

Reagent Cost vCCN N Q, after N stages of electrolysis
v 1ppm
N ln (15)
kA C
After T years of operation for a capital cost R, combining Eqs. (9) and (15):

kAN
1 1 R T
NPV N ,T vC Q 1 e v NE
PiA N (16)
R

Table 1. Flow cell sizes required for regulation sufficient conversion of cyanide.
v (m3/day) v (m3/h) A(m2) N
5000 208 81,814 2727
10,000 417 163,628 5454
15,000 625 245,442 8181
20,000 833 327,256 10,909
25,000 1042 409,070 13,636
30,000 1250 490,884 16,363
35,000 1458 572,698 19,090
40,000 1667 654,512 21,817
Maximum Cost Savings The local optima of Eq. (16) can be found by setting the
derivative to 0.

kAN
kQC e Pitot E 0
v
(17)

Equations (6) and (7) can then be reintroduced into the above equation. Rearranging,
the instantaneous current efficiency at the point of greatest cost savings follows
Eq. (18).
PEnF
N (18)
Q
Intuitively, the target current efficiency after N cells can be understood as the ratio
between the cost of 1 mole of oxidizing reagents to the cost of 1 mole worth of charge
transfer across the anode. Notably, the target current efficiency is independent from
electrode area, flow rate, or initial concentration. Instead, those values will all impact
the number of unit cells required to achieve the target efficiency.
v kQC
N max ln (19)
kA Pitot E
Note that Nmax is only positive when kQC∘ > PitotE. This allows for a quick test to see
what conditions are worth considering (Table 2).
Under experimental conditions and theoretically ideal conditions, the best-case
operating cost for the INCO SO2/Air process is always more cost effective than electro-
lytic cyanide oxidation. However, electrolysis becomes even more efficient the greater
the initial cyanide concentration becomes. While this analysis assumes C∘ = 0.25 kg/m3,
INCO reported concentrations between 0.1 and 0.8 kg/m3, meaning that there are pos-
sible operating conditions for which early-stage electrolysis is cost effective in terms of
operating costs. Figure 4 shows the operating cost savings of using electrolysis in series
with INCO SO2/Air for N stages of electrolysis pretreatment. For very large flowrates
under theoretically optimal voltages, Nmax exceeds the validity of Eq. 19. At the median
operating cost, electrolysis in the early stages is always cost effective.
Maximizing IRR The internal rate of return is the rate of return at which the net pres-
ent value is 0 dollars. It is the maximum capital cost that can be assumed in order to
break even on a project. Maximizing the IRR is, therefore, ideal. Over an infinite times-
pan, the IRR could be taken as the ratio of the Capital Cost to the Cost Savings.
Realistically, however, the annuity factor in Eq. (16) decays exponentially from 0 to a
Table 2. Ratio of kQC∘/PitotE by voltage and reagent cost.
Reagent cost E = 4.5 V E = 2.8 V

Q = 1.22 USD/kg 0.56 0.9
Q = 2.86 USD/kg 1.31 2.11
E = 4.6V
100 = 5000m 3/day

= 10000m 3/day
= 20000m 3/day
USD/hr
= 40000m 3/day
50
0
0 100 200 300 400 500 600 700 800 900 1000
N
E = 2.8V
100
USD/hr
50
0
0 100 200 300 400 500 600 700 800 900 1000
N
Fig. 4. Savings per hour of electrolysis with the INCO SO2/Air process in series over just INCO SO2/
Air assuming an initial cyanide concentration of 250 ppm and an SO2/Air process operating cost of
2.86 USD/kg.
value of R−1. The lifetime of a project is unlikely to approach that limit of R−1 unless R
is very large, or the project lifetime is unrealistically long.
If Capital Cost was fixed or independent of N, Eq. (19) could be taken as the most
cost-effective flow cell size. However, if the growth in capital cost as flow cell size
grows is significant, as is true in most cases, Eq. (19) serves as an upper bound for a
financially feasible reactor size.
The maximum IRR occurs at the minimum payback period. The goal, then, is to
determine the number of cells N that minimizes the right side of Eq. (20):
1 1 R T N

(20)
R
kAN

vC Q 1 e v PiA
NE

Figure 5 shows the plots of the payback period for a flow cell with a 2 cm spacing.
On inspection, the system is not financially feasible, showing a payback period on the
order of hundreds of years. Table 3 also demonstrate that IRR is never positive for an a
typical mine life of 20 years. With a 6.5 cm spacing, the performance was even worse,
with payback periods exceeding 600 years.
Payback Period, E = 2.8V

400
3
= 5000m /day
380
= 10000m 3 /day
360 = 20000m 3 /day

3
= 40000m /day
340
320
years
300
280
260
240
220
200
0 50 100 150 200 250
N
Fig. 5. Payback period of electrolysis with the INCO SO2/air in series over just SO2/air assuming an
initial cyanide concentration of 250 ppm and an SO2/air operating cost of 2.86 USD/kg.
Table 3. Optimal number of electrolytic reactors with 2 cm electrode spacing and corresponding
payback period and IRR as a function of daily flow rate of 250 ppm cyanide effluent.
ν m3/day Optimal N Payback period IRR for 20 year project
5000 16 259 years −17.4%
10,000 32 254 years −17.3%
20,000 64 249 years −17.2%
40,000 127 244 years −17.1%
4 Conclusion
The results showed that electrolysis in series with the INCO SO2/Air process is cost
effective but only over unrealistically long time frames. Some approaches are worth
considering to bring the payback period down to a reasonable value. The major hin-
drance in financial feasibility is the capital cost as the flow cell requires a very large area
to be effective.
References
1. Yarar B. Long term persistence of cyanide species in mine waste environments. In: Tailings and
mine waste 2002: proceedings of the 9th international conference. 2002.
2. Eisler R, Wiemeyer SN. Cyanide hazards to plants and animals from gold mining and related water
issues. Rev Environ Contam Toxicol. 2004;183:21–54.
3. SCHEDULE 4: authorized limits of deleterious substances. 15 May 2021. [Online]. Available:
https://laws-l ois.justice.gc.ca/eng/regulations/SOR-2 002-2 22/section-s ched685007.html.
Accessed 17 May 2021.
4. Breuer P, Sutcliffe C, Meakin R. Comparison of industrial cyanide destruction processes. In: 25th
International mineral processing congress – IMPC2010, Brisbane, 2010.
5. Mudder TI, Botz MM, Smith A. Chemistry and treatment. London: Mining Journal Books
Ltd; 2006.
6. Cheng SC, Gatrell M, Guena T, MacDougal B. The electrochemical oxidation of alkaline copper
cyanide solutions. Electrochim Acta. 2002;47:3245–56.
7. Hwang J-Y, Wang Y-Y, Wan C-C. Electrolytic oxidation of cuprocyanide electroplating waste
waters under different pH conditions. J Appl Electrochem. 1987;17:684.
8. Lanza MR, Bertazzoli R. Cyanide oxidation from wastewater in a flow electrochemical reactor. Ind
Eng Chem Res. 2002;41:22.
9. Breuer PL, Sutcliffe CA, Meakin RL. Cyanide measurement by silver nitrate titration: comparison
of rhodanine and potentiometric end-points. Hydrometallurgy. 2010;106:135.
10. Stinn C, Allanore A. Estimating the capital costs of electrowinning processes. Electrochem Soc.
2020;29:44.
11. CRC handbook of chemistry and physics., 91th ed., W. M. Haynes (Ed.), CRC Press: Ohio, 2004.
Part V Light Metals for the Transportation
Industry
Titanium: From Ore Extraction and Processing to Its
Applications in the Transportation Industry
Carsten Siemers1 (*) and Fabian Haase1

1
Technische Universität Braunschweig, Institute for Materials Science, Braunschweig, Germany
c.siemers@tu-braunschweig.de
Abstract. Titanium alloys combine outstanding mechanical properties with excellent

corrosion resistance, making them desirable for challenging light-weight constructions in
the transportation industry. On the other hand, titanium production is energy consuming
and expensive since the current method of extraction of titanium from the ore includes
smelting, chlorination, and reduction by magnesium (Kroll’s process). Hence, cost reduc-
tion is one of the driving factors for titanium research to broaden its field of application.
In this overview paper, the basics of titanium metallurgy are first presented. Afterward, the
conventional production route from titanium ore to semi-finished products and compo-
nents is discussed. This includes ore winning and processing, melting and remelting for
ingot production, forging, casting, machining of (semi-finished) products, scrap recy-
cling, and additive manufacturing. In the end, a brief overview of titanium application in
the transportation industry is given.
Keywords: Titanium · Titanium alloys · Kroll’s process · Titanium ore processing ·

Melting · Forging · Casting · Machining · Additive manufacturing · Powder bed
fusion
1 Titanium’s Metallurgy
Titanium belongs to the allotropic metals and can exist in more than one equilibrium
lattice modification [1]. Pure titanium crystallizes at 1668 °C in a body-centered cubic
(bcc) structure called β-titanium which at 882 °C is transformed to a hexagonal close-
packed structure (hcp), the α-titanium, having anisotropic properties. In addition, at
very low temperatures and/or high pressures, α-titanium transforms to the ω-phase
(hexagonal). A martensitic β-to-α-transformation (called α′- or α″-phase depending on
the crystal structure) is possible at significant cooling rates. The transformation tem-
perature named β-transus temperature (Tβ) can be influenced by alloying elements [1].
For titanium alloy production, typical alloying elements are aluminum (Al) and oxy-
gen (O), both α-stabilizers shifting the β-transus temperature to higher temperatures,
whereas tin (Sn) and zirconium (Zr) only show a limited influence on the β-transus
temperature. The β-stabilizing elements are subdivided into two groups: Niobium (Nb),
molybdenum (Mo), tantalum (Ta), and vanadium (V) stabilize the β-phase between the
melting point and room temperature and are, therefore, called isomorphous β-stabilizers.
Elements like copper (Cu), iron (Fe), or silicon (Si) also stabilize the β-phase to lower
https://doi.org/10.1007/978-3-031-17425-4_41
294 C. Siemers and F. Haase
temperatures but undergo a eutectoid reaction during cooling. β-titanium then is disso-
ciated to α-titanium and an intermetallic compound. Consequently, these elements are
called eutectoid β-stabilizers [2].
Titanium alloy classification and related phase compositions are shown in the sche-
matic quasi-binary phase diagram of titanium with increasing amounts of isomorphous
β-stabilizers (see Fig. 1).
According to the phases present at room temperature, titanium alloys are divided into
different main groups (see Fig. 1, top line), starting with CP-Titanium (commercially
pure titanium, containing low amounts of oxygen, iron, nitrogen, and carbon only), con-
sisting of nearly 100% α-phase. α- and near-α-alloys are composed of α-phase and up to
5% of β-phase at room temperature, respectively. Near-β-alloys contain more than 95%
of β-phase and β-alloys are composed of almost 100% β-phase [1]. Finally, the two-
phase alloys consist of more than 5% and less than 95% β-phase at room temperature.
Typically, one or more α- and β-stabilizers are present in such alloys. They are divided
into two subgroups: In (α + β)-alloys, a martensitic transformation during quenching
from above β-transus is possible and depending of the exact chemical composition,
either the hexagonal α′- or the orthorhombic α′′-martensite can form. The microstructure
of (α + β)-alloys can, therefore, be refined and hardened by a (partial) decomposition of
martensite. In metastable β-alloys, the martensite start temperature is below room tem-
perature. Therefore, after water quenching, metastable β-alloys consist of supercooled
β-phase, and in the case of solute-lean metastable β-alloys, ω-phase might be present [3].
The ω-transus temperature of solute-rich metastable alloys, on the other hand, is below
room temperature so that such an alloy consists of a supercooled β-phase only after
Fig. 1. Schematic quasi-binary phase diagram of titanium and increasing amounts of isomorphous
β-stabilizers. The β-transus temperature, the different equilibrium phases and the two types of martens-
ite and their starting temperatures (MS) as well as the alloy classification are shown.
Titanium: From Ore Extraction and Processing to Its Applications in the Transportation… 295
water quenching from above β-transus. Aging of supercooled β-phase leads to the for-
mation of fine-dispersed α-phase [4] and the metastable ω-phase in some alloys [5].
Thus, in metastable β-alloys, a partial or complete precipitation hardening of the micro-
structure can be carried out to reach ultimate tensile strengths up to 1400 MPa [6].
Three intermetallic titanium aluminides are known: Ti3Al, TiAl, and TiAl3. In particu-
lar, the phases α2-Ti3Al (hexagonal) and γ-TiAl (tetragonal) exhibit low-density and good
high-temperature properties, but both phases show limited ductility at room temperature.
Nevertheless, in the last decades, different γ-TiAl alloys have been developed [3].
2 Ore Winning, Semi-finished Part Production and Recycling
The most crucial titanium minerals are anatase, rutile (both TiO2), and ilmenite
(FeTiO3), which contains up to 53% TiO2. Primary deposits occur in Australia, Canada,
South Africa, Brazil, India, and Norway. The most useful mineral for the extraction of
titanium is rutile. Although it is rarer than ilmenite, its TiO2 content is higher. Titanium
metal is mainly produced by reducing titanium tetrachloride (TiCl4) which is manufac-
tured from natural rutile or from ilmenite, called Kroll’s process. Iron is separated from
the ilmenite during a smelting operation, and a TiO2-rich slag is produced which is then
transformed to TiCl4 by chlorination. In a final step, TiCl4 is reduced to pure titanium
by liquid magnesium reacting to magnesium chloride (mainly MgCl2). Finally, a tita-
nium sponge is obtained by vacuum distillation to remove the remaining MgCl2 and
other byproducts [6].
During the industrial titanium alloy production, the titanium sponge coming from
Kroll’s process is mixed with alloying elements and compressed into parts. The result-
ing compacts are welded together to form an electrode [1]. The electrodes are often
molten to the first ingot in a vacuum arc furnace (VAR). To ensure sufficient homogene-
ity of the alloy, the first ingot has to be remolten once for standard applications or twice
for safety-critical applications, e.g., compressor disks of aircraft engines. For first ingot
production or in the case of non-safety-critical applications, nowadays, electron-beam
cold hearth (re)melting (EBCHR) is used [7]. The two most essential melting processes
(VAR and EBCHR) in industrial titanium production are shown in Fig. 2.
In the case of vacuum arc remelting of alloys containing refractory elements with
high melting temperatures (e.g., Nb or Mo), the resulting ingots might contain small
particles of these elements, the so-called high-density inclusions (HDI). To some extent,
this is unavoidable, as during vacuum arc remelting, superheating of the melt is not
possible, and crystallization occurs naturally in the mold (see, Fig. 2, left). Increasing
electric currents would only lead to increased melting velocity. If refractory elements
are required in titanium alloys, master alloys like Al-Mo or Al-Cr-V are advisable to
reduce their melting temperatures [1]. In electron-beam melting, the process-related
risk of HDI formation is drastically reduced as unmolten refractory element particles
normally sink to the ground of the water-cooled hearth and are trapped in the skull
before they can reach the ingot (see Fig. 2, right).
Fig. 2. Schematic sketches of the VAR process (left) and the EBCHR process (right). In vacuum arc
remelting, an electric arc heats the electrode, drops of the molten material fall down and crystallize in
the mold to form the ingot. In electron-beam melting, the material coming from a feedstock is molten
by electron guns (EB guns), flows along a water-cooled hearth and crystallizes in a mold.
Independent of the melting procedure, the final ingots are typically deformed by
(hot) forging, rotary swaging, rod extrusion, or rolling to form bars, rods, plates, sheets,
or semi-finished products [1]. As these operations are normally performed in air, oxy-
gen can diffuse into the surface of processed parts as oxygen is a strong α-stabilizer.
The so-called α-case formation can occur at the subsurface, a partial transformation to
a-phase. Along with the phase transformation, increased hardness, reduced toughness,
and notch sensitivity are observed in the α-case due to the interstitially dissolved oxy-
gen. Therefore, the α-case is generally removed by stripping or grinding before apply-
ing semi-finished products [3]. Usually, a recrystallization treatment and subsequent
aging produce the final microstructure (see section microstructure).
Ingots coming from VAR or EBCHR can be remolten and used for precision casting
of semi-finished parts or components. Also, skull melting of titanium sponge and the
addition of alloying elements is possible, followed by casting to reduce the number of
necessary production steps. After casting, a hipping treatment or further heat treatments
might be applied to improve the part’s quality and mechanical properties and produce
the final microstructure [8].
Machining is a standard operation in titanium component manufacturing, especially
if complex geometries have to be produced. Here, up to 50% of the semi-finished parts
are removed. However, the machining of titanium alloys involves relatively high pro-
duction costs because of their poor machinability. This difficulty arises from titanium’s
physical, chemical, and mechanical properties [2]. Due to titanium’s relatively poor
thermal conductivity, heat generated by the cutting action cannot diffuse quickly into
the chip’s material, so heat is concentrated in front of the tool’s rake face [9]. Titanium’s
relatively low modulus of elasticity results in the spring back of the workpiece during
the cutting action, causing tool rubbing at the flank face, chatter marks, and tolerance
problems. The high strength of most of the titanium alloys only allows relatively low
cutting speeds compared to the machining of aluminum and steel. Furthermore, titani-
um’s high chemical reactivity limits the number of possible materials for the cutting
tools. Finally, titanium’s strong alloying tendency causes galling, welding, and smear-
ing, leading to the rapid destruction of the tool and decreasing the finished workpiece’s
quality. In addition, titanium machining operations like turning or drilling cannot be
automated due to the formation of long chips. These chips can wrap around the turning
chisel during turning or might get stuck between the drill and the hole, causing poor
surface quality and tool failure [10]. In recent years, severe progress in titanium machin-
ing has been made by the introduction of new tool coatings [11], the application of
enhanced machining processes [12], minimum quantity lubrication [13], and high-
pressure cooling [14].
With increasing titanium production, titanium scrap as a raw material instead of
titanium sponge has become increasingly important. Titanium scrap is produced during
the production of semi-finished (e.g., during titanium ingot production in which poten-
tially contaminated surfaces are often removed by machining between the different
remelting steps) and finished (mainly chips coming from machining operations) prod-
ucts. Before reuse, titanium scrap must be pretreated. Adhering particles and other resi-
dues must be removed from the surface by abrasive blasting or pickling. Depending on
the melting process, the scrap may be mixed with a new sponge for EBCHR melting or
can be welded together for addition to a consumable electrode to be used in titanium
recycling, EBCHR melting is often preferred to avoid HDI formation as scrap might be
contaminated with cemented-carbide particles coming from machining operations [6].
3 Microstructures
α- and β-alloys consist of a single-phase microstructure of typically equiaxed α- or

β-grains. Due to the relatively large self-diffusion rate in the β-phase, grain growth and,
thus, coarsening of the microstructure might become a problem if a related alloy is
annealed above β-transus.
(α + β)-alloys are normally applied with one of three different basic microstructures,
namely, (i) a globular, (ii) a lamellar, or (iii) a duplex structure, a combination of a
globular and a lamellar structure (see Fig. 3). Vacuum arc or electron-beam melting
typically leads to lamellar microstructures with relatively large grains. After deforma-
tion, the microstructure typically consists of strongly deformed lamellae which could
be broken up completely to allow subsequent recrystallization [3].
Globular microstructures (see Fig. 3, left) in (α + β)-alloys are typically achieved by
recrystallization at comparably low temperatures. According to the binary phase dia-
gram, as shown in Fig. 2, the amount of α-phase at annealing temperature (the so-called
primary α-phase, αp) is comparably large (typically >60%). After recrystallization, the
alloy is air cooled. Strongly deformed materials heat treated at short annealing times
typically obtain fine-grained microstructures with fully equiaxed α-grains surrounded
by small amounts of β-phase. Structures with αp grain sizes <5 μm show excellent
mechanical properties.
Fig. 3. Three typical microstructures of (α + β)-alloys. Left: Globular microstructure consisting of

equiaxed αp-grains (bright) surrounded by β-phase (dark), Center: Lamellar microstructure of altering
α- and β-lamellae, colonies of parallel laths are clearly visible. Right: Duplex microstructure consisting
of αp-phase in a lamellar matrix.
As shown in Fig. 3, center, lamellar microstructures are produced by annealing the

alloy above β-transus followed by air or furnace cooling. After crossing β-transus,
α-seeds develop at grain boundaries or other areas of increased energy and grow into
the former β-grains following the burgers relationship [3]. Due to element partitioning,
β-stabilizers are concentrated in the remaining β-phase (as well as the α-stabilizers are
diffused into the newly formed α-phase) so that parallel, altering laths of α- and β-phase
develop the so-called colonies. Depending on the number of active α-seeds and the
cooling conditions, the colony size and the length and width of individual lamellae can
be varied. In general, lamellar microstructures are relatively damage tolerant and show
the best impact properties of all three microstructures [1]. Water quenching from above
β-transus leads to a fully (metastable, as it is supersaturated with β-stabilizing elements)
martensitic structure. Due to its notch sensitivity, relatively low fatigue limit, and the
difficulty to obtain homogeneous microstructures after quenching, primarily if more
significant parts are concerned, fully martensitic structures are not applied [2].
Duplex structures (see Fig. 3, right) consist of globular αp-grains surrounded by
lamellar structures, i.e., they are composed of a mixture of globular and lamellar micro-
structures. Annealing temperatures are typically chosen close to (between 30 and 50 K
below) β-transus, followed by air cooling. Optimized mechanical properties are
achieved once a volume fraction between 10% and 20% of αp-phase is reached.
Subsequent aging is possible [3].
Besides these basic microstructures, multi-step heat treatments might be applied to
(α + β)-alloys, including precipitation hardening of α- (e.g., by Ti3Al) or β-phase (e.g.,
by α-phase) and martensite decomposition once the duplex heat treatment is followed
by water quenching instead of air cooling. During subsequent aging at relatively low
temperatures, the martensite platelets are transformed to α- and β-phase and, thus, fur-
ther refine the structure [15].
Metastable β-alloys are typically used with a certain amount of αp-phase (approx.
15%), which is achieved by heat-treating related alloys below β-transus followed by
fast cooling or water quenching [1]. As a martensitic transformation is not possible,
besides αp-phase, retained β-phase is present. Subsequent aging leads to the precipita-
tion of fine-dispersed α-particles and, thus, to excellent mechanical properties, i.e., the
highest strengths among all titanium alloys [2].
4 Additive Manufacturing
Additive manufacturing (AM) is a modern technology in titanium manufacturing in

which a component is built-up layer-wise from a powder according to a digital model
[16]. In general, spherical powders (made of CP-Titanium as well as of titanium alloys),
mainly being produced by gas atomization like electrode induction melting gas atomi-
zation (EIGA) or plasma melting induction guiding gas atomization (PIGA) with par-
ticle sizes between approx. 10 μm and 100 μm are used in titanium additive
manufacturing. Both powder production processes are shown in Fig. 4.
In titanium additive manufacturing, besides sintering processes like metal injection
molding (MIM), in which titanium powder is processed together with an organic binder
[8], a direct energy source, such as a laser or electron beam, is applied to melt the alloy
powders locally [17]. Related processes are laser- or electron-beam powder bed fusion
(PBF-L or PBF-EB, respectively). Consequently, crystallization occurs at high solidifi-
cation rates in these processes, and the titanium alloys undergo a unique thermal history
so that specific microstructures develop, see below [16].
Mainly two different titanium alloys, i.e., CP-Titanium and Ti-6Al-4V, have been
successfully used to produce dense and reliable components [18]. From the materials’
point of view, one of the most significant advantages of powder bed fusion is the
Fig. 4. Processes used for the production of spherical titanium (alloy) powders applied in additive
manufacturing. Left: Electrode induction melting gas atomization (EIGA). Right: Plasma melting
induction guiding gas atomization (PIGA). The powder production process is shielded from oxygen
containing atmospheres.
n ear-net-shape forming ability. Parts with complex geometries can be produced in one
step without intensive and expensive machining procedures. On the other hand, post-
processing is required, e.g., thermo-mechanical treatments like hipping to close exist-
ing pores and improve the microstructure or surface finishing of functional surfaces by
grinding or machining. Nevertheless, besides reducing raw materials’ scrap, powder
bed fusion is particularly effective and allows designs containing undercuts and, hence,
to realize high-level functional integration in light-weight constructions.
In titanium powder bed fusion, two significant problems occur: First of all, even if
the process is well shielded from the air by the use of inert gases (PBF-L) or vacuum
(PBF-EB), during manufacturing, the amount of oxygen in the alloys processed is
slightly increased (as the powder particles typically contain a thin oxide layer).
Therefore, high-purity (in particular with low-oxygen levels) and, hence, more expen-
sive powders are needed so that the final component fulfills the requirements of the
related materials’ standards regarding the chemical composition. Second, due to the
rapid cooling, the sharp thermal gradient, and the resulting directional solidification,
columnar, textured β-grains develop having their <100> directions oriented approxi-
mately parallel to the building direction [19]. The β-grains either nucleate at the build-
ing platform or grow at the interface between the melt pool and the previously deposited
layer, extending the β-grains to a length far larger than the deposition layer thickness
[16]. During further cooling, most of this β phase is transformed into α′-martensite with
a limited number of orientations [20]. Even if post-processing can transform the mar-
tensite into lamellar structures, then the remaining texture and, thus, anisotropic proper-
ties of the part are still present [21]. Globular or duplex structures cannot be produced
by powder bed fusion. These two problems are the subject of ongoing research activi-
ties, which is aimed at process design, i.e., choosing adequate process parameters or a
dedicated post-processing procedure or developing new titanium alloys dedicated to
additive manufacturing.
5 Application of Titanium in the Transportation Industry
The largest consumer of CP-Titanium and titanium alloys in the transportation sector is
the aircraft industry, in which the used amounts continuously increase [1]. In modern
aircrafts, the amount of titanium (mainly applied in airframe structures and aircraft
engines) can reach 15% of the airplanes’ weight, e.g., 14% of the aerostructure of the
Airbus A350-900 XWB and 15% of the Boeing 787 are made of titanium. In addition,
25% of the GE CF6 aero-engine consists of titanium alloys [22]. A typical buy-to-fly
ratio for the aerospace industry is 6:1 and can be as high as 20:1 for some complex aero-
engine parts [22]. Typical alloys used for airframe structures are Ti-6Al-4V, a medium-
strength (α + β) alloy applied, e.g., in fuselage frames or fasteners, Ti-10V-2Fe-3Al and
Ti-5Al-5V-5Mo-3Cr, both high-strength metastable β-alloys mainly used in landing
gear components. In aircraft engines, fan components (e.g., blades and disks) are made
of Ti-6Al-4V, Ti-6Al-2Sn-4Zr-6Mo, or Ti-5Al-2Sn-2Zr-4Mo-4Cr (Ti-17). Ti-6Al-2Sn-
4Zr-6Mo and Ti-17 are primarily used as fan disks at higher strength levels in some
newer, larger aircraft engines [23], whereas Ti-6Al-4V can be found in several fans of
small and medium-sized commercial aircraft engines.
The current temperature application limit for conventional titanium alloys is approx.
530 °C. Hence, titanium alloys can only be used in the low-pressure compressor. This
can be explained by the relatively poor oxidation resistance of titanium alloys. At room
temperature, titanium spontaneously forms a TiO2 layer once a fresh metallic surface is
exposed to air [3]. However, at temperatures above 550 °C, this oxidation resistance is
dramatically lost. This can be explained by (i) a partial transformation of Ti4+ ions pres-
ent in TiO2 to a mixture of Ti4+ and Ti3+ ions producing vacancies in the oxygen sub-
lattice and (ii) a partial transformation of TiO2 to TiO, TiO2, and Ti2O3, producing
additional phase boundaries and lattice mismatches. Both effects support accelerated
oxygen diffusion through the oxide layer [1]. Consequently, at the metal-oxide inter-
face, additional oxidation occurs, leading to a volume increase so that the oxide layer
finally peels off, and the process starts again. The poor oxidation resistance hinders the
application of titanium alloys in high-temperature environments, e.g., in aircraft
engines’ high-pressure compressor or turbine applications, for which they are other-
wise very well suited [1]. Considerable efforts have been made since the late 1980ies to
improve the oxidation and creep resistance at elevated temperatures of both conven-
tional titanium alloys and titanium aluminides [24]. The two most compelling elements
are Si and Nb [25]. In the case of Si, this can be explained by (1) the formation of a thin
SiO2 layer at the metal-oxide interface forming a diffusion barrier for oxygen penetra-
tion and, thus, reducing further oxidation, and (2) the precipitation of Ti3Si or Ti5Si3
intermetallic phase at the grain boundaries decelerating grain boundary oxidation [25].
On the other hand, the mechanisms of Nb additions leading to improved oxidation
resistance of titanium alloys have been intensively discussed for several years [26].
Still, a satisfying, final explanation has not been found so far.
Typical alloys used in low-pressure compressor applications are Ti-6Al-2Sn-4Zr-
2Mo- 0.1Si and Ti-5.8Al-4Sn-3.5Zr-0.7Nb-0.5Mo-0.35Si-0.06C both Si-containing
near-α-alloys and in case of the latter Nb containing as well. In low-pressure turbines of
modern aircraft engines, titanium aluminides (γ-TiAl) such as Ti-48Al-2Cr-2Nb, a cast
alloy, and Ti-43.5Al-4Nb-1Mo-0.1B (TNM-B7), a wrought alloy, are applied, e.g., in
the PW 1000 engine series of Pratt & Whitney [27].
Besides the application in the (aero)space industry, in racing cars and high-quality
bikes, titanium in the transportation sector is rather limited due to the relatively high
cost of titanium and its alloys. Nevertheless, parts of Diesel engines like compressor
impeller wheels of turbochargers used in trucks, locomotives, and large ships are manu-
factured of titanium alloys or γ-TiAl-based intermetallics, and thus, this shows that
serial applications of titanium exist; whenever, a high specific fatigue limit in combina-
tion with corrosion resistance is needed.
6 Summary
This review paper presents a brief overview of titanium’s metallurgy, and the state-of-
the-art titanium production from ore extraction and processing, melting and remelting,
and semi-finished-product manufacturing, including casting, forging, and machining,
powder production, and additive manufacturing to scrap recycling has been given.
Finally, the application of titanium in the transportation industry has been discussed.
Even if titanium and its alloys are expensive materials, the unique combination of a
high specific (fatigue) strength with corrosion resistance makes them desirable for chal-
lenging light-weight constructions, especially in the transportation industry.
References
1. Peters M, Leyens C, editors. Titanium and titanium alloys. Weinheim: Wiley-VCH; 2002.
2. Boyer R, Welsh G, Collings EW, editors. Materials properties handbook: titanium alloys. Ohio:
ASM International; 1994.
3. Lütjering G, Williams JC. Titanium. Berlin: Springer; 2007.
4. Bowen AW. Strength enhancement in a metastable β-titanium alloy: Ti-15Mo. J Mater Sci.
1977;12:1355–60.
5. Ng HP, Douguet E, Bettles CJ, Muddle BC. Age-hardening behaviour of two metastable beta-
titanium alloys. Mater Sci Eng A. 2010;527:7017–26.
6. Sibum H, Güther V, Roidl O, Habashi F, Wolf HU, Siemers C. Titanium, titanium alloys and tita-
nium compounds. In: Ullmann’s encyclopedia of industrial chemistry. Weinheim: Wiley-VCH;
2017. https://doi.org/10.1002/14356007.a27_095.pub2.
7. Okano H, Hatta Y, Tada O, Tanaka H. Titanium ingot production at Toho Titanium Co., Ltd. In:
Niinomi M, Akiyama S, Hagiwara M, Ikeda M, Maruyama K, editors. Eleventh world conference
on titanium. Kyoto; 2007. p. 155–8.
8. Siemers C, Stöcker C. Developments in titanium research and applications in Germany. MATEC
Web Conf. 2020;321:01003. https://doi.org/10.1051/matecconf/202032101003.
9. Siemers C, Bäker M, Mukherji D, Rösler J. Microstructure evolution in shear bands during the chip
formation of Ti 6Al 4V. In: Lütjering G, Albrecht J, editors. Tenth world conference on titanium
(Ti-2003). Hamburg: Wiley VCH; 2003. p. 839–46.
10. Siemers C, Laukart J, Zahra B, Rösler J, Spotz Z, Saksl K. Development of advanced and free-
machining titanium alloys. In: Gallienne D, Bilodeau M, editors. Fortynineth conference of met-
allurgists, section light metals 2010 – advances in materials and processes. Vancouver; 2010.
p. 311–22.
11. Li G, Lü W, Liu S, Li C, Zhou Y, Wang Q. Multilayer-growth of TiAlN/WS self-lubricating com-
posite coatings with high adhesion and their cutting performance on titanium alloy. Compos Part
B. 2021;211:108620. https://doi.org/10.1016/j.compositesb.2021.108620.
12. Muhammad R, Hussain MS, Maurotto A, Siemers C, Roy A, Silberschmidt VV. Analysis of a
free machining α+β titanium alloy using conventional and ultrasonically assisted turning. J Mater
Process Technol. 2014;214:906–15. https://doi.org/10.1016/j.jmatprotec.2013.12.002.
13. Roushan A, Rao US, Patra K, Sahoo P. Performance evaluation of tool coatings and nanofluid
MQL on the micro-machinability of Ti-6Al-4V. J Manuf Process. 2020;73:595–610. https://doi.
org/10.1016/j.jmapro.2021.11.030.
14. Lu Z, Zhang D, Zhang X, Peng Z. Effects of high-pressure coolant on cutting performance of
high-speed ultrasonic vibration cutting titanium alloy. J Mater Process Technol. 2020;279:116584.
https://doi.org/10.1016/j.jmatprotec.2019.116584.
15. Morita T, Hatsuoka K, Iizuka T, Kawasaki K. Strengthening of Ti-6Al-4V alloy by short-time

duplex heat treatment. Mater Trans. 2005;46(7):1681–6.
16. Zhang T, Liu C-T. Design of titanium alloys by additive manufacturing: a critical review. Adv
Powder Mater (in print). 2021. https://doi.org/10.1016/j.apmate.2021.11.001.
17. Martin J, Yahata B, Hundley J, Mayer JA, Schaedler TA, Pollock TM. 3D printing of high-strength
aluminium alloys. Nature. 2017;549:365–9. https://doi.org/10.1038/nature23894.
18. Lewandowski JJ, Seifi M. Metal additive manufacturing: a review of mechanical properties. Annu
Rev Mater Res. 2016;46:151–86. https://doi.org/10.1146/annurev-matsci-070115-032024.
20. Liu S, Shin YC. Additive manufacturing of Ti6Al4V alloy: a review. Mater Des. 2019;164:107552.
https://doi.org/10.1016/j.matdes.2018.107552.
21. Barriobero-Vila P, Gussone J, Stark A, Schell N, Haubrich J, Requena G. Peritectic titanium alloys
for 3D printing. Nat Commun. 2018;9(1):3426. https://doi.org/10.1038/s41467-018-05819-9.
22. Shokrani A, Al-Samarrai I, Newman ST. Hybrid cryogenic MQL for improving tool life in machin-
ing of Ti-6Al-4V titanium alloy. J Manuf Process A. 2019;43:229–43. https://doi.org/10.1016/j.
jmapro.2019.05.006.
23. Williams JB, Boyer RR. Review – opportunities and issues in the application of titanium alloys for
aerospace components. Metals. 2020;10(6):705. https://doi.org/10.3390/met10060705.
24. Tsuyama S, Mitao S, Minakawa K. Alloy modification of γ-base titanium aluminide for improved
oxidation resistance, creep strength and fracture toughness. Mater Sci Eng A. 1992;153:451–6.
25. Maki K, Shioda M, Sayashi M. Effect of silicon and niobium on oxidation resistance of TiAl inter-
metallics. Mater Sci Eng A. 1992;153:591–6.
26. Tegner BE, Zhu L, Siemers C, Saksl K, Ackland JG. High temperature oxidation resistance
in titanium-niobium alloys. J Alloys Compd. 2015;643:100–5. https://doi.org/10.1016/J.
JALLCOM.2015.04.115.
27. Siemers C, Kiese J. Developments in titanium research and applications in Germany. In: Venkatesh
V, Pilchak AL, Allison JE, editors. Thirteenth world conference on titanium (Ti-2015). San Diego;
2015. p. 41–55. https://doi.org/10.1002/9781119296126.ch5.
HSBC Back-Meniscus Stability
for a Double-Impingement Liquid Metal Feeding
System
D. R. Gonzalez-Morales1, M. M. Isac1 (*), and R. I. L. Guthrie1
1
McGill Metals Processing Centre, McGill University, Montreal, QC, Canada
Mihaiela.Isac@McGill.ca
Abstract. Metallurgically, financially, and environmentally, Horizontal Single-Belt

Casting (HSBC) enjoys several advantages over traditional Continuous Casting (CC) and
Direct-Chill (DC) casting methods for producing thin strips of aluminum alloys.
Nonetheless, “back-meniscus” stability is a critical factor for obtaining good-quality cast
strips. In this work, the effects of air-gap dimensions, and belt speeds, on the stability of
the “back-meniscus” have been studied for a double-impingement metal-feeding system
used for casting AA2024 aluminum alloy. Using the Computational Fluid Dynamics
(CFD) software ANSYS-Fluent 19.1, various combinations of these process parameters
were tested, to obtain optimum results for promoting “back-meniscus” stability.
Additionally, the effect of these parameters on other phenomena, e.g., air entrainment and
heat transfer, are assessed. A new feeding system, without “free stream fall” is proposed,
and for this case, the effect on the “back-meniscus” stability is also analyzed.
Keywords: CFD modeling · Horizontal Single-Belt Casting (HSBC) · Meniscus ·

Thin strips
1 Introduction
Horizontal Single-Belt Casting (HSBC) is a novel Near-Net-Shape Casting Process

(NSSC), which has shown several advantages against traditional Continuous Casting
(CC) and Direct-Chill (DC) casting processes for steel and aluminum, respectively.
These advantages have been widely documented and discussed [1]. In Europe, the pro-
cess is now commercialized by Salzgitter GmbH, at their steel plant in Peine, Germany,
under the name “Belt Cast Technology.” Similarly, in North America, the Horizontal
Single-Belt Casting process and equipment is now being used commercially for pro-
ducing thin strips of copper alloys, by Materion Brush, USA. The process’s schematic
including the components of the HSBC machine can be seen in Fig. 1a, where the
molten metal is fed to an intensively water-cooled moving belt, where it solidifies pro-
gressively. Some process parameters are crucial to promote a stable casting and to
obtain good strip quality, e.g., belt speed, the melt inlet velocity, and the back gap size.
https://doi.org/10.1007/978-3-031-17425-4_42
306 D. R. Gonzalez-Morales et al.
Fig. 1. (a) HSBC schematic. (b) Double-impingement feeding systems for the HSBC process, with a
45° slope entry design.
Depending on the type of alloy to be cast, mainly two types of feeding systems have
been implemented: single and double impingements [2–4]. Double impingement refers
to the encounter of the melt stream with the refractory back wall, and the moving belt.
A schematic of the double-impingement system is presented in Fig. 1b.
The back-meniscus, formed within the back gap between the back-refractory wall
and the moving belt, is crucial to promote a good surface quality, as shown experimen-
tally [1]. Recent mathematical modelling studies have defined the optimal process
parameters in a single-impingement feeding system to avoid backflow and promote
back-meniscus stability, concluding that a back gap of 0.5 mm and a belt speed of 1 m/s
will promote a stable back-meniscus, for the casting of an aluminum alloy AA2024 [1].
Similarly, the objective of the present study is to define the optimal operation conditions
(belt speed and back gap size) to promote a stable back-meniscus behavior for casting
HSBC Back-Meniscus Stability for a Double-Impingement Liquid Metal Feeding System 307
AA2024 alloy with a double-impingement feeding system, through mathematical mod-

elling, using the CFD (Computational Fluid Dynamics) ANSYS-Fluent 19.1 code.
2 Mathematical Modeling of Meniscus Behavior
A two-dimensional, transient state, iso-thermal model of the HSBC system with a

double-impingement feeding was considered in the present research. Two air-gap sizes
and three belt speeds were used in the numerical simulations, giving a total of six cases.
The modified conditions are presented in Table 1.
The geometry for the system was generated using the ANSYS SpaceClaim Software
and is presented in Fig. 2, for a back gap size of 1 mm. Additionally, the inlet and outlet
boundary conditions are specified.
The velocity inlet (3 mm) was defined with a Liquid Aluminum/AA2024 Alloy
Volume Fraction of 1. The moving belt condition was defined as a moving wall with
only an X positive velocity component, and all other walls were defined with a no slip
condition. The properties of the materials, Liquid Aluminum/AA2024 Alloy, and Air
Table 1. Modified operating conditions for the HSBC mathematical modeling.
Feeding system type Double impingement

Back gap size (mm) 0.5 and 1.0
Belt speed (m/s) 0.1, 0.5 and 1.0
Fig. 2. Double-impingement geometry generated in SpaceClaim.
Table 2. Materials properties and operating conditions.

Material Liquid Aluminum/AA2024 Alloy Air
Density, ρ (kg/m3) 2300 1.225
Viscosity, μ (kg/m s) 0.001338 1.7894 × 10−5
Surface tension (N/m) 0.914 –
Contact angle along the moving belt (degrees) 120 –
Inlet nozzle velocity (m/s) 1.0
Fig. 3. (a) Mesh generated in ANSYS Meshing and (b) refinement detail in the second impingement
zone.
were considered constant and are presented in Table 2. The model was defined as being
transient/unsteady state, using a constant time step interval of 1 × 10−6 s.
The mesh for the systems was generated using ANSYS Meshing, with cell refine-
ment in the impingements zones for a better inter-phase accuracy. The meshes gener-
ated were mostly rectangular, with an average cell size of 2 × 10−4 m2. The average
number of cells for these geometries was 40,000 cells, with an average orthogonality of
0.99. The generated mesh for a 1 mm back gap is presented in Fig. 3a, and a detail of
the refinement done on the second impingement zone is shown in Fig. 3b.
The VOF multiphase model was used to solve the Liquid Aluminum/AA2024 Alloy-
Air-Walls interactions and the k-ω Shear-Stress Transport (SST) Turbulence model to
describe the fluid flow. The use of these models has been validated, [5] and the equa-
tions and their details can be found in a previous publication [1].
Fig. 4. Phase 2 (Liquid Al) Volume fraction contours and Velocity Vectors maps, at different times for
a 1 mm gap, and 0.5 m/s belt speed.
3 Discussion
3.1 Double Impingement, 1 mm Air Gap
The mathematical results, liquid aluminum volume fraction, and velocity vectors, for a
double-impingement feeding system, with a back-wall gap of 1 mm and a belt speed of
0.5 m/s, are presented in Fig. 4. Similar results for a 0.1 and 1 m/s belt speed are omit-
ted, since no big differences were observed. The predicted volume fraction contours
show that initially, at 0.07 s, a little backflow penetrates the back gap zone, but very
shortly afterward, it is dragged forward for quasi-steady state operation, at 0.215 s.
With the three different belt speeds, no considerable backflow was generated, similarly
to previous experimental and simulation results [2, 6, 7], when casting AA2024 alloy
using a double-impingement feeding system.
A detailed view of the volume fraction contours on the meniscus zone for the three
different belt speeds, with a back gap of 1 mm is presented in Fig. 5. Initially, air
entrainment between the molten metal and the inclined back wall is observed at 0.13 s
for the three cases. At 0.18 s, the air entrainment reaches the back-meniscus, where the
volume fraction of liquid aluminum drops to around 0.8, generating difficulties to
track the interface accurately. The effect of the air entrainment is enhanced with a
higher belt speed, making the back-meniscus more unstable. Air entrainment with a
double-impingement system had been observed previously, but that was due to the
oscillation in the upper free fall zone and not to oscillations in the zone adjacent to the
inclined back wall.
Velocity contours and vectors within the back-meniscus zone, for the three belt
speeds, are shown in Fig. 6 Regarding the velocity magnitude, with a higher belt speed,
a higher velocity will be reached generally in the system, as expected. But regarding the
Fig. 5. Phase 2 (Liquid Al) Volume fraction contours at the meniscus zone, at different times, for the
3 different belt velocities, and for a 1 mm back gap.
Fig. 6. Velocity contours with vectors map for the three different belt speed velocities, at 0.22 s, with
a 1 mm back gap.
vectors, indicating the flow direction, the ones adjacent to the inclined wall suddenly
change direction chaotically when reaching the back gap, due to the generated air
entrainment reaching the gap, and consequently, exiting the system through the back
gap zone. This confirms the instabilities generated by the potential air entrainment,
which can generate undesired conditions on the final strip. Consequently, a double-
impingement system with a 1 mm back gap is not recommended for the processing of
AA2024 alloy, nor any aluminum alloy, most likely.
Fig. 7. Phase 2 (Liquid Al) Volume fraction contours and Velocity Vectors maps, at different times, for
a 0.5 mm gap, and 0.5 m/s belt speed.
3.2 Double Impingement, 0.5 mm Air Gap
Phase 2 (liquid Al) volume fraction contours and Velocity Vector maps for the 0.5 mm
back gap, and 0.5 m/s belt speed are presented in Fig. 7. According to the volume frac-
tion contours, predictions reveal that no backflow is generated at 40 ms. Similarly, after
150 ms, no air entrainment is observed, in the presence of a stable back-meniscus. This
same behavior was observed for 0.1 and 1.0 m/s belt speeds, hence, the omission of
those repetitive results.
In Fig. 8, volume fraction contours in the meniscus zone, at different times for the
three cases, are shown for analysis. It can be observed that with a 0.5 mm gap, no air
entrainment is predicted, at all belt speeds investigated. Similarly, regardless of the belt
speed, a stable meniscus, with no instabilities caused by air entrainment and with no
oscillations, is promoted during quasi-steady operation, at 150 ms. As such, a 0.5 mm
back gap exhibited considerable improvements over the 1 mm gap. Consequently, pre-
dicted mathematical results recommend using a 0.5 mm back gap for casting AA2024
alloys, and for most other aluminum alloys, given the density, surface tensions, and
viscosity values are not considerably changed.
Figure 9 depicts the velocity contours and vectors for the three different belt speeds
on the back-meniscus zone, where no big differences in velocity magnitude, and direc-
tion are observed. However, compared to Fig. 6 with a 1 mm gap, the direction of the
vectors adjacent to the inclined wall remains constant, since no predicted air entrain-
ment is to be eliminated through the back gap. Given predictions for a 0.5 mm gap, the
improvements compared to the use of a 1 mm gap are evident. As such, a 0.5. mm back
is recommended, as it will promote back-meniscus stability, and avoid air
entrainment.
Fig. 8. Phase 2 (Liquid Al) Volume fraction contours at the meniscus zone, at different times, for the
three different belt velocities, with a 0.5 mm back gap.
Fig. 9. Velocity contours with vectors map for the three different belt speed velocities, at 0.22 s, with
a 0.5 mm back gap.
4 Conclusions
The effects of the back-wall gap and the belt speed, on the HSBC system when using
double-impingement feeding system were studied through mathematical modelling.
Predictions show that a 1 mm back-wall gap will generate air entrainment, and affect
the back-meniscus stability, regardless of the belt speed. In contrast, a 0.5 mm gap will
promote good stability of the meniscus and will avoid air entrainment, for the three belt
speeds studied. These predictions confirm the adequacy of a 0.5 mm gap observed for
the experimental casting of an AA2024 alloy, validating the present mathematical
results.
Acknowledgments The authors acknowledge the financial support received from the Natural
Sciences and Engineering Research Council of Canada (NSERC), and the Aluminum Research and
Development Centre of Quebec (CQRDA), as well as member companies of the McGill Metals
Processing Centre, for this work. The authors would also like to acknowledge the support in software
licensing received from ANSYS Inc. to facilitate this research. Daniel R. Gonzalez-Morales would like
to thank the National Council of Science and Technology of Mexico (CONACyT) for their additional
financial support provided during his graduate studies (CVU:1036046).
References
1. Morales DG, Riviere C, Isac MM, Guthrie RIL. Predicted back-meniscus stability for the hori-
zontal single belt casting (HSBC) process. In: Paper presented at the 60th COM Annual Meeting,
virtual, 17–19 Aug 2021.
2. Xu M, Isac M, Guthrie RIL. A numerical simulation of transport phenomena during the horizontal
single belt casting process using an inclined feeding system. Metall Mater Trans B. 2018;49:1003–13.
https://doi.org/10.1007/s11663-018-1237-4.
3. Ge S, Isac M, Guthrie RIL. The computational fluid dynamic (CFD) modeling of the horizon-
tal single belt casting (HSBC) processing of Al-Mg-Sc-Zr Alloy Strips. Metall Mater Trans
B. 2015;46B:2264–77. https://doi.org/10.1007/s11663-015-0375-1.
4. Xu M, Isac M, Guthrie RIL. Flow instabilities in the horizontal single belt casting process with
an inclined feeding system. ISIJ Int. 2018;59(1):69–75. https://doi.org/10.2355/isijinternational.
ISIJINT-2018-514.
5. Ge S, Celikin M, Isac M, Guthrie RIL. Mathematical modeling and microstructure analy-
sis of Al-Mg-Sc-Zr alloy strips produced by horizontal single belt casting (HSBC). ISIJ Int.
2014;54(2):294–303. https://doi.org/10.2355/isijinternational.54.294.
6. Niaz U, Isac M, Guthrie RIL. Numerical modeling of transport phenomena in the horizontal single
belt casting (HSBC) process for the production of AA6111 aluminum alloy strip. PRO. 2020;8:529.
https://doi.org/10.3390/pr8050529.
7. Niaz U, Isac M, Guthrie RIL. The casting of AA6111 aluminum alloy strips using single and
double impingement types of liquid metal feeding system. J Min Sci Mater. 2020;1:1018.
Development of Fe-Based Metallic Glass Coating
on Magnesium Alloys by Cold Spray
G. Asala1 (*), O. T. Ola1, and O. A. Ojo2

1
Technology Access Centre for Aerospace and Manufacturing, Red River College Polytechnic,
Winnipeg, MB, Canada
gasala@rrc.ca
2
Department of Mechanical Engineering, University of Manitoba, Winnipeg, MB, Canada
Abstract. Magnesium alloys (Mg alloys) are an attractive class of materials for structural
engineering applications in the defense and transportation sectors due to their unique
properties, including low density and high specific strength. Unfortunately, their adoption
has been limited due to these alloys’ inherent poor corrosion and wear resistance. One of
the successful methods to mitigate some of these limitations is applying a high-dense,
well-adhered protective-coating layer to the alloy. Metallic glass (MG) is a unique set of
amorphous materials that possess excellent corrosion and favorable mechanical proper-
ties, including high hardness, toughness, and good wear resistance. It has shown the
potential to be applied as a protective coating on several substrate materials. However,
metallic glass materials are temperature sensitive, so producing dense coatings that retain
the materials’ amorphous structure and avoid the formation of process-related defects
such as oxidation by using conventional thermal-coating techniques is challenging. In this
work, developmental coating of Fe-based MG was deposited on Mg-ZE41A alloy using
cold spray – a solid-state materials deposition process. The influence of the cold spray
deposition parameters on the physical, microstructural, and mechanical properties of the
coating is evaluated.
Keywords: Cold spray · Mg alloy · Fe-based metallic glass · Metal deposition ·

Coating · Wear resistance
1 Introduction
Magnesium alloys (Mg alloys) are attractive as structural materials in the aerospace,
military and automotive industries due to their high specific strength and low density.
Over the last decade, the growth of Mg alloy usage in the automotive industry has
increased and is projected to continue at a rate of 12%/year into the next decade [1]. In
addition, recent developments in processing have led to renewed efforts in the use of
Mg alloys for military applications, from personnel protection to vehicle structures [2].
Although weight reduction is a fundamental consideration in many of these applica-
tions, poor corrosion resistance has limited the adoption of Mg alloys for many applica-
tions. One of the promising methods to mitigate the inherent poor corrosion resistance
https://doi.org/10.1007/978-3-031-17425-4_43
316 G. Asala et al.
of the Mg alloys is to apply a protective coating that effectively separates the alloy from
the corrosive environment. Current coating solutions include electroless plating, organic
coating, conversion coating, laser surface alloying/cladding, and other specialized
anodizing methods [2, 3]. However, even with applying some of these barrier coatings,
normal usage of Mg alloy parts can result in wear, abrasion and mechanical damage,
eventually leading to severe corrosion [2]. Thus, using a well-adhered protective coat-
ing with high abrasion and wear resistance is vital to protecting Mg alloys in service.
Metallic glass (MG) materials are amorphous alloys with a short-range-ordered
atomic arrangement that is different from the typical long-range order in crystalline
materials. As a consequence of the disordered structure in MG, they exhibit a series of
unique properties such as excellent corrosion and wear resistance [4]. The application
of MG as protective coatings is being used in areas in which corrosion and wear degra-
dation constitute a significant issue, especially in the oil and gas, and energy sectors [5].
Fe-based metallic glass is uniquely attractive as a coating material among the MG
materials due to its low cost and a notable glass-forming capacity. Thermal spray meth-
ods such as plasma spraying (PS), arc spraying (AS) and high-velocity oxy-fuel spray-
ing (HVOF) are commonly used to produce Fe-based MG coatings on several materials,
including Mg alloys [3]. Due to the thermal sensitivity of MG, process-related defects
such as potentials for crystallization of the coating, oxidation, and porosity formation
are some of the challenges associated with the thermal process. Cold spray is a solid-
state materials deposition method that uses plastic deformation of the impacting solid
particles for joining feedstock materials instead of melting as in thermal spray methods
[6]. The cold spray technique has shown great potential to produce protective MG coat-
ings, which display better coating properties and performance compared to thermally
produced MG coatings [7]. Although non-structural repairs are being performed on Mg
alloys’ components using the cold spray technique, little work has been done to pro-
duce MG coatings for protection against wear and corrosion degradation. In this work,
a developmental coating of Fe-based MG was deposited on Mg-ZE41A alloy using
cold spray. The influence of the cold spray deposition parameters on the physical,
microstructural, and mechanical properties of the coating was evaluated and
discussed.
2.1 Feedstock and Substrate Material
The Fe-based MG feedstock powder used in this investigation was produced and sup-
plied by Epson Atmix Corporation, Japan. The feedstock powder was produced using
water atomization, and the chemical composition is presented in Table 1. The charac-
teristic of the powder was analyzed using scanning electron microscopy (SEM), and
X-ray computed tomography (X-ray CT) system. The X-ray CT system used is a GE
Phoenix v = tome| × m (GE Sensing & Inspection Technologies GmbH, Wunstorf,
Germany) with dual tubes (180 kV & 300 kV tubes) located at the Smart Factory of Red
River College Polytechnic (RRC Polytech). The X-ray CT analysis was used to
Development of Fe-Based Metallic Glass Coating on Magnesium Alloys by Cold Spray 317
Table 1. Composition of Fe-based MG feedstock powder.

ELement Fe Cr C B Si
Wt% Bal. 2.51 0.75 2.51 6.68
Table 2. X-ray CT scanning parameters for feedstock material and coating.

Voltage
Scan parameter (kV) Current (μA) Voxel size (μm) # of projections
Fe-based MG powder 68 65 1 2300
Coating 150 110 3.6 2400
evaluate the internal powder porosity, powder morphology, particle size distribution
(PSD) and sphericity. The scan parameter for the powder is presented in Table 2. In
addition, the amorphous nature of the feedstock powder was evaluated with a Bruker
D5000 X-ray diffractometer using the Cu Kα radiation collected from 20 to 100° 2θ at
0.02° step size and scan at a rate of 1°/min.
The substrate material is a cast Mg alloy ZE41A with a nominal composition of (%
weight) 4.3 Zn, 0.8 Zr, 0.7 Ce, 1.2 RE (rare earth elements) and the balance Mg. The
alloys were received in the artificially aged T5 condition and were machined into cou-
pons having dimensions of 152.4 mm × 38.1 mm × 6.35 mm. Before applying the coat-
ings, the coupons were ground using silicon-carbide paper down to 600 grit and then
cleaned with acetone in an Ultrasonic bath.
2.2 Fe-Based MG Coating by Cold Spray
The Fe-based MG coatings in this work were produced using the Impact Innovation
cold spray system at the Centre for Aerospace Technology and Training (CATT),
Winnipeg, Manitoba, Canada (see Fig. 1). Nitrogen gas (N2) was used as the process
gas. The cold-sprayed coatings were produced using five (5) process temperatures
(750 °C, 800 °C, 850 °C, 900 °C, and 950 °C). All other deposition parameters such as
the pressure (40 bar), powder feed rates (4 rpm), standoff distance (25 mm), and the
nozzle travel speed (250 mm/s) were kept constant. Ten (10) layers of the MG coating
were deposited on the coupons at each process parameter.
The coated coupons were sectioned perpendicular to the spraying gun's travel direc-
tion and prepared for microscopy using standard metallographic procedures. XRD
analysis was performed on the coated coupons to evaluate the influence of the deposi-
tion parameters on the amorphous nature of the material after the deposition. In addi-
tion, X-ray CT analyses were performed on 10 mm × 10 mm × 10 mm specimens
extracted from the coated coupons to investigate the characteristics and extent of defects
in the MG coatings. The X-ray CT parameters used for analyzing the coupons are also
presented in Table 2.
The micro-Vickers hardness of the coatings was measured using a LECO M-400
Hardness tester, under an applied load of 300 gf at 10 s dwell time. Twenty (20) hard-
ness indentations were taken for each sample and performed according to ASTM E384-
17 standard.
318 G. Asala et al.
Fig. 1. Impact Innovation’s cold spray system at the Centre for Aerospace Technology and Training
(CATT), Winnipeg, Manitoba, Canada.
3.1 Feedstock Powder Characterization
The SEM micrograph and the X-ray CT analysis of the feedstock materials are pre-
sented in Figs. 2 and 3, respectively. The SEM micrograph shows that the powder has a
spherical morphology with a smooth surface, as seen in Fig. 2. The 2D X-ray CT color-
mapped image and the histogram in Fig. 3a, b, respectively, summarize the powder's
particle size distribution (PSD). The size analysis of the MG powder shows the PSD as
D10 = 8 μm, D50 = 11 μm and D90 = 21 μm. The powder sphericity is found to be between
0.5 and 0.65. In cold spray, particle morphology is known to influence deposition effi-
ciency and defect formation; spherical feedstock powders are preferred [8]. The inter-
nal porosity in the powder feedstock is estimated to be about 0.45%, and the distribution
is shown in the 2D colored image and 3D X-ray CT transparent volume in Fig. 3a, c,
respectively. Since cold spray deposition technique is a solid-state materials consolida-
tion process, the extent of internal porosity in the starting feedstock material is crucial
because it is transferred into the coating. The XRD analysis of the powder in Fig. 4
showed a broad peak that is typical of an amorphous material.
3.2 Fe-Based MG Coating Characteristics
Representative macrographs of the Fe-based MG-coated Mg-ZE41A coupons are

shown in Fig. 5. It is seen that the coatings are smooth, continuous, and uniformly
applied to the surfaces of the substrate alloys.
Fig. 2. SEM micrograph of the Fe-based metallic glass used for coating.
Fig. 3. X-ray CT analysis of powder (a) 2D colored mapped size distribution, (b) histogram PSD, (c)
histogram sphericity, and (d) 3D volume showing porosity distribution.
The XRD analyses of coated coupons at different processing temperatures are pre-
sented in Fig. 6a–e. In addition, plot 6(f) shows the XRD spectrum from a coated cou-
pon subjected to post-deposition annealing at 950 °C for 0.5 h. The plotted data from
the as-processed MG coating display a broad peak similar to those observed from the
powder. Also, the characteristic XRD patterns observed in the post-deposited annealed
coating, Fig. 6 plot (f), are absent in all the as-processed MG coatings. This indicates
320 G. Asala et al.
Fig. 4. XRD analysis of the Fe-based MG powder.
Fig. 5. Photographs of Fe-based MG-coated coupon deposited at (a) 800 °C and (b) 900 °C.
that there was little or no crystallization of the amorphous material during the coating
process for all the process temperatures. Although the process temperature may be as
high as 900 °C, the powder particles are not subjected to such high temperatures as they
are introduced at the expanding region of the Laval nozzle, where the temperatures are
much lower [9]. In addition, the time spent by the particles in flight is short and has
been reported to be less than the time required for significant crystallization of the
amorphous material to occur [9]. This explains the minimal crystallization observed in
the coating. Maintaining the level of amorphicity of coating is important; studies have
shown that the primary factor influencing the properties of MG coating, especially the
corrosion resistance, is the amorphicity [10].
The cross-sectional micrographs of the coated coupon, produced with different pro-
cess temperatures, are presented in Fig. 7. The micrographs show a well-adhered coat-
ing with no signs of delamination, at the deposited MG/substrate interface. Some
Fig. 6. XRD analyses of the Fe-based coatings showing the diffraction pattern of the metallic glass
(a–e) as-coated at 750–950 °C, and (f) annealed coating at 950 °C for 0.5 h.
un-deformed particles can be seen concentrated mainly along the substrate/MG glass
interface. This observation could be attributed to the Mg alloy's low hardness, allowing
much harder MG particles to penetrate the substrate without undergoing substantial
plastic deformation. From Fig. 7a, micro-fissure formation can be observed between
splats of the coating produced at 750 °C, which is absent in other coatings produced at
higher temperatures. In addition, the thickness of the coating produced at each process
temperature and their relative deposition efficiency – a ratio of efficiencies at the vari-
ous deposition temperatures to that at 950 °C – is summarized in Table 3 and Fig. 8,
respectively. It is observed that the coating thickness and the deposition efficiency
increase with temperature up until 850 °C, after which no significant increase in the
deposition efficiency occurred. For MG particles to deform homogeneously and obtain
good coating by cold spray, it has been shown that the temperature of the impacting
particles needs to have reached the super-cooled liquid region of the material [11].
Therefore, the presence of micro-fissure at 750 °C and the low efficiency could be
attributed to the inhomogeneous deformation of the particles at lower process
temperatures.
3.3 Defect in Fe-Based MG Coating
The plot summarizing the percentage volume of porosity in the MG coating using
X-ray CT analysis is presented in Fig. 9a. In addition, the 3D volumetric image of the
porosity is shown in Fig. 9b. Generally, the level of porosity in the coating decreases as
322 G. Asala et al.
Fig. 7. SEM micrographs of Fe-based MG coating at (a) 750 °C, (b) 800 °C, (c) 850 °C, (d) 900 °C,
and (e) 950 °C process temperatures.
Table 3. Measured coating thickness at various process temperatures.

Process temperature (°C) 750 800 850 900 950
Coating thickness (mm) 0.568 1.354 1.928 1.964 2.002
SD (mm) ± 0.037 ± 0.023 ± 0.020 ± 0.015 ± 0.014
Fig. 8. Relative efficiency of Fe-based coating as a function of process temperature.
the deposition temperature increases from 750 °C to 950 °C. It can be observed from
the plot that the porosity levels initially decreased from 1.45% to 0.38% as the process
temperature increased from 750 °C to 850 °C and then increased to about 0.79% at
950 °C.
Henao et al [12] reported a similar trend in porosity levels for Fe-based MG coating
with deposition temperatures. The decreased level of porosity within a certain process
temperature range and the increased porosity levels at high deposition temperatures
were attributed to the homogeneous deformation tendency and powder crystallization
at these temperatures, respectively. In this work, however, minimal crystallization of
the MG coatings occurred from the XRD data, and the observed marginal porosity
increase could not be attributed to particle recrystallization. During cold spray, increas-
ing process temperature also acts to increase particle acceleration; this could affect the
critical impact velocities of the particle, reducing the homogenous deformation capac-
ity of the powder and leading to higher porosity.
From Fig. 7, cracking could be seen in some of the MG coatings. X-ray CT analysis
was used to investigate the nature and extent of cracking in the coating. The volume of
defects obtained using the X-ray CT is summarized in Fig. 10a. It is seen that cracks
begin to appear in the coating produced using process temperature above 850 °C, and
the volume of the cracks increases with the deposition temperatures. Figure 10b shows
the X-ray CT 3D volume of cracking in the coating at the highest process temperature
(950 °C). The cracks can be seen running perpendicularly to the travel direction of the
coating. These cracks seem to be created by the thermal stress generated in the coupon
during coating. In addition, the thermal expansion difference between the substrate and
324 G. Asala et al.
Fig. 9. X-ray CT porosity analysis of coating. (a) A plot of the % volume of porosity vs. processing
temperature. (b) 3D XCT volume showing the distribution of porosity.
Fig. 10. X-ray CT analysis of cracks in the coatings. (a) %volume of cracks in coating. (b) XCT 3D
volume of cracks in the deposits.
Table 4. Vickers hardness measurements across the Fe-based MG coatings.

Fe-based metallic glass coating
ZE41A substrate 750 °C 800 °C 850 °C 900 °C 950 °C
Average 65.24 781.1 753.5 716.0 722.5 747.8
S.D ± 1.79 ± 24.61 ± 64.72 ± 30.05 ± 25.77 ± 69.44
Fig. 11. Hardness measurements of the MG-coating deposition at different process temperatures.
the coating materials may contribute to the cracking propensity. Higher process tem-
peratures, to an extent, can help improve coating efficiency and reduce porosity, but
they enhance thermal stresses, which can cause increased cracking in coatings. Using
moderate process temperatures and applying inter-layer heat treatments could be fur-
ther explored to minimize thermal stresses during deposition and reduce or eliminate
cracking in coatings.
3.4 Hardness Properties of the Fe-Based MG Coatings
The Vickers hardness values of the substrate materials and the Fe-based MG coatings
produced at each operating temperature are summarized in Table 4 and plotted in
Fig. 11. There is no significant difference between the measured hardness values at all
processing temperatures. This can be expected as there is no noticeable change in the
amorphous nature of the coating during deposition. Compared to the substrate material,
the hardness of the Fe-based coating is significantly higher, indicating an increased
wear and abrasion resistance tendencies of the MG coating.
326 G. Asala et al.
4 Conclusions and Outlook
In this work, cold spray developmental coating of Fe-based MG on Mg-ZE41A alloy is

performed. The following conclusions can be drawn from the results:
• The Fe-based MG coatings produced using cold spray technique are smooth, con-
tinuous, and uniformly applied to the surface of the substrate alloy. As a result, a
well-adhered bonding was observed at the substrate coating interface.
• The deposition efficiency of the coating is found to increase with operating tempera-
ture up to 850 °C, after which no noticeable change is observed. In addition, micro-
fissures are seen at the interface between the MG splats at low process temperature
(750 °C), indicating poor bonding.
• The porosity in the coating decreases with increasing process temperature up to 850
C and thereafter increases slightly with temperature increase.
• The propensity of the coating to thermal-induced cracking is found to increase with
increasing deposition temperatures.
Outlook
• The Fe-based MG deposition by cold spray technique appears promising with proper
optimization of the deposition parameters. Further process optimization and coating
performance testing are ongoing.
Acknowledgments This work was supported in part by funding from the Innovation for Defence
Excellence and Security (IDEaS) program from the Department of National Defence (DND). The
authors would like to acknowledge the support of Turker Sengoz and Vui Mai Nguyen for helping with
cold spray robot programming and X-ray CT analysis, respectively.
References
1. Luo AA. Materials comparison and potential applications of magnesium in automo-

biles. In: Essential readings in magnesium technology; 2014. p. 25–34. https://doi.
org/10.1002/9781118859803.ch4
2. Mathaudhu SN, Nyberg EA. Magnesium alloys in U.S. military applications: past, current and
future solutions. In: Essential readings in magnesium technology; 2014. p. 71–6. https://doi.
org/10.1002/9781118859803.ch10
3. Zhang H, Wang S, Yang X, Hao S, Chen Y, Li H, Pan D. Interfacial characteristic and microstruc-
ture of Fe-based amorphous coating on magnesium alloy. Surf Coat Technol. 2021;425:127659.
https://doi.org/10.1016/j.surfcoat.2021.127659.
4. Wang Q, Han P, Yin S, Niu WJ, Zhai L, Li X, Mao X, Han Y. Current research status on cold
sprayed amorphous alloy coatings: a review. Coatings. 2021;11:1–21. https://doi.org/10.3390/
coatings11020206.
5. NASA. Metallic glass coatings improve power plant. Oil Rig Productivity; 2020. https://spinoff.
nasa.gov/Spinoff2020/ip_1.html
6. Champagne VK Jr. Practical cold spray; 2021. https://doi.org/10.1007/978-3-030-70056-0.
7. Su J, Kang JJ, Yue W, Ma GZ, Fu ZQ, Zhu LN, She DS, Wang HD, Wang CB. Comparison of tribo-
logical behavior of Fe-based metallic glass coatings fabricated by cold spraying and high velocity
air fuel spraying. J Non Cryst Solids. 2019;522 https://doi.org/10.1016/j.jnoncrysol.2019.119582.
8. Champagne VK. The cold spray materials deposition process: fundamentals and applications.
Woodhead Publishing Limited; 2007. https://doi.org/10.1533/9781845693787.
9. Wang Z, Ma L, Han B, Huang G, Cao Q, Sun M. Influence of parameters on the cold spraying
FeCoCrMoBCY amorphous coatings. Surf Eng. 2021;37:545–57. https://doi.org/10.1080/0267084
4.2020.1805716.
10. S.N.A.K.A.B. Sarkar. Corrosion behavior of Fe-based amorphous/nanocrystalline composite coat-
ing: correlating the influence of porosity and amorphicity. In: Corrosion and materials degradation
web conference; 2021.
11. Concustell A, Henao J, Dosta S, Cinca N, Cano IG, Guilemany JM. On the formation of metallic
glass coatings by means of Cold Gas Spray technology. J Alloys Compd. 2015;651:764–72. https://
doi.org/10.1016/j.jallcom.2015.07.270.
12. Henao J, Concustell A, Cano IG, Cinca N, Dosta S, Guilemany JM. Influence of Cold Gas Spray
process conditions on the microstructure of Fe-based amorphous coatings. J Alloys Compd.
2015;622:995–9. https://doi.org/10.1016/j.jallcom.2014.11.037.
Applications of Advanced Thermal Analysis
for Aluminum Foundries
François Audet1 (*) and Yohan Tremblay1

1
Foundry Solutions, Metallurgical Services Inc, Magog, QC, Canada
Info@solutionsfonderie.com
Abstract. Foundries can take different samples of the aluminum melt before and during
the casting process to measure melt properties: RPT, chemistry, temperature, hydrogen,
inclusions, oxides, and thermal analysis. The thermal analysis sample provides informa-
tion on the solidification properties of the melt. Traditionally, thermal analysis systems
rely solely on the temperature curve to assess a few criteria. However, recent advances in
thermocouple signal processing allow automatic cooling curve analysis to provide more
than the well-known grain refinement and eutectic modification potentials of aluminum-
silicon alloys. Using a variant of the Newtonian approach and the zero curve, phases
forming during solidification of the sample are measured as an energy percentage over the
total energy absorbed or released. This opens the possibility for advanced applications
such as the terminal freezing to assess the hot-tearing susceptibility of aluminum alloys.
The fraction solid curve from the thermal analysis result can be used for casting filling and
solidification simulation inputs.
Keywords: Aluminum thermal analysis · Cooling curve · Zero curve · Solid

fraction curve · Hot tearing
1 Introduction
Thermal analysis is the study of the temperature change during the solidification of a
sample of molten metal [1]. In the aluminum industry, thermal analysis is mainly used
to assess the transformations which modify the properties of the casting. For example,
the two main applications are the control of the grain refinement and eutectic modifica-
tion levels for aluminum-silicon alloys such as A356 and 319. With the development of
faster computers, research on the zero curve, numeric data filters and the use of the
derivatives, advanced thermal analysis became a valuable tool that gives the metallur-
gist a deeper understanding of the solidification of aluminum alloys.
https://doi.org/10.1007/978-3-031-17425-4_44
330 F. Audet and Y. Tremblay
Fig. 1. A356 Cooling Cur & Zero Curve.
2 Zero Curve
The zero curve, also known as the baseline, is a cooling rate curve assuming that there
is no phase transformation. Figure 1 presents an example of a typical cooling curve
obtained with the SFTA system. The blue plot is the temperature, the orange plot is the
cooling rate, and the pink plot is the zero curve.
According to Sokolowski [2, 3], the solid fraction fs(t) can be obtained by calculating
the area between the zero curve ZC(t) and the cooling rate CR(t) between the time at
which the sample is liquid tl and is solid ts.
ZC ( t ′ ) − CR ( t ′ ) dt ′
t
∫
fs (t ) = tl
. (1)
ZC ( t ′ ) − CR ( t ′ ) dt ′
ts
∫ tl

Therefore, determining the solid fraction strongly depends on the zero curve. Two mod-
els are commonly known to determine the zero curve: the Fourier model and the Newton
model [4]. A novel variant of the Newton method is used in the SFTA system.
2.1 Newton Model
Although the Newton model is a simpler method to model the temperature variation of
an object in cooling, the literature shows that it is possible to precisely predict the
melt properties [1, 5]. Also, it is easier to set up on the foundry floor, since only one
thermocouple is required. The Newton model used in the SFTA system assumes the
following hypothesis:
• The heat capacity of the mold is small enough to be neglected.
• The heat capacity is different depending on the liquid or solid state.
Applications of Advanced Thermal Analysis for Aluminum Foundries 331
Table 1. Average confidence interval of thermal arrests.
Measurement 95% confidence interval

Temperature ± 0.2 °C
Cooling rate ± 0.02 °C/s
Solid fraction ± 0.4%
Relative energy per phase ± 0.4%
• The initial temperature of the mold is taken into account (without measuring it
directly).
3 Repeatability of SFTA System
A repeatability test was performed in order to evaluate the reproductivity and repeat-
ability of the measurements of the SFTA thermal analysis system. Four thermal analy-
sis stands were used side by side to measure a stable A356 furnace. During a four-hour
window, 48 samples were taken at 725 °C. As shown in Table 1, the resulting confi-
dence intervals indicating a high level of reproducibility. It should be noted that the
confidence interval is smaller than that of the thermocouples. This can be explained by
the use of a numerical filter and the use of the cooling rate to detect all the thermal
arrests. The resulting arrest temperature is calculated based of the arrest cooling rate.
4 Applications of Advanced Thermal Analysis
4.1 Terminal Freezing Range and Hot-Tearing Susceptibility
Certain aluminum alloys such as Al-Cu, Al-Zn, and Al-Mg are prone to hot tearing. A
complete dendritic solidification tends to have the adverse effect of extending the time
period that the alloy is exposed to the hot-tearing susceptible zone, which causes more
solidification defects in comparison to eutectic alloy (Al-Si) [6].
Using thermal analysis, it is possible to determine the hot-tearing index using the
solid fraction and the temperature (Terminal freezing range). For example, the tempera-
ture range between 90% and 99% solid gives an indication of how at risk the alloy is to
hot tearing before casting. The casting and feeding designs need to account for the hot-
tearing tendency of the alloy still, the liquid metal condition should be within the
acceptable hot-tearing susceptibility index range. If not, the melt should be treated
before casting.
The SFTA system can be used to control parameters that are known to reduce the
hot-tearing susceptibility of the alloy such as grain refinement and gating design.
Also, the cooling curve can be used in simulation software for a more representative
simulation while designing the mold.
332 F. Audet and Y. Tremblay
Fig. 2. Typical A356 fraction solid curve.
4.2 Solid Fraction Curve in Simulation Software
The solid fraction obtained through thermal analysis can be used in simulation software
in order to better design the gating of the mold. For example, thermal analysis samples
can be taken at different states. Then, the engineer can use the solid fraction curves and
simulate the same part under different conditions. Therefore, the engineer can deter-
mine what are the optimal conditions to produce the part. The main advantage is that
the simulated solidification of the part will be significatively more representative of the
shop floor. This is particularly the case when casting proprietary alloys that are not
documented in the database. An example of a solid fraction curve is shown in Fig. 2.
SFTA results can be used as a control loop on the shop floor in order to know if the
simulated melt properties are within the range of the real shop floor melt properties.
References
1. Erbaş KC. A new baseline for the Newtonian thermal analysis of casting: two-capacitive sys-
tem baseline: modeling the effects of the thermal capacity of mold. J Therm Anal Calorim.
2014;119(1):183–9. https://doi.org/10.1007/s10973-014-4143-2.
2. Smith TF, Waterman MS. Identification of common molecular subsequences. J Mol Biol.
1981;147:195–7. https://doi.org/10.1016/0022-2836(81)90087-5.
3. Djurdjevic MB, Chen J, Sokolowski JH. A novel algorithm for the calculation of latent heat of solid-
ification for multi-component aluminum alloys. Trans Am Foundry Soc. 2003;111(03–094):1–11.
4. Palkowski H, Günther L. Comparison of newtonian and fourier thermal analysis techniques for
calculation of latent heat and solid fraction of aluminum alloys. Metall Mater Eng. 2005;3:215–24.
Applications of Advanced Thermal Analysis for Aluminum Foundries 333
5. Erbaş KC. Analytically solved solid fraction model for the Newtonian thermal analysis of cast-
ing. Metall Mater Trans A Phys Metall Mater Sci. 2016;47(6):3026–30. https://doi.org/10.1007/
s11661-016-3464-0.
6. Nabawy AM, et al. A review on the criteria of hot tearing susceptibility of aluminum cast alloys. Int
J Met. 2021;15(4):1362–74. https://doi.org/10.1007/s40962-020-00559-3.
Microstructure and Texture in Additively
Manufactured Maraging Steel Lattice Structures
Farzad Khodabakhshi1, 2 (*) and Mohsen Mohammadi2

1
Marine Additive Manufacturing Centre of Excellence (MAMCE), University of New
Brunswick, Fredericton, NB, Canada
fkhodaba@unb.ca; fkhodabakhshi@ut.ac.ir
2
School of Metallurgy and Materials Engineering, College of Engineering, University of Tehran,
Tehran, NB, Iran
Abstract. Cellular lattice structures additively manufactured using the laser powder bed
fusion (LPBF) technique was considered in this study. In this case, a body-centered cubic
(BCC) design was deposited using maraging steel powder. The diagonal cross section of
the cellular lattice was considered to characterize the microstructural details. Electron
backscattering diffraction (EBSD) analysis revealed the microstructural evolutions and
crystallographic texture across the struts and junctions depending on the lattice geometry.
The progression of martensitic phase transformation in the material upon layer-by-layer
process was studied tracking the fraction of austenite and martensite phases.
Keywords: Maraging steel · Cellular lattice structure · Laser powder bed fusion ·
Texture
1 Introduction
Additive manufacturing (AM) is a newly developed technology developed to fabricate

metallic materials undergoing layer-by-layer deposition methodology mainly based on
two well-known techniques of powder bed fusion (PBF) and directed energy deposition
(DED) [1]. In this case, the PBF approach using laser or electron beam heat sources is
outstanding to overcome the solid complex geometries [1, 2]. One of the leading classes
of complex structures is cellular lattices, classified as porous integrate components hav-
ing a regular or irregular network of cells geometry and opening, with high-application
potentials for various possible industries such as biomedical sectors (implanting and
cell culture) as well as aerospace or automotive regarding the superior crush perfor-
mance of these designs [3]. Traditionally, these cellular structures’ stochastic or deter-
ministic architectures were prepared using powder metallurgy and investment casting,
which were very difficult to manufacture and limited in design [4]. Laser powder bed
fusion (LPBF) has recently displayed exceptional capabilities for precise manufactur-
ing of these complex architectures with a broad range of geometries non-comparable to
conventional manufacturing techniques [4, 5]. Maraging steels are a unique class of
metals mainly used in aerospace, automotive, and marine and offshore applications due
https://doi.org/10.1007/978-3-031-17425-4_45
336 F. Khodabakhshi and M. Mohammadi
to their superb combination of high strength and good elongation to failure [6]. Hence,
the construction of cellular lattice structures from these materials using the LPBF tech-
nique has attracted recent attention, whereas few reports were published on this type
geometry and material choice [7–9]. Tallon et al. [9] studied the crash performance of
these maraging steel lattice structures prepared by LPBF. The focus of this article is on
characterizing the microstructural details and crystallographic texture of microlattice
struts using electron backscattering diffraction (EBSD) analysis.
2 Experimental Methodology
Maraging steel powder with the standard designation of CL50WS and chemical com-
position of Table 1 supplied from GE company were considered for laser powder bed
fusion (LPBF) using a GE Concept Laser M2 V2 machine. After constructing a cellular
lattice structure with a body-centered cubic (BCC) design, the shear plane cross section,
including all struts and the related junctions, was considered for microstructural exami-
nations. At first, the geometry was sectioned by electrical discharge machining (EDM),
and after that, mechanical grinding and polishing procedures were accomplished. The
final sample preparation step was conducted by vibratory polishing on colloidal silica
for around 10 h to remove the possible surface mechanical damages undergoing a
chemo-mechanical treatment. The structure of additively manufactured cellular lattice
design of maraging steel was investigated with field emission-scanning electron micros-
copy (FE-SEM) along with energy-dispersive X-ray spectroscopy (EDS) analysis,
using a Thermo-scientific Scios 2 microscope. Then instrument was equipped with a
solid high-speed scan rate EBSD detector of OXFORD Instruments SYMMETRY ded-
icated for high-resolution analysis and EBSD microstructural mapping.
Figure 1 illustrates the macro-structure of additively manufactured BCC lattice struc-

ture section across the shear plane, captured by combining FE-SEM micro-images.
This map validates the soundness of 3D printing to fabricate the complex geometry of
the cellular lattice. The dimensional uniformity of the struts with no obvious macro-
defects and the regularity of lattice units’ periodic geometry are the general matters that
can be highlighted by assessing this macro-profile. Meanwhile, the formation of some
dispersed micro-pores due to solidification shrinkage with white contrast is visible
through the struts of the lattice structure. High-magnification FE-SEM images from the
Table 1. Typical chemical composition ranges for the primary CL 50WS maraging steel powder
supplied from GE Additive company (in wt%).
Element Ni Co Mo Ti C Si Mn Cr Fe
Content 17–19 8.5–10 4.5–5.2 0.8–1.2 ≤0.03 ≤0.1 ≤0.15 ≤0.25 Base
Microstructure and Texture in Additively Manufactured Maraging Steel Lattice Structures 337
microstructure of deposited braces and corresponding EDS elemental distribution maps

are presented in Figs. 2 and 3, respectively. The mentioned micro-pores are addressed
as black-contrast distinctive features in Fig. 2. The formation of fine cellular grains
inside the elongated martensitic laths of maraging steel is evident. Based on EDS maps
in Fig. 3, homogenous distribution of alloying elements inside the steel matrix after
LPBF and no segregation or precipitation of secondary phase agents are the main
imperative phenomena that can be reported.
Centered struts junction in the cross section of the designed lattice structure, as shown
in Fig. 1, was analyzed for EBSD mapping, while the related Euler map grain structural
combined plot is demonstrated in Fig. 4. In this broad map, the microstructure of differ-
ent regions at the junction and strut parts, and their corresponding soundness are revealed.
The structure of LPBF-deposited maraging steel seems martensitic for the whole body,
while the fraction of the martensite phase and its geometry can be varied across the
design depending on the applied heat input index and realistic cooling rates depending
the volume of material under deposition at each layer. The fraction of porosities caused
by solidification shrinkage is heterogeneous across this analyzed area, as black-color
holes on the map. This signifies the contribution of functionality and gradient on the
microstructure and soundness of the consolidated part. In strut junction, thanks to higher
metal deposition and larger volume of the surrounded solid area, the corresponding
lower cooling rates induced a structure with a lower fraction of porosities and larger
martensitic laths. Meanwhile, the scenario for struts is reverse. Refined martensitic struc-
tures and a higher fraction of porosities were indexed inside those four wings.
More detailed EBSD investigations from the middle of the junction region are dem-
onstrated in Figs. 5 and 6, respectively. The formation of large martensitic laths and
Fig. 1. Macro cross section of BCC lattice structure across the diagonal shear plane.
Fig. 2. FE-SEM microstructure of lattice struts.
Fig. 3. EDS maps from different elements distribution for the additive manufactured maraging steel.
Fig. 4. The Euler EBSD grain structure map from the centered strut cross-junction in the middle of the
lattice structure.
Fig. 5. (a) Band contrast, (b) Euler microstructure, and (c) phase-color map determined by EBSD
analysis from a region in the middle of struts junction.
Fig. 6. The IPF grain textural maps along with different directions of (a) X, (b) Y, and (c) Z from the
analyzed area show preferred orientations for martensitic laths.
delicate cells inside them upon rapid cooling of layer-by-layer metal deposition during
LPBF is well illustrated in Fig. 5a, b. The contribution of retained austenite phase with
red color contrast can be noted along the cellular boundaries of martensitic laths as
indexed in the EBSD phase structure of Fig. 5c. Another outcome of the EBSD analysis
could be the crystallographic textural tendency, while it has been assessed in this work
by plotting the related inverse pole-figure (IPF) plots along with different directions of
X, Y, and Z, as shown in Fig. 6. Typical assessment of these IPF grain structural maps
suggests a weak directionality of martensitic laths toward the building direction. The
reason for this feeble orientation dominancy lies with the fact that the firm solidification
texture of LPBF was primarily affected and disappeared by the subsequent solid-state
austenite to ferrite/martensite phase transformation, operated in this steel 6.
4 Conclusions
EBSD characterization of additively manufactured maraging BCC lattice structures

prepared by laser powder bed fusion was assessed for the first time. By conducting
large-area EBSD mapping, microstructural features and crystallographic macro-texture
across the shear plane of printed design were studied. As revealed, the fraction of
porosities caused by solidification shrinkage varied depending on geometrical location

across the lattice structure, attributing to the solidification path alteration. The forma-
tion of larger martensitic laths at the junctions and finer cells at the struts were induced
a global orientation dependency toward the building direction in connection to the vari-
able local cooling rate cycles for different layers.
References
1. DebRoy T, et al. Additive manufacturing of metallic components – process, structure and proper-
2. Zitelli C, Folgarait P, Di Schino A. Laser powder bed fusion of stainless steel grades: a review.
Metals. 2019:9. https://doi.org/10.3390/met9070731.
3. Hussein A, Hao L, Yan C, Everson R, Young P. Advanced lattice support structures for metal
additive manufacturing. J Mater Process Technol. 2013;213:1019–26. https://doi.org/10.1016/j.
jmatprotec.2013.01.020.
4. Chen LY, Liang SX, Liu Y, Zhang LC. Additive manufacturing of metallic lattice structures: uncon-
strained design, accurate fabrication, fascinated performances, and challenges. Mater Sci Eng R
Rep. 2021;146 https://doi.org/10.1016/j.mser.2021.100648.
5. Panesar A, Abdi M, Hickman D, Ashcroft I. Strategies for functionally graded lattice structures
derived using topology optimisation for Additive Manufacturing. Addit Manuf. 2018;19:81–94.
https://doi.org/10.1016/j.addma.2017.11.008.
6. Shamsdini S, et al. A relationship between the build and texture orientation in tensile loading of the
additively manufactured maraging steels. Addit Manuf. 2021;41:101954. https://doi.org/10.1016/j.
addma.2021.101954.
7. Concli F, Gilioli A. Numerical and experimental assessment of the mechanical properties of
3D printed 18-Ni300 steel trabecular structures produced by Selective Laser Melting–a lean
design approach. Virtual Phys Prototyp. 2019;14:267–76. https://doi.org/10.1080/17452759.201
9.1565596.
8. Hanzl P, Zetková I, Kučerová L. Structural changes and microstructure of maraging steel lattice
structures using additive manufacturing. Manuf Technol. 2019;19:37–41. https://doi.org/10.21062/
ujep/241.2019/a/1213-2489/mt/19/1/37.
9. Tallon J, Cyr E, Lloyd A, Mohammadi M. Crush performance of additively manufactured mar-
aging steel microlattice reinforced plates. Eng Fail Anal. 2020:108. https://doi.org/10.1016/j.
engfailanal.2019.104231.
Si-Containing Titanium Alloys for Laser Powder Bed
Fusion (PBF-L)
Fabian Haase1 (*), Carsten Siemers1, Maximilian Goldapp1, and Joachim Rösler1
1
Technische Universität Braunschweig, Institute for Materials Science, Braunschweig, Germany
F.Haase@tu-braunschweig.de
Abstract. Titanium alloys are very well suited for high-demanding applications in the
transportation industry, but comparable high costs are currently a significant drawback. In
regard to the latter, additive manufacturing, for example, via laser powder bed fusion
(PBF-L) offers several advantages compared to traditional manufacturing. However, a
columnar structure of prior β-grains is formed during printing of conventional titanium
alloys, which usually leads to a crystallographic texture after β to α transformation and,
hence, to anisotropic mechanical properties. The use of printed titanium parts is, conse-
quently, limited to non-safety-critical applications so far. Therefore, in the present study,
silicon-containing titanium alloys with Si contents between 0.4 and 1.5 weight percent
have been developed, produced in laboratory scale, and studied in order to achieve better
solidification conditions. Silicon was used as alloying element because literature research,
on the one hand, suggests silicon as an effective element for β-grain refinement during
conventional casting and additive manufacturing due to its beneficial impact on constitu-
tional supercooling. On the other hand, silicon promotes the formation of Ti5Si3 silicides,
which might act as solidification nuclei. Both effects have been investigated in the present
study in the as-cast state, after multi-track laser-melting experiments conducted on a
PBF-L machine, and via Thermo-Calc© simulations. Additionally, CP-Titanium was ana-
lyzed as reference material. The results obtained indicate an improved solidification of the
studied alloys compared to CP-Titanium during PBF-L due to constitutional supercool-
ing. However, Ti5Si3 formation seems to be unsuitable to promote an equiaxed dendritic
solidification to reduce anisotropy of printed titanium parts.
Keywords: Titanium · Titanium alloy · Alloy development · PBF-L · Silicon ·

Solidification
1 Introduction
Titanium alloys are well suited as components in the transportation industry due to the
highest specific (fatigue) strength among all light metal alloys and high corrosion resis-
tance [1–3]. However, despite these benefits, the use of titanium and titanium alloys, for
example, in the automotive industry is limited due to comparable high costs [3–5].
Additive manufacturing, for example, via laser powder bed fusion (PBF-L) might
change this situation, since PBF-L allows the fabrication of complex geometries with
fewer manufacturing steps and less waste, therefore, leading to lower production costs
https://doi.org/10.1007/978-3-031-17425-4_46
344 F. Haase et al.
[6, 7]. However, during 3D printing of conventional titanium alloys via PBF-L, an epi-
taxial growth and the formation of columnar prior β-grains take place [6–9]. These
columnar grains, caused by, among other reasons, the sharp thermal gradient as well as
the material (lack of constitutional supercooling and of potent solidification nuclei)
[7–9], exhibit a grain orientation with one of the <100> directions aligned approxi-
mately with the building direction [7], which usually leads to anisotropic mechanical
properties [6–9]. However, in terms of Ti-6Al-4V, it was also reported that the mechani-
cal properties do not show a distinct anisotropy [7].
Zhang et al. [9] summarize different strategies for promoting the formation of
equiaxed grains in titanium alloys during fusion-based additive manufacturing. Isotropic
mechanical properties would be achievable by promoting an equiaxed dendritic solidi-
fication [7–9]. For this, in general, a suitable amount of alloying elements leading to
constitutional supercooling as well as potent solidification nuclei are necessary [8, 9].
The lack of the latter seems to be the main reason why equiaxed prior β-grains are not
achieved throughout the microstructure [9, 10]. Among others, silicon was used to
refine the microstructure during conventional casting, for example, by Zhu et al. [11]
and Bermingham et al. [12] as well as additive manufacturing [10]. A refinement of
prior β-grains was observed in all cases [10–12], which was attributed to a higher con-
stitutional supercooling caused by silicon [10, 12]. Additionally, Ti5Si3 could act as
solidification nuclei when present within the melt [12], since coherent relationships
were reported to exist between crystallographic planes of Ti5Si3 and β-titanium [12, 13].
However, in their studies, Bermingham et al. [12] did not find indications of such nucle-
ation effects.
In the present study, silicon-containing titanium alloys with the target composition
Ti-0.44O-0.5Fe-0.08C-0.1Au-(0.4/0.7/1.1/1.5)Si were developed for use within the
PBF-L process. They are based on Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au – an alloy devel-
oped within the authors’ working group for use in medical applications [14]. The pres-
ent alloys as well as a CP-Titanium Grade 1 reference sample were produced in
laboratory scale and studied in the as-cast state as well as after multi-track laser-melting
experiments conducted on a PBF-L machine and via Thermo-Calc© simulations. The
focus was laid on the determination of the role of silicon in terms of constitutional
supercooling and the formation of Ti5Si3 silicides within the liquid as potential solidifi-
cation nuclei during PBF-L to reduce the anisotropy of as-printed titanium parts.
2 Methods
2.1 Simulations
Thermo-Calc© software version 4.1 and TTTI3 database version 3 (2005) were used to
calculate the equilibrium mass fraction of stable phases with respect to temperature
(property diagram simulations), the composition and constitution of phases at specific
temperatures (single equilibrium calculations) as well as phases formed during solidifi-
cation (Scheil-Gulliver simulations; no fast diffusing elements within the solidified
material). In terms of the silicon-containing alloys, the targeted chemical compositions
Si-Containing Titanium Alloys for Laser Powder Bed Fusion (PBF-L) 345
without the alloying element gold and, in terms of CP-Titanium, the chemical composi-
tion according to the provided inspection certificate were considered. Gold had to be
excluded, since this alloying element is not covered by the database. This seems to be
acceptable, as the amount of gold atoms is very low. In terms of property diagram simu-
lations and single equilibrium calculations, a pressure of 1 bar as well as a system size
of 1 mole was used. In case of property diagram and solidification simulations, a tem-
perature step width of 10 K and 0.1 K, respectively, was used. Default values were kept
for all other relevant options. In general, the simulation results obtained have to be
considered as rough estimates, since the studied alloys differ significantly from conven-
tional titanium alloys. Consequently, temperatures simulated were round to integers.
2.2 Alloy Production
Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au pre-melted and cast material and elemental silicon

chips were used to weigh out alloys with the targeted chemical composition of Ti-0.44O-
0.5Fe-0.08C-0.1Au-(0.4/0.7/1.1/1.5)Si using a precision scale with the readability of
0.001 g, which resulted in specimens with an individual mass of approx. 4 g. For com-
parison, a small sample of commercial CP-Titanium (rod material) with the chemical
composition Ti-0.14O-0.09Fe-0.01C was cut off and subsequently processed the same
way. The chemical composition corresponds to CP-Ti Grade 1 according to ASTM
B348-13 [15]. Melting was conducted using a small plasma beam cold hearth melting
facility. Prior to melting, the chamber was evacuated four times to a pressure of 1–6 ∙
10−2 mbar with argon of purity 99.999%. Melting was performed with an argon pres-
sure of approx. 700 mbar. All specimens have been turned and remelted three times to
ensure chemical homogeneity. Prior to each melting, a small piece of titanium foil was
melted first to bind remaining oxygen within the processing chamber. After final melt-
ing, the solidification conditions were kept as constant as possible by keeping the but-
ton shape specimens in the chamber under the protective argon atmosphere for 4 min.
During the first 3 min, the water cooling of the copper plate was kept switched on.
All samples did not show any signs of colorization during each (re-)melting; how-
ever, three samples exhibited a mass increase of up to 0.002 g compared to the weighed
material. Such an increase is often observed and might be caused by either inaccuracies
of the precision scale regarding such small mass changes, or an oxygen pickup of the
samples. In case of the latter, the oxygen content would have increased to approx.
0.49%, 0.46%, and 0.16% in terms of Ti-0.44O-0.5Fe-0.08C-0.1Au-0.7Si, Ti-0.44O-
0.5Fe-0.08C-0.1Au-1.5Si, and CP-Titanium, respectively. These values are calculated
based on the weighed material and the mass change due to melting. Even if this is the
case, it is nevertheless expected that the solidification conditions remained comparable,
although oxygen is also promoting constitutional supercooling [9, 16, 17], since the
oxygen pickup would be small compared to the changes in the amount of silicon. Other
alloying elements differed to the ideal chemical composition by max. 0.01%.
346 F. Haase et al.
2.3 Multi-Track Laser-Melting Experiments
Multi-track laser-melting experiments were conducted on a PBF-L machine (SLM®125

of SLM Solutions Group AG, Germany) equipped with a single laser with a maximum
power of 400 W. Small disks with a thickness of approx. 1–2 mm were cut of the button
shape specimens in the as-cast condition along the solidification direction, of which
surface was subsequently ground with P2500 SiC grinding paper and cleaned in ethanol
using an ultrasonic bath. All specimens were put on a preheated (200 °C) titanium sub-
strate plate within the processing chamber of the PBF-L machine. The chamber was
subsequently flushed with Argon of purity 99.999% several times and once during the
experiments, leading to an oxygen content and pressure of approx. 0.03–0.05% and
approx. 12–17 mbar, respectively, during the experiments. Parallel weld pools with
their longitudinal axis perpendicular to the gas flow direction were generated on each
specimen with a zigzag scanning strategy and within a rectangular field of size approx.
8 × 5 mm. The overall progression of the laser was, thereby, against the direction of the
gas flow. The laser power, hatch distance, and scanning speed were set to 250 W,
0.1 mm, and 1000 mm/s, respectively. These parameters were chosen on the basis of
experiments conducted by Wang et al. [18], who optimized the PBF-L processing
parameters in terms of Ti-6Al-4V. The velocity of the gas flow within the processing
chamber was set to 6 m/s.
Since bulk specimens and no powder bed were used during the experiments, align-
ment of the specimen’s surface toward the focus of the laser was difficult. As a result, it
is expected that (a) the specimen’s surface was not in focus of the laser and that (b) the
focus shift was not constant among the specimens. Additionally, the thickness of indi-
vidual disks differed up to 0.15 mm, which leads to a non-constant position of the focus
of the laser toward the specimen’s surface. These influences may have altered solidifi-
cation conditions but are seen to be small compared to the fact that bulk material was
melted instead of powder.
2.4 Microstructural Investigations and Phase Analysis
For microstructural investigations, the button shape, as-cast specimens were cut in half
at the approximate middle of each specimen along the solidification direction and the
specimens of the multi-track laser-melting experiments were cut approximately per-
pendicular to the melting pools. Afterward, all specimens were warm embedded into
EpoMet™ G and Bakelite compound and were ground with SiC-grinding paper, pol-
ished with MasterPrep™ polishing suspension together with oxalic acid, and etched for
10–20 s with an etchant consisting of 86 mL H2O, 12 mL H2O2, 4.5 mL HNO3, and
5 mL Hf. The exact polishing and etching routine were set with regard to chemical
composition and type (after melting or multi-track laser-melting experiment) of the
specimen.
Microstructure analyses were performed using an Olympus BX51M optical light
microscope with Stream Motion software version 1.7. Panorama images of specimens
were automatically generated by the software consisting of individual images taken
with 50x magnification, which were stitched together automatically. These images were
taken either in the polished state using polarized light to qualitatively study the different
orientations of α-lamellae or α-colonies [5] or after etching using bright field mode. In
terms of Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au and Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si,
the latter panorama images were used to estimate the size of prior β-grains. The other
alloys of the present study were not investigated in this way, since the results obtained
exhibited a great uncertainty, as described later. For measurement, a grid consisting of
seven horizontal lines and 12 vertical lines was placed over the panorama images.
Afterward, the length of segments of these lines falling within prior β-grains was mea-
sured with Fiji software [19]. For high-magnification analysis, SEM images were
obtained by using a Hitachi TM3000 tabletop microscope or a ZEISS LEO 1550
FE-SEM.
Additionally, phase analysis was performed on Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si
in the as-cast state via X-ray diffraction using a GE XRD 3003 PTS diffractometer
(copper anode) operated in Bragg-Brentano setup with the exception that the speci-
men’s tilt angle was set to 40° in order to get sufficient results in terms of peak intensi-
ties. The beam path consisted of a 1 mm pinhole collimator and – in terms of the
secondary side – a 6 mm and 2 mm anti-scatter and receiving slit, respectively (no soller
slit was used). The measurement was conducted over a 2ϴ range of 34–80° with a step
width of 0.005° and an integration time of 5 s. The diffractogram was subsequently
analyzed with PDF2 database (release 2005) and CMPR software [20].
3.1 Simulation Results
According to the property diagram simulations, titanium carbide, TiC, and TiFe should
be stable in all alloys in addition to α- und β-Ti in certain temperature ranges. Moreover,
the formation of Ti5Si3 should be possible in all Si-containing alloys. However, TiC and
TiFe are not present in CP-Titanium Grade 1. Furthermore, these phases as well as
Ti5Si3 were not detected in Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au, an alloy intensively
studied in the working group. Consequently, the simulation results are either inaccurate
at least regarding these phases and alloys, or the precipitation kinetics are very low.
Despite of this, the simulation results indicate – as expected – that an increasing silicon
content increases the temperature range of Ti5Si3 stability as well as the amount, which
could be precipitated, since Ti5Si3 solvus temperature increases from approx. 1325 K
(0.7% silicon) to approx. 1509 K (1.5% silicon) and since the equilibrium mass fraction
of Ti5Si3 increases. In terms of the alloys with 0.7–1.5% silicon, Ti5Si3 solvus tempera-
ture should be 75 K (0.7% silicon)–255 K (1.5% silicon) above β-transus. This suggests
a higher probability of Ti5Si3 formation with increasing silicon content, if kinetic
aspects are disregarded.
The solidification simulations, in which an equilibrium state is always present at the
interface between the solid and the liquid [21], show comparable findings. With regard
to CP-Titanium Grade 1, as expected, only β-phase is stable besides liquid at the inter-
face during solidification and the solidification interval of approx. 38 K is very narrow.
348 F. Haase et al.
Compared to this, the silicon-containing alloys should possess a much higher solidifica-
tion range, which should decrease slightly with increasing silicon content from approx.
533 K (0.4% silicon) to 500 K (1.5% silicon). This increased range is almost exclu-
sively attributed to a much lower solidus temperature. In addition, Ti5Si3 should be
stable at the solid–liquid interface during the last stages of solidification in all silicon-
containing alloys. In terms of Ti-0.44O-0.5Fe-0.08C-0.4Si, Ti5Si3 should become stable
when a mass fraction of solid of approx. 99% is reached. This threshold value is
decreased to approx. 98%, 95%, and 92% in terms of 0.7%, 1.1%, and 1.5% silicon,
respectively. Consequently, an increasing silicon content could lead to a higher amount
of Ti5Si3 before and after complete solidification. Moreover, in terms of the alloys with
0.4% and 0.7% silicon, it has to be noted that the simulations also indicate the stability
of TiC at the interface within a very narrow solidification range. This range corresponds
to a difference of the mass fraction of solid of less than 1.3%.
So far, only the stability of Ti5Si3 was considered. An early stability and nucleation
of Ti5Si3 particles during solidification of the alloy are necessary so that these particles
have a considerable effect on the nucleation of β-grains. Consequently, the present
results indicate that a comparable high silicon content seems to be necessary to increase
the solidification range in which Ti5Si3 is stable at the interface and the formation of
particles is, thus, possible. However, besides stability, kinetic aspects have to be consid-
ered. The formation of Ti5Si3 nuclei and the subsequent nucleation of β-grains on Ti5Si3
particles have to occur considerably fast compared to the growth of existing β-grains in
order that Ti5Si3 particles have a significant effect on the solidification of the alloy.
Despite the lack of information regarding the kinetics of nucleation mechanisms, the
simulation results described above can only be seen as a reference, since the presump-
tions of the Scheil–Gulliver simulations, meaning an infinite diffusion in the liquid
[21], are not valid in terms of PBF-L. Instead, a concentration gradient of the alloying
elements is present in the liquid at the solid–liquid interface [8].
3.2 Microstructure and Phase Analyses
As-cast Condition Figure 1 shows panorama images (polarized light) of CP-Titanium

Grade 1 (a) and Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au (b) in the as-cast condition. As can
be seen, CP-Titanium exhibits coarse α-colonies consisting of individual α-lamellae
with (almost) same orientation of the unit cell [5] and, consequently, only a limited
number of different unit cell orientations leading to a considerable texture and anisot-
ropy. These large α-colonies indicate the presence of coarse and elongated prior
β-grains. This is expected since the alloy only contains a very small amount of oxygen,
carbon, and iron, which would promote constitutional supercooling [9, 16, 17]. In con-
trast, the microstructure of Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au is much finer showing
α-colonies with a much smaller size. Moreover, Fig. 1b seems to indicate that the over-
all amount of unit cell orientations was increased substantially, which could be attrib-
uted to a much higher amount of constitutional supercooling. However, since coarse
and elongated prior β-grains are still visible in the etched condition, which apply to all
silicon-containing alloys, this is also attributed to a change to a Widmannstätten-type
microstructure. This change in microstructure certainly increases the number of
Fig. 1. Microstructure (polarized light) of CP-Titanium Grade 1 (a) and Ti-0.44O-0.5Fe-0.08C-0.4Si-

0.1Au (b) in the as-cast condition.
Fig. 2. Microstructure of Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si in the as-cast condition, (a) optical

micrograph, (b) SEM image (BSE detector, COMPO mode) with arrows indicating small cavities
along grain boundaries.
orientations of α-lamellae or α-colonies and, thus, leads to more isotropic mechanical

properties. However, since the β-grains are still coarse and the Burgers relationship is
obeyed during β to α transformation [5], the effect is limited.
Moreover, Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si exhibited a dendritic solidification as
evident in Fig. 2a: Primary as well as secondary dendrite arms are visible, most likely
due to a different corrosive attack of interdendritic regions due to gradients in chemical
composition. The 1.1% Si-containing sample shows either a dendritic or cellular solidi-
fication in this way. Additionally, compared to CP-Titanium Grade 1, all silicon-
containing alloys show shrinkage porosity at the top or bottom of the button shape
specimens. The appearance of such porosity can mirror the appearance of interdendritic
regions [22], which, therefore, seems to be the case here. This indicates that a transition
from planar to cellular or columnar dendritic solidification was achieved in the silicon-
containing alloys compared to CP-Titanium due to constitutional supercooling.
Consequently, the interdendritic regions have a much lower solidification temperature
due to solute segregation and, in regard to the solidification simulations performed,
Ti5Si3 could form in these regions during solidification, when a sufficient high silicon
350 F. Haase et al.
content is reached, so that Ti5Si3 becomes stable, and when the nucleation rate is high
enough.
In terms of prior β-grain size measurements, an average grain size of roughly 720 μm
and 470 μm was determined for Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au and Ti-0.44O-
0.5Fe-0.08C-0.1Au-1.5Si, respectively. In both cases, the calculated standard deviation
is higher than the average value showing that there are considerable differences in the
line lengths measured within each grain. This can be attributed to actual differences in
grain size but is also caused due to the pronounced columnar structure of some prior
β-grains. Moreover, identification of prior β-grain boundaries was in parts difficult.
Consequently, there is a great uncertainty, but, nevertheless, the results are consistent
with observations in the literature by Zhu et al. [11] and Bermingham et al. [12] in
terms of β-grain size reduction by silicon additions. Therefore, compared to CP-Titanium
Grade 1, a much higher constitutional supercooling seems to be achieved, but a transi-
tion to an equiaxed dendritic solidification with the formation of only equiaxed prior
β-grains was not obtained.
Moreover, the microstructure of Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si consists of
small round and needle-like precipitates present within α-lamellae and along their grain
boundaries, as can be seen in Fig. 2b. These precipitates are forming clusters, which are
not evenly distributed over the longitudinal section. X-ray diffraction analysis of this
specimen suggests that silicides of type Ti5Si3 are present within the sample, as stated
in Fig. 3, which shows the complete as well as a smaller part of the diffractogram
together with ideal peak positions of several phases and their respective PDF2 database
entry number, which were considered during analysis. Identified peaks are labeled with
their respective phase. Two peaks were identified to belong to Ti5Si3. In case of the first
Fig. 3. Diffractogram of Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si in the as-cast state. The insert shows a

smaller part of the diffractogram with ideal peak positions of several phases and their respective PDF2
database entry number. Identified peaks are labeled with their respective phase.
peak, two individual peaks seem to overlap. The bright or darker appearance of precipi-
tates could be attributed to a slight topography contrast of the BSE detector [23] or to
differences in chemical composition, which would imply that the precipitations might
not be solely silicides. In terms of the latter, there is one slight increase of the ambient
noise level of the diffractogram at an angle of approx. 43.2°, which can hardly be
denoted a peak but would fit to the main peak of the intermetallic phase TiFe.
Furthermore, as can be seen in Fig. 3, three peaks cannot be identified. Despite of these
precipitations, the microstructure of all silicon-containing alloys indicates that small
cavities seem to exist along grain boundaries of α-lamellae. Some of these cavities are
marked with arrows in Fig. 2b. It can be seen that they exhibit a comparable size to sili-
cides, which are also present at grain boundaries or within cavities. Cavities along grain
boundaries are also visible regarding Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au. But, how-
ever, in comprehensive previous studies regarding this alloy, there were no indications
of silicide precipitations, as already described. It is, therefore, concluded that cavities
along grain boundaries of α-lamellae are caused due to prolonged etching of the speci-
mens and do not necessarily indicate silicide precipitation. A considerable corrosive
attack of grain boundaries is often observed. Precipitations similar to those in Fig. 2b
were not present in the alloys containing 0.4–1.1% silicon.
Additionally, Fig. 3 indicates that the analyzed volume of the sample exhibits a
strong texture, since the main α-titanium peak {101} at an angle of approx. 40.2°
(according to database) is not the highest peak in the diffractogram. This might be
caused due to use of a comparable small pinhole collimator leading to a small irradiated
surface area in combination with the still comparable high size of former β-grains,
therefore, limiting the number of orientations of α-lamellae. However, similar observa-
tions can be made in terms of Ti5Si3, since the main peak of this phase at an angle of
approx. 41° is not present as well as a second peak at approx. 42.7°, which should both
exhibit a higher peak intensity compared to the identified peaks according to PDF2
database. This might be an indicator that Ti5Si3 has a crystallographic relationship
toward α- or β-Ti in this specimen, which would be consistent with the fact that coher-
ent relationships were already reported to exist between Ti5Si3 and β-Ti [13].
To summarize, the present results show that the alloying element silicon effectively
reduces the size of prior β-grains. However, high amounts of this alloying element seem
to be necessary in order to promote Ti5Si3 formation during solidification. Moreover,
silicides present along α-grain boundaries probably precipitated after β to α transforma-
tion and, hence, had no impact on solidification and nucleation of β-grains.
Multi-Track Laser-Melting Experiments Compared to CP-Titanium, the micro-
structure of the melting pools of the silicon-containing alloys is much finer, showing
martensite, whereas the microstructure of CP-Titanium shows more similarities to a
lamellar than to an acicular martensitic microstructure. Additionally, in terms of the
silicon-containing alloys, the first and, partly, the last melting pools generated by the
laser exhibit a significantly finer microstructure compared to the rest of the scanning
field. This might be caused due to different thermal influences. Due to the findings in
terms of silicide precipitation within Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si in the as-cast
state, the melting pools of this alloy were investigated in detail in the present study.
Figure 4a shows an SEM image (BSE detector, COMPO mode) of the microstructure
352 F. Haase et al.
within the first melting pool of this alloy at a position closer to the melting pool bound-
ary. Several features can be observed, which have been partly marked in Fig. 4a. On the
one hand, prior β-grain boundaries marked with arrows (a) and martensite plates with
widths of several hundred nanometers are present. On the other hand, bright, elongated
superstructures – partly indicated by arrows (b) – are present, which seem to be crossed
by martensitic plates. These structures can also be observed in the alloys with 0.7% and
1.1% silicon. Figure 4b shows a high-magnification image (InLens detector) of these
structures in terms of the 1.1% Si-containing alloy. As can be seen, these structures are
topographical features and differ from polishing artifacts, which are also present. Since
these superstructures are well aligned and differ in orientation in terms of the two prior
β-grains indicated in Fig. 4a, these structures might indicate a cellular solidification.
The topographical contrast would be the result of a different corrosive attack of inter-
cellular regions during specimen preparation.
Additionally, in terms of Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si, small precipitates (not
present in Fig. 4) are present beside of martensite close to the base metal, which was not
remelted or heat treated above β-transus during PBF-L. Although phase analysis via
X-ray diffraction or chemical analysis via EDS measurement was not possible, based
on the results obtained in terms of the as-cast state, it is assumed that these precipitates
are silicides. These precipitations can be observed at different locations, but only at or
close to the transition to the base material. Consequently, and according to the property
diagram simulations, it is believed that the temperature at these locations during PBF-L
was lower than Ti5Si3 solvus temperature but above β-transus, so that the silicides were
not dissolved, but martensite formed during cooling.
Furthermore, small round cavities are visible alongside of martensite plates in all
silicon-containing alloys, especially in the first melting pool of Ti-0.44O-0.5Fe-0.08C-
0.1Au-1.5Si. These could be artifacts from metallographic sample preparation, for
example, due to a prolonged etching time, which is more likely than being leftovers of
silicides, which were etched out. Consequently, it is assumed that either the silicon
content or the nucleation rate was not high enough for (a detectable) Ti5Si3 precipitation
during solidification under PBF-L solidification conditions. However, higher silicon
contents would probably have detrimental effects on mechanical properties, especially
Fig. 4. Microstructure of the first melting pool of (a) Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si (BSE detec-
tor, COMPO mode) and (b) Ti-0.44O-0.5Fe-0.08C-0.1Au-1.1Si (InLens detector) after conducting
multi-track laser-melting experiments.
if a subsequent annealing treatment for example during HIPing has to be performed.

Therefore, in situ formation of Ti5Si3 during solidification to promote an equiaxed den-
dritic solidification seems to be an unsuitable method for reducing the anisotropy of
PBF-L-printed titanium parts.
4 Conclusion
This study shows that the analyzed silicon-containing alloys, especially in terms of
Ti-0.44O-0.5Fe-0.08C-0.1Au-1.1Si and Ti-0.44O-0.5Fe-0.08C-0.1Au-1.5Si, exhibit a
significant higher amount of constitutional supercooling in the as-cast state compared
to CP-Titanium Grade 1. Moreover, the much finer colony size reduces the anisotropy
to a certain, but limited extent. In terms of the performed multi-track laser-melting
experiments, there are indications of a cellular solidification within the first melting
pool regarding the alloys with 0.7%, 1.1%, and 1.5% silicon. Nevertheless, despite of
undissolved Ti5Si3 near the transition to the base material, there are no clear indications
of silicide precipitation within the melting pools. As a result, Ti5Si3 formation during
solidification seems to be unsuitable to promote an equiaxed dendritic solidification and
to reduce the anisotropy of PBF-L-printed titanium parts.
Further studies will focus on finding suitable solidification nuclei for the β-phase to
promote an equiaxed dendritic solidification during PBF-L rather than promoting con-
stitutional supercooling. Different potential nuclei are described in the literature, sum-
marized for example by Zhang et al. [9]. The addition of low amounts of rare-earth
elements to, for example, Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au, are of particular
interest.
Acknowledgments The present study is part of project IGF 21671 N of the AiF Arbeitsgemeinschaft
industrieller Forschungsvereinigungen “Otto von Guericke” e.V. This project is funded by the Federal
Ministry for Economic Affairs and Climate Action on the basis of a resolution of the German Bundestag,
which is gratefully acknowledged. Furthermore, the authors would like to thank F. Graef and
S. Sternberg for assistance in terms of XRD measurements and specimen preparation, respectively.
References
1. Peters M, Hemptenmacher J, Kumpfert J, Leyens C. Structure and properties of titanium and titanium
alloys. In: Leyens C, Peters M, editors. Titanium and titanium alloys: fundamentals and applica-
tions. Weinheim, Germany: Wiley-VCH; 2003. p. 1–36. https://doi.org/10.1002/3527602119.ch1.
2. Peters M, Kumpfert J, Ward CH, Leyens C. Titanium alloys for aerospace applications. In: Leyens
C, Peters M, editors. Titanium and titanium alloys: fundamentals and applications. Weinheim,
Germany: Wiley-VCH; 2003. p. 333–50. https://doi.org/10.1002/3527602119.ch13.
3. Schauerte O. Titanium in automotive production. In: Leyens C, Peters M, editors. Titanium and tita-
nium alloys: fundamentals and applications. Weinheim, Germany: Wiley-VCH; 2003. p. 467–82.
https://doi.org/10.1002/3527602119.ch18.
354 F. Haase et al.
4. Peters M, Leyens C. Non-aerospace applications of titanium and titanium alloys. In: Leyens
C, Peters M, editors. Titanium and titanium alloys: fundamentals and applications. Weinheim,
Germany: Wiley-VCH; 2003. p. 393–422. https://doi.org/10.1002/3527602119.ch15.
5. Lütjering G, Williams JC. Titanium. 2nd ed. Berlin: Springer; 2007. https://doi.
org/10.1007/978-3-540-73036-1.
6. Gorsse S, Hutchinson C, Gouné M, Banerjee R. Additive manufacturing of metals: a brief review
of the characteristic microstructures and properties of steels, Ti-6Al-4V and high-entropy alloys.
Sci Technol Adv Mater. 2017;18:584–610. https://doi.org/10.1080/14686996.2017.1361305.
8. StJohn DH, McDonald SD, Bermingham MJ, Mereddy S, Prasad A, Dargusch M. The challenges
associated with the formation of equiaxed grains during additive manufacturing of titanium alloys.
Key Eng Mater. 2018;770:155–64. https://doi.org/10.4028/www.scientific.net/KEM.770.155.
9. Zhang D, Prasad A, Bermingham MJ, Todaro CJ, Benoit MJ, Patel MN, Qiu D, StJohn DH, Qian
M, Easton MA. Grain refinement of alloys in fusion-based additive manufacturing processes.
Metall Mater Trans A. 2020;51:4341–59. https://doi.org/10.1007/s11661-020-05880-4.
10. Mereddy S, Bermingham MJ, StJohn DH, Dargusch MS. Grain refinement of wire arc additively
manufactured titanium by the addition of silicon. J Alloy Compd. 2017;695:2097–103. https://doi.
org/10.1016/j.jallcom.2016.11.049.
11. Zhu J, Kamiya A, Yamada T, Watazu A, Shi W, Naganuma K. Effect of silicon addition on micro-
structure and mechanical properties of cast titanium alloys. Mater Trans. 2001;42:336–41. https://
doi.org/10.2320/matertrans.42.336.
12. Bermingham MJ, McDonald SD, Dargusch MS, StJohn DH. The mechanism of grain refinement of
titanium by silicon. Scr Mater. 2008;58:1050–3. https://doi.org/10.1016/j.scriptamat.2008.01.041.
13. Choe B-H, Kim J-W, Jeong H-W, Hyun Y-T, Kim S-E, Lee Y-T. On phase relationship between
Ti-Si precipitates and beta matrix in Ti-6Al-4Fe-0.5Si alloy. Korean J Met Mater. 2004;42:1–6.
14. Haase F, Siemers C, Klinge L, Lu C, Lang P, Lederer S, König T, Rösler J. Aluminum- and
vanadium-free titanium alloys for medical applications. MATEC Web Conf. 2020;321:05008.
https://doi.org/10.1051/matecconf/202032105008.
15. ASTM B348-13. Standard specification for titanium and titanium alloy bars and billets. West
Conshohocken, PA: ASTM International; 2013. https://doi.org/10.1520/B0348-13.
16. Bermingham MJ, McDonald SD, StJohn DH, Dargusch MS. Beryllium as a grain refiner in tita-
nium alloys. J Alloy Compd. 2009;481:L20–3. https://doi.org/10.1016/j.jallcom.2009.03.016.
17. Bermingham MJ, McDonald SD, Dargusch MS, StJohn DH. Grain-refinement mechanisms in
titanium alloys. J Mater Res. 2008;23:97–104. https://doi.org/10.1557/JMR.2008.0002.
18. Wang Z, Xiao Z, Tse Y, Huang C, Zhang W. Optimization of processing parameters and establish-
ment of a relationship between microstructure and mechanical properties of SLM titanium alloy.
Opt Laser Technol. 2019;112:159–67. https://doi.org/10.1016/j.optlastec.2018.11.014.
19. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden
C, Saalfeld S, Schmid B, Tinevez J-Y, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona
A. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9:676–82.
https://doi.org/10.1038/nmeth.2019.
20. Toby BH. CMPR – a powder diffraction toolkit. J Appl Crystallogr. 2005;38:1040–1. https://doi.
org/10.1107/S0021889805030232.
21. Porter DA, Easterling KE, Sherif MY. Phase transformations in metals and alloys. 3rd ed. Boca
Raton, FL: CRC Press; 2009.
22. Stefanescu DM. Computer simulation of shrinkage related defects in metal castings – a review. Int
J Cast Met Res. 2005;18:129–43. https://doi.org/10.1179/136404605225023018.
23. Goldstein JI, Newbury DE, Michael JR, Ritchie NWM, Scott JHJ, Joy DC. Scanning elec-
tron microscopy and X-ray microanalysis. 4th ed. New York, NY: Springer; 2018. https://doi.
org/10.1007/978-1-4939-6676-9.
Aluminum Auxetic Structures Produced Using
Combined Stereolithography and Investment Casting
N. Alfano1 (*), H. Simha1, and A. Elsayed1

1
nalfano@uoguelph.ca
Abstract. Investment casting produces structurally complex metal parts with good sur-
face finish and attractive mechanical properties. This study combines polymer additive
manufacturing through the utilization of a stereolithography 3D printer and metal invest-
ment casting to produce complex geometry and energy absorption structures. Investment
cast 319 aluminum alloy with several investment-mold temperatures, melt-pouring tem-
peratures, and vacuum levels was used to produce open-celled, non-stochastic auxetic
structures. The aluminum auxetic castings are produced using higher vacuum levels, and
higher investment-mold temperatures had fewer filling defects and most closely matched
the geometry of the stereolithography model. Typical casting defects included the pres-
ence of shrinkage pores and misruns at areas far away from the casting sprue. It is
expected that optimization of the casting parameters can help eliminate these casting
defects. The resulting aluminum auxetic structures and their production route offer an
accessible method to produce high-integrity, geometrically complex castings for trans-
portation applications.
Keywords: 3D printing · Aluminum alloys · Auxetic · Stereolithography ·

Investment casting
1 Introduction
Achieving a structure capable of utilizing minimal material without compromising

mechanical properties is a task sought out by many modern-day researchers in the auto-
motive, aerospace and medical industries [1]. Ideally, desirable properties are obtained
while also reducing expenditures on production costs as well as energy of production.
Three-dimensional, open-celled lattice structures are a concept that in recent years has
been heavily studied due to their ability to maintain mechanical strength, high modulus,
wear resistance, and thermal stability among others [2]. Several structural designs have
been examined with auxetic configurations showing high energy absorption capabilities
[3]. Auxetic materials or structures have a negative Poisson’s ratio, causing them to
expand in the lateral direction when stretched and shrink when compressed [4].
Researchers have utilized direct metal additive manufacturing, or fused deposition
modeling (FDM) to produce a polymer template and then utilized investment casting to
produce the final metal part. However, both of these approaches have high costs, suffer
from quality/surface finish issues as well as defects on a macroscale in the final product
https://doi.org/10.1007/978-3-031-17425-4_47
356 N. Alfano et al.
[5]. Stereolithography (SLA) using a castable resin and then subsequent investment
casting results in a highly detailed cast part and can be produced with minimal equip-
ment costs and allows for the use of the large catalog of existing casting alloys [6]. This
research examines the possibility to use combined SLA with investment casting to pro-
duce complex aluminum auxetic structures.
2.1 Mold Preparation
The procedure to prepare the aluminum auxetic structures is shown in Fig. 1. The open-
celled, non-stochastic auxetic structures were initially designed using the Solidworks®
Computer-Aided Design (CAD) modeling software having struts 1 mm thick and an
overall dimension of 50 × 50 × 30 mm. With the prepared CAD model, a Standard
Tessellation Language (STL) file format was exported and sliced into printing layers
using the Chitubox® software so that it can be imported to a Phrozen Mighty 4 K SLA
printer. The SLA print material was Resinworks 3D’s EasyCast castable resin. The print
parameters included a layer height of 0.05 mm and lifting speed of 60 mm/s.
The investment mold was prepared by mixing and vacuuming a 42/100 ratio of water
to Ransom & Randolph’s Plasticast investment utilizing a hand mixer and vacuum
chamber at 100 kPa. This allowed for the entire steel flask containing the SLA-printed
model to fill entirely. After leaving the investment mold to set at room temperature for
2 h, the investment mold begun to cure where it was then placed in a kiln to run a burn-
out process as per Resinworks 3D’s time and temperature specifications starting from
150 °C ramping to 730 °C and ending at 538 °C, having a ramp rate of 2 °C/min.
2.2 Investment Casting Process
The 319 aluminum alloy was melted in an electric-resistance furnace to a temperature

of 720 °C. Once melted, it was subsequently degassed, grain refined using 1 wt.% of
Al-5Ti-1B grain refiner, and modified using an Al-10 wt.% Sr master alloy to bring the
Sr concentration to 0.015 wt.%. The investment-mold temperature was 700 °C and the
melt was poured at 720 °C while being subjected to a continuous vacuum pressure of
80 kPa. The poured melts were left undisturbed for a duration of 15 min before quench-
ing in water to remove the investment mold and subsequently cool the sample.
Fig. 1. Process steps to produce aluminum auxetic structures using SLA and investment casting.
Aluminum Auxetic Structures Produced Using Combined Stereolithography… 357
2.3 Compression Testing
The compression tests of the casted auxetic samples were conducted on an Instron 5969
with 50 kN load cell. A plot of applied load with respective extension was collected.
The displacement rate was 2.5 mm/min. Molybdenum disulfide lubricant paste was
applied to both the top and bottom faces of the sample as well as the compression plates
to limit the friction during testing.
Figure 2 is the auxetic sample during compression testing, specifically at its max load
of 6358 N which takes place at an extension of 1.38 mm. From Solidworks®, the model
mass and volume were found to be 26.84 grams and 9620.11 mm3, respectively, yield-
ing a density of 0.00279 g/mm3. This having a total volume of 88% less than that of a
filled, non-porous model of the same overall dimensions and composition.
Figure 2b depicts the plot of load versus extension as the sample underwent com-
pressive loading. It is evident through this plot and figure that the specimen is exhibiting
traits of brittle failure with sections twisting and buckling rather than compressing lead-
ing to quick failure. This twisting and buckling of particular struts resulted in the loads
being transferred to other struts and a quick reduction in load with extension. It is
anticipated that the cause of failure within the struts was related to turbulent filling dur-
ing casting, resulting in pores and misruns that compromised struts within the sample.
A line of best fit based on a level-six polynomial function was generated and integrated
to determine the area under the plot from zero load to 10% beyond that of the max load
attained. This estimated that the energy absorbed during compression was
11,604 J. Improvements to the energy absorption capabilities of the auxetic structures
can be made by using a more ductile alloy composition. With optimized temperatures
and vacuum levels, the struts can deform in unison without cracking resulting in uni-
form deformation and higher energy absorption.
Acknowledgments The authors would like to thank the Natural Science and Engineering Research
Council of Canada for financial support and machine shop support personal at the University of
Fig. 2. (a) Compression testing snapshot at a max load of 6358 N occurring at an extension of
1.38 mm. (b) Load versus displacement plot.
358 N. Alfano et al.
Guelph. The authors also thank S. Kotiadis and M. Zavala-Serret for their help during the
experiments.
References
1. Yan C, Hao L, Hussein A, Bubb SL, Young P, Raymont D. Evaluation of light-weight AlSi10Mg
periodic cellular lattice structures fabricated via direct metal laser sintering. J Mater Process
Technol. 2014;214(4):856–64. https://doi.org/10.1016/j.jmatprotec.2013.12.004.
2. Gecu R, Karaaslan A. A comparative study on titanium-reinforced aluminium matrix composites
produced by melt infiltration casting and squeeze infiltration. Int J Met. 2019;13(2):311–9. https://
doi.org/10.1007/s40962-018-0253-0.
3. Carneiro VH, Puga H, Meireles J. Positive, zero and negative Poisson’s ratio non-stochastic metal-
lic cellular solids: dependence between static and dynamic mechanical properties. Compos Struct.
2019;226:111239. https://doi.org/10.1016/j.compstruct.2019.111239.
4. Xue Y, Wang W, Han F. Enhanced compressive mechanical properties of aluminum based aux-
etic lattice structures filled with polymers. Compos Part B Eng. 2019;171:183–91. https://doi.
org/10.1016/j.compositesb.2019.05.002.
5. Carneiro VH, Rawson SD, Puga H, Meireles J, Withers PJ. Additive manufacturing assisted invest-
ment casting: a low-cost method to fabricate periodic metallic cellular lattices. Addit Manuf.
2020;33:101085. https://doi.org/10.1016/j.addma.2020.101085.
6. Pattnaik S, Karunakar DB, Jha PK. Developments in investment casting process—a review. J Mater
Process Technol. 2012;212(11):2332–48. https://doi.org/10.1016/j.jmatprotec.2012.06.003.
High-Strength, Low-Modulus Nanostructured: Ti
13Nb 13Zr Alloy
L. Klinge1 (*), C. Siemers1, and C. Jahnke1

1
TU Braunschweig – Institute for Materials Science, Braunschweig, Germany
lina.klinge@tu-braunschweig.de
Abstract. High-strength nanostructured titanium alloys with a low Young’s modulus

play a key role in biomedical applications, especially in the field of dentistry. On the
one hand, the high strength avoids implant and abutment breakage, and on the other
hand, the low Young’s modulus improves the growth of bone cells as stress shielding
is diminished. Furthermore, the nanostructured implant surface also improves the
adherence of bone cells and additionally reduces the colonization of bacteria which
in turn reduces the occurrence of the so-called periimplantitis. Periimplantitis is a
disease which in the worst case can lead to the loosening and loss of the implant. By
using the Ti 13Nb 13Zr (TNZ) alloy, these risks can be minimized.
TNZ is a second-generation biomedical β-rich (α + β)-titanium alloy with an
improved biocompatibility as critical alloying elements such as aluminium and vana-
dium are not present. In addition, the corrosion resistance is improved compared to
the standard Ti 6Al 4 V alloy used for implants. Nanostructured materials can be
produced through Severe Plastic Deformation (SPD) processes like Equal Channel
Angular Pressing (ECAP), which was used for this study. To strengthen the material
and to regain a sufficient elongation at fracture to fulfil the requirements of related
medical standards, the material must be recrystallized (short-time heat treatments)
and precipitation hardened after the forming process. Thereby, different phase consti-
tutions, namely α- and β-phase as well as α″-martensite, can be used to optimize the
Young’s modulus and the mechanical properties.
The studied effects should be transferrable to other β-rich (α + β)-titanium alloys
like Ti 6Al 2Sn 4Zr 6Mo which is applied in aircraft engines so that the thermo-
mechanical treatment strategies presented here are of interest in other fields of appli-
cation as well, e.g. in the transportation industry.
Keywords: Dental implants · ECAP · High strength · Low-modulus

Nanostructured titanium · Severe Plastic Deformation (SPD) · Titanium · Titanium
alloys · Ti 13Nb 13Zr · TNZ
1 Introduction
Amongst all metals, titanium has the highest specific fatigue strength, it is biocompat-
ible and has excellent corrosion resistance which makes titanium and its alloys attrac-
tive for many applications, e.g. in the transportation sector (especially for aerospace
applications), the chemical industry, or for biomedical applications [1]. As titanium
https://doi.org/10.1007/978-3-031-17425-4_48
360 L. Klinge et al.
alloys can be used at elevated temperatures up to 550 °C, they are applied in low-
pressure compressors in aircraft engines. Also, first 3D-printed titanium parts like
locket shafts or brackets find their application in aviation and, therefore, in transporta-
tion industry [2].
In addition to the before mentioned characteristics, if biomedical applications are
concerned, a low Young’s modulus close to that of the bones (5–20 GPa) is aimed for to
avoid stress shielding. As titanium alloys can be composed of different equilibrium (α
and β) and non-equilibrium (α´- and α″-martensite) phases, a wide range of Young’s
moduli (50–120 GPa) can be obtained through optimized heat treatments, which is a
major advantage compared to other standard medical alloys like steels and CoCr alloys
(approx. 210 GPa). The Ti 13Nb 13Zr (TNZ) alloy used in this study is a second-
generation biomedical β-rich (α + β) titanium alloy which is standardized in ASTM
F1713. Due to the avoidance of aluminium (Al) and vanadium (V), which are consid-
ered to cause severe health issues [3], the biocompatibility of TNZ is improved com-
pared to commercially used alloys for implant applications like Ti 6Al 4 V (ELI) or Ti
6Al 7Nb. Also, the corrosion resistance of TNZ is better than that of Ti 6Al 4 V (ELI)
[4].
Under equilibrium conditions, TNZ can consist of α- and β-phase. After fast cooling,
i.e. water quenching from temperatures close to or above β-transus temperature, the
β-phase can transform to α″-martensite. Compared to α- and β-phase, the α″-martensite
is soft with good ductility and a very low Young’s modulus [5]. Therefore, the
α″-martensite is, on the one hand, good formable which allows processing even at low
temperatures, and on the other hand, the low Young’s modulus should diminish the
stress shielding effect when used as implant material. As the strength of the α″-martensite
is too low, a fully martensitic microstructure is not applicable for dental implants. To
strengthen the material and to fulfil the requirements of ASTM F1713, optimized heat
treatments can be used to establish microstructures consisting of different amounts of
primary-α (αP) and β-phase as well as α″-martensite. Ageing of these microstructures
leads to the precipitation of secondary-α (αS) and/or a (partial) martensite decomposi-
tion (α″ → α + β). Besides of the strengthening, the ageing process also leads to further
refinement (i.e. nano-structuring) of the microstructure, which is essential for the
osseointegration [6].
Standard alloys used for dental implant applications are CP-Titanium Grade 4
(implant) and Ti 6Al 4 V ELI (abutment). These alloys possess a comparably high
Young’s modulus of approx. 105–110 GPa [7] which can lead to stress shielding, bone
degradations, and inflammations can occur due to bacterial infections. In addition, due
to the difference in strength, wear of the softer implant (typically at the upper edge) can
occur as a result of friction during relative movements of implant and abutment (e.g. if
screw loosening occurred) so that titanium particles as well as titanium, aluminium, and
vanadium ions might be released. To ensure a good bone cell adhesion, the implants are
either etched or coated to establish a nanostructured surface to which the bone cells can
easily attach and grow. In case stress shielding occurs, bacteria can colonize in the gaps
between the implant and the bone together with the formation of a carbon-rich biofilm,
the so-called periimplantitis [8]. The biofilm shields the bacteria and, at the same time,
prevents the bone from re-growing on the implant surface. The periimplantitis therapy
includes the use of antibiotics and mechanical cleaning (grinding and polishing) of the
High-Strength, Low-Modulus Nanostructured: Ti 13Nb 13Zr Alloy 361
implant surface to remove the biofilm. At the same time, etched surfaces and previously
applied coatings are being removed and (re-) osseointegration becomes more difficult.
To ensure a good bone cell adhesion and to avoid bacterial colonization after implan-
tation, a per se nanostructured microstructure (grain size smaller than 1 μm) is aimed
for [9]. Together with a low Young’s modulus stress shielding and therefore, the risk of
periimplantitis should be minimized. Even if a periimplantitis therapy becomes neces-
sary, a nanostructured surface would redevelop naturally after mechanical cleaning of
the implant surface, because of the different mechanical properties of the grains and the
grain boundaries, as well as the different phases, and the bone cells can re-attach more
easily.
To obtain a fully nanostructured material, Severe Plastic Deformation (SPD) can be
used. Due to high shear stresses, the initially coarse-grained material becomes nano-
structured. A commonly used method is the so-called Equal Channel Angular Pressing
(ECAP) in which a bar is pressed through a die with an angle up to 90°. The ECAP
process requires high forming forces and friction occurs, which is a great challenge for
producing nanostructured titanium and the process length is limited. An alternative
process which is still under development is the so-called Equal-Channel-Angular-
Swaging (ECAS) process [10].
In this study, TNZ bars produced by Conform-ECAP, a modified ECAP process,
were subjected to heat treatments to determine the parameters to receive the most suit-
able mechanical properties for use as dental implants. This includes a high strength, low
modulus, and sufficient elongation at fracture together with a nanostructured (grain size
smaller than 1 μm) microstructure.
The material used for this study was delivered by Shaanxi Yunzhong Industry
Development Co., LTD, China, according to ASTM F1713 specification. The chemical
composition is stated in Table 1. The diameter of the bars was 10 mm.
The Conform-ECAP was done at COMTES FHT, Czech Republic. After ECAP, the
bars were rotary swaged from 10 mm diameter to 6 mm diameter as this would be a
typical semi-finished product for implant production.
From the ECAP-ed rods, small cylinders (approx. 8 mm in length) were cut using a
water-cooled precision cutting machine (JeanWirtz, Cuto 20) with corundum cutting
discs from ATM.
For the subsequent heat treatments, chamber furnaces Carbolite CWF 1300 using
standard air atmosphere were used. Recrystallization took place at 700 °C between 10
and 30 min with subsequent water quenching (WQ). The ageing was performed between
300 and 600 °C and holding times of 15 min to 24 h with subsequent air cooling (AC).
Table 1. Chemical composition of Ti 13Nb 13Zr in wt-%.

Ti Nb Zr O Fe C N H
Balance 12.99 13.15 0.053 0.022 0.005 0.010 0.001
Microstructural investigations were done on cross sections. Samples were embedded

into EpoMet® amorphous, glass fibre-reinforced polymer by warm embedding (Buehler,
SimpliMet 4000). Afterwards, the samples were prepared by water-cooled mechanical
grinding with SiC-grinding papers (size P320, P400, P600, P800, P1200, and P2500)
and polishing (9 μm, 6 μm, and 3 μm with Kulzer NewLam® diamond suspension
including lubrication followed by OPS polishing with OPS (25 ml) + H2O (10 ml) + KOH
(0.7 g) + distilled water). Finally, etching was carried out using a special titanium-
etching reagent (86 ml H2O + 12 ml H2O2 + 4.5 ml HNO3 + 5 ml HF) for 3–5 s. The
fully α″-martensitic microstructure was etched with 30 ml H2O + 60 ml H2O2 + 10 ml
HF for 20 s.
The microstructure has been analysed by means of optical microscopy (ZEISS
Imager.M2m) and scanning electron microscopy (LEO 1550) using an In-lens detector.
Hardness tests were carried out using an automated Vickers hardness tester (LECO
LV100AT) with HV10 (10 indentations per specimen distributed along two straight
lines at the cross sections) and an indentation time of 15 s.
Tensile tests were carried out according to DIN EN ISO 6892-1 by a Zwick/Roell
universal tensile testing machine Z.MART.PRO at room temperature together with the
testXpert software by Zwick. Tensile test specimens were produced according to DIN
50125 B 4 × 20.
3.1 Initial Microstructure
The initial microstructure in the “as-received” (AR) state is shown in Fig. 1a. The
microstructure consists of lamellar primary-α (αP) and β-phase. Within the β-phase, fine
dispersed secondary-α (αS) phase has been precipitated, which indicates a slow cooling
after annealing (the exact heat treatment has not been recorded by the material sup-
plier). The average hardness is 258 HV10. A first attempt of ECAP forming was made
on the AR state which led to the destruction of the forming tools as high forces became
necessary and strong wear occurred. The hardness after ECAP forming is 284 HV10.
The microstructure presented in Fig. 1b shows a slight deformation of the material, but
some αp-lamellae are still present. Also, some relatively coarse (>1 μm) αp-grains are
visible. That indicates that the ECAP forming of the initial microstructure was insuffi-
cient to break up all the αp-lamellae or to establish a fine-grained microstructure after
one pass.
As the forming process leads to the destruction of the forming tools, the initial
microstructure was heat treated at 750 °C for 30 minutes with subsequent water quench-
ing to establish a soft, orthorhombic α″-martensitic structure (Fig. 2a) which is a good
formable microstructure due to its moderate strength (approx. 632 MPa) and good elon-
gation at fracture (approx. 23%).
Fig. 1. SEM images: (a) Initial microstructure in the “as-received” state. (b) Microstructure after one
pass of Conform-ECAP with a tool angle of 90°.
Fig. 2. (a) OM image: α″-martensitic starting structure, the α″-martensite platelets have built in for-
mer β-grains; (b) SEM image: Microstructure after one pass of Conform-ECAP (90°) with subsequent
rotary swaging to diameter 6 mm. The microstructure consists of fine- dispersed α-(dark) and β-(bright)
phases.
3.2 Martensitic Starting Structure and Microstructure After ECAP Forming
In the optical microscope (OM) image of the α″-martensitic starting structure in Fig. 2a,
the former β-grain boundaries are still visible. The grain size was approx. 100 μm. The
hardness of the α″-martensite is 192 HV10 and, therefore, very soft.
During Conform-ECAP (Fig. 2b), the α″-martensite platelets break up into small
α- and β-grains. Also, due to higher temperatures caused by friction, the α″-martensite
started to decompose. Therefore, after forming, a fine-grained microstructure consisting
of α- and β-phase is present. The grain size is approx. 200 nm and the hardness increased
to 306 HV10.
In Table 2, the tensile test results of the α″-martensitic starting structure and the
ECAP-ed + rotary swaged state are shown. Also, the requirements of ASTM F1713 are
stated. The results showed the necessity of post-heat treatments of the ECAP-ed micro-
structure. To fulfil the requirements of ASTM F1713, an ultimate tensile strength (UTS)
of 860 MPa at an elongation at fracture of minimum 8% is required. The Young’s modu-
lus and the yield strength (YTS) are not specified in the ASTM standard, but FEM analy-
Table 2. Tensile test results of the α″-martensitic starting structure and the ECAP-ed + rotary-swaged
state. Also, the requirements of ASTM F1713 are stated.
E YTS UTS A
GPa MPa MPa %
ASTM F1713 n. def. n. def. 860 8
α″-martensitic-starting structure 58 439 632 23
Microstructure after ECAP and rotary swaging 74 1015 1109 5
ses of the bone-implant interfaces by our project partner ADVANTIQX (to be published)
showed that Young’s modulus between 70 and 80 GPa is a good compromise to prevent
stress shielding (low Young’s modulus required) and overloading of the newly formed
bone (high Young’s modulus required) at the bone-implant interface.
The α″-martensitic starting structure shows a very low Young’s modulus, which is
typical for the α″-martensite [5]. The tensile strength is moderate (approx. 632 MPa) at
a large elongation at fracture of approx. 23%, which is the most important factor for the
good formability. After the forming process, the Young’s modulus increases to 74 GPa
due to the newly formed α + β microstructure. At the same time, the strength increases
to more than 1100 MPa, but the elongation at fracture decreases to 5% due to work
hardening and is, therefore, not sufficient to comply with ASTM F1713.
3.3 Heat Treatments
To improve the elongation at fracture, recrystallization of the deformed material is nec-

essary. The Severe Plastic Deformation leads to an increased dislocation density which
in turn should provide a large number of recrystallization seeds [10]. Hence, short hold-
ing times during the recrystallization treatment are necessary to prevent grain growth.
The overall aim of the multi-step heat treatments is to achieve the highest possible
strength with sufficient elongation at fracture, a low Young’s modulus together with a
nanostructured microstructure. Therefore, in the first step, the recrystallization sets the
αP-grain size, removes the work hardening and simultaneously improves the elongation
at fracture. But at the same time, the tensile strength decreases. Thus, in a second step,
a precipitation hardening becomes necessary, to re-strengthen the material. In addition,
due to αS-precipitations and/or α″-martensite decomposition, the microstructure
becomes nanostructured again. Finally, after the heat treatment procedure is completed,
the material must fulfil the requirements of ASTM F1713.
Previous studies [11] showed that only recrystallization treatments slightly below
β-transus (β-transus ≈ 740 °C) are applicable, as after quenching the amount of
α″-martensite is high enough to decrease the Young’s modulus. A heat treatment above
β-transus completely resolves the αP and instantaneously leads to grain growth.
A recrystallization temperature of 700 °C was chosen, because after water quench-
ing (WQ), a microstructure consisting of αP (low amounts of αP are recommended to
improve the fatigue properties [1]) and α″-martensite is present, which possesses a very
low Young’s modulus and gives the opportunity for a (partial) martensite decomposi-
tion during ageing. To prevent severe grain growth, the holding times still have to be
kept as short as possible but need to be long enough to allow industrial production of
dental implants. Therefore, holding times of 10–30 min (5 min steps) have been chosen.
But, as can be seen from Fig. 3, severe grain growth has already occurred after 10 min.
However, the hardness values only vary between 198 and 204 HV10 for the recrys-
tallized microstructures which lies within the 3% measurement uncertainty of the hard-
ness tester. However, tensile tests showed that the microstructure after 10 min holding
time has the highest strength (725 MPa) due to the smallest grain size, while the Young’s
modulus (average of 64 GPa) and the elongation at fracture (average of 19.5%) are
quite similar between the different holding times. Because of the small grain size and
the highest strength, the heat treatment at 700 °C and 10 min holding time was chosen
to evaluate the most suitable parameter set for the ageing process.
For the ageing, temperatures between 300 and 600 °C were applied on the recrystal-
lized material for 1 h with subsequent air cooling (AC) to allow an α″-martensite
decomposition into small α- and β-grains. Figure 4 shows the hardness values and the
resulting microstructures after ageing at 400, 500, and 600 °C. Compared to the recrys-
tallized microstructure (700 °C/10 min/WQ – 208 HV10), the hardness increased at any
ageing temperature. In addition, the hardness increases steadily with increasing ageing
temperature until a maximum is reached at 500 °C, followed by overageing of the struc-
ture. As can be seen from Fig. 4, the microstructural investigations supported the hard-
ness measurements. After ageing at 400 °C, the αP-grains are approx. 800 nm in
diameter and the soft α″-martensite has not been fully decomposed (darker spots within
the former β-grains). At 500 °C, the αP-grains are smaller (approx. 500 nm) and the
α″-martensite has been fully decomposed into very small α- and β-grains and, therefore,
reaches the maximum in hardness due to the very fine microstructure. After exceeding
500 °C, the microstructure started to overage with an accompanying coarsening of the
αS-needles, and therefore, the hardness decreases.
In a next step, different holding times at 500 °C were investigated. Here, holdings
times between 15 min and 2 h with 15-min steps, between 2 and 8 h with 2-h steps and
finally 24 h have been applied. The starting point was, again, the recrystallized micro-
structure after 700 °C for 10 min WQ. Table 3 summarizes the hardness values after
ageing.
Fig. 3. SEM images of the recrystallized microstructures: (a) 10 min holding time; (b) 30 min holding
time.
Fig. 4. Hardness values and microstructures after ageing at different temperatures.
Table 3. Hardness values in HV10 after ageing at 500 °C with different holding times.
2 4 6 8
15 min 30 min 45 min 60 min 75 min 90 min 105 min h h h h 24 h
272 280 283 291 286 281 286 280 286 292 290 266
It can be seen that the hardness values are quite similar. From 15 to 60 min, there is
a slight increase visible due to the ongoing α″-martensite decomposition. Between
30 min and 8 h, the values are almost constant taking the 3% measurement uncertainty
and standard deviation into account. This indicates that the α″-martensite was fully
decomposed. After 24 h, the hardness has strongly decreased because of overageing.
Due to the results of the hardness measurements, tensile test specimens have been
prepared. Holding times of 15–60 min have been chosen as the presence of small
amounts of α″-martensite after these heat treatments should lower the Young’s modu-
lus. The results of the tensile tests are summarized in Table 4; the corresponding SEM
images of the microstructures are shown in Fig. 5. During the tensile tests, a hysteresis
loop has been implemented for improved determination of the Young’s modulus. It
becomes evident that the Young’s modulus increases with increasing holding time as
expected, due to the ongoing α″-martensite decomposition as shown in Fig. 5. At the
same time, the ultimate tensile strength increases due to the nano-structuring because
of the newly formed α- and β-grains. However, the elongation at fracture decreases with
increasing tensile strength.
All four specimens fulfil the requirements of ASTM F1713, but as the goal was to
reach the highest possible strength with a Young’s modulus between 70 and 80 GPa and
an elongation at fracture of minimum 8%, the specimen after 60 min holding time is
most suitable for dental implant applications.
Table 4. Tensile test results after ageing at 500 °C with different holding times.
E/GPa EHysteresis/GPa YTS/MPa UTS/MPa A/%
15 min 76 68 873 938 10,0
30 min 84 71 898 963 9,3
45 min 78 71 902 964 9,5
60 min 85 76 894 970 9,3
Fig. 5. SEM images of the aged microstructures at 500 °C for (a) 15 min; (b) 30 min; (c) 45 min; and
(d) 60 min with subsequent air cooling.
4 Summary and Outlook
In this study, heat treatments at ECAP-deformed material was performed to evaluate

temperatures and holding times to establish a nanostructured microstructure with the
highest possible strength, a low Young’s modulus, and a sufficient elongation at fracture
to fulfil the requirements of ASTM F1713. The heat treatment strategy combines
recrystallization with ageing. For recrystallization, a temperature of 700 °C was used,
which is slightly below β-transus to obtain a high amount of α″-martensite after water
quenching to reduce the Young’s modulus. In addition, a small amount of αP keeps the
β-grain size small at annealing temperature and improves the fatigue properties.
Nevertheless, the holding time has to be limited to 10 min to prevent severe grain
growth. Optimal ageing was possible at 500 °C for 1 h with subsequent air cooling to
obtain a high strength due to α″-martensite decomposition and nano-structuring of the
alloy. The resulting mechanical properties are a Young’s modulus of 76 GPa, and an
ultimate tensile strength of 970 MPa at an elongation at fracture of 9.3%, fulfilling the
requirements of the ASTM F1713. The results presented here open an opportunity to
produce fully nanostructured dental implants made of TNZ in the future.
Future work will include a further microstructure and properties optimization, e.g.
by additional parameter studies regarding the ageing treatments. Besides tensile tests,
fatigue tests will be carried out at standard samples and real dental implants. Furthermore,
for medical certification, biological testing must be implemented to evaluate, if the
nanostructured microstructures are suitable for promoting bone cell adhesion and grow-
ing as well as to prevent bacterial colonization.
In large modern aircraft engines, β-rich (α + β) alloys like Ti 6Al 2Sn 4Zr 6Mo are
applied [12]. To increase their strength levels locally, a partial martensitic transforma-
tion at the surface of related parts followed by a martensite decomposition leading to
severe grain refinement might be feasible. The results of our study presented here
should in general be transferable to other β-rich (α + β) titanium alloys like Ti 6Al 2Sn
4Zr 6Mo so that short-time heat treatments might be applicable in aerospace compo-
nents as well.
Acknowledgements The research leading to the results presented here has been funded by the BMBF
(Bundesministerium fuer Bildung und Forschung, Germany), project number 13XP5093C. Financial
support is, therefore, gratefully acknowledged. Also, the authors would like to thank COMTES FHT
(Czech Republic) for producing the Conform-ECAP rods as well as our project partner ADVANTIQX
Dr. Johannes Scherer.
References
1. Lütjering G, Williams JC. Titanium. 2nd ed. Berlin Heidelberg: Springer; 2007.
2. Blakey-Milner B, Gradl P, Snedden G, Brooks M, Pitot J, Lopez E, Leary M, Berto F, du Plessis
A. Metal additive manufacturing in aerospace: a review. Mater Des. 2021;209:110008. https://doi.
org/10.1016/j.matdes.2021.110008.
3. Geetha M, Singh AK, Asokamani R, Gogia A. Ti based biomaterials, the ultimate choice for
orthopaedic implants – a review. Prog Mater Sci. 2009;54:397–425. https://doi.org/10.1016/j.
pmatsci.2008.06.004.
4. Davidson JA, Mishra AK, Kovacs P, Poggie RA. New surface-hardened, low-modulus, corrosion-
resistant Ti-13Nb-13Zr alloy for total hip arthroplasty. Biomed Mater Eng. 1994;4(3):231–43.
5. Brunke F. Ti 15Mo und Ti 13Nb 13Zr: Qualifizierung von Titanlegierungen der zweiten Generation
für der Einsatz in der Medizintechnik (Published dissertation) TU Braunschweig, Germany; 2019.
6. Gongadze E, Kabaso D, Bauer S, Slivnik T, Schmuki P, van Rienen U, Iglič A. Adhesion of osteo-
blasts to a nanorough titanium implant surface. Int J Nanomedicine. 2011;6:1801–16. https://doi.
org/10.2147/IJN.S21755.
7. Peters M, Leyens C. Titanium and titanium alloys. Weinheim, Germany: Wiley-VCH; 2002.
8. Mouhyi J, Dehan Ehrenfest DM, Albrektsson T. The peri-implantitis: implant surfaces, mirocstruc-
ture, and physicochemical aspects. Wiley Periodicals, Inc; 2009. https://doi.org/10.1111/j.1708-82
08.2009.00244.x.
9. Venkatsurya PKC, Girase B, Misra RDK, Pesacreta TC, Somani MC, Karjalainen LP. The interplay
between osteoblast functions and the degree of nanoscale roughness induced by grain boundary
grooving of nanograined materials. Mater Sci Eng C. 2012;32:330–40. https://doi.org/10.1016/j.
msec.2011.10.036.
10. Görtan M. Severe plastic deformation of metallic materials by equal channel angular swaging: the-
ory, experiment and numerical simulation (Published Dissertation). TU Darmstadt, Germany; 2014.
11. Klinge L, Siemers C, Sobotta B, Krempin H, Kluy L, Groche P. Nanocrystalline Ti 13Nb 13Zr
for dental implant applications. Proceedings of the 60th Conference of Metallurgists, COM
2021 Paper No. 0033 The Canadian Institute of Mining, Metallurgy and Petroleum, ISBN:
978-1-926872-53-7; 2021.
12. Williams JB, Boyer RR. Review – opportunities and issues in the application of titanium alloys for
aerospace components. Metals. 2020;10(6):705. https://doi.org/10.3390/met10060705.
A Novel Experimental Set-up for Generating
Microbubbles for Removal of Inclusions
A. Panicker1, D. Lu1, R. Tiwari1, L. Calzado1, M. M. Isac1 (*), and R. I. L. Guthrie1

1
McGill Metals Processing Center, McGill University, Montreal, QC, Canada
Mihaiela.Isac@mcgill.ca
Abstract. The empirical relations that govern multiphase interactions between micro-
bubbles and inclusions indicate a strong dependence between bubble size and overall
inclusion capture efficacy. Established probability models highlight the need for bubbles
less than 1 mm in diameter for optimal inclusion capture. To achieve bubbles in this size
regime, tangential shearing flows and turbulence can be used to facilitate bubble forma-
tion and fragmentation. A novel set-up consisting of nozzles submerged within a melt or
an aqueous bath analog has been designed, manufactured, and operated, to test the bubble
sizes produced for different speeds of rotation and effluent gas flow rates. Increasing gas
flow rates lead to larger bubbles, whereas larger RPMs lead to much smaller bubble sizes.
Bubbles are captured through high-speed digital photography and then analyzed using
ImageJ postprocessing software. All experiments were conducted in an aqueous medium.
Keywords: Microbubble generator · Bubble fragmentation · Turbulence shear ·

ImageJ · Rotation · Flow rate
1 Introduction
The concept of metal cleanliness was first invented in the mid-twentieth century to refer
the number of inclusions per volume of metals [1, 2]. The presence of inclusions in
metals can cause detrimental effects on mechanical properties, including tensile
strength, formability, toughness, weldability, cracking resistance, corrosion resistance,
and fatigue resistance [3]. Thus, the control of inclusion level is critical in manufactur-
ing premium-quality metals with predictable properties. Especially for structural com-
ponents of which mechanical properties are essential in their applications, inclusions
must be minimized to suppress their harmful effects. Along with inclusion levels,
another important factor is the inclusion size distribution. In fact, all metals include
some level of inclusions, but not all of them are equally harmful. Generally, small inclu-
sions (<4 μm) do not constitute a concern [4]. However, inclusions that aggregate to
form larger macro-inclusions are detrimental to the metal quality, since a catastrophic
defect is usually caused by just a single large inclusion. Therefore, clean steel entails
not only minimizing inclusion levels but also avoiding inclusions larger than the critical
size that are harmful to the given product [5].
https://doi.org/10.1007/978-3-031-17425-4_49
370 A. Panicker et al.
Inclusion removal by bubble floatation/gas dispersion has been widely accepted as

an effective approach for improving metal cleanliness while the metal is molten. The
process injects non-reactive gas into the liquid metal, bringing non-wetted inclusions to
the top surface by their attachment to rising gas bubbles. The upward flow caused by the
floating bubbles can also promote the homogenization of the liquid metal’s composition
and temperature [6]. The efficiency of inclusion removal using this method differs
according to bubble size. In continuous casting processes, large bubbles float upward
(because of their higher buoyancy forces and shorter residence time compared to the
smaller ones) and entrain mold powder/slag, whereas smaller bubbles tend to move
towards sidewalls and may be captured by the solidifying shell, again resulting in
defects in steel products [7]. On the other hand, the equivalent volume of smaller bub-
bles offers substantially larger net interfacial areas, which can then lead to higher
removal rates. Studies by Chang et al., and by Wang et al. [6, 7] concluded that the
bubbles with a diameter of 0.5 mm to 1 mm should have the optimal inclusion removal
rate. These bubbles are classified as microbubbles as their diameter is less than 1 mm.
One approach to create such tiny bubbles is to use liquid shears. The process splits
the generated bubbles into smaller bubbles by application of shear stresses. When bub-
bles are subjected to flow stresses, they lose their ability to maintain a steady shape as
soon as flow-induced stresses exceed the interfacial tension stress. Rather, they undergo
a transient, continuous stretching, which eventually results in bubble breakage, and the
formation of smaller bubbles. In addition, research by Marshall et al. [8] has revealed
the dependence of bubble radius on gas velocity, crossflow velocity, and orifice radius
in shearing flows. Surface tension is also important in determining the final bubble
sizes. Bubbles forming in water are expected to be much smaller compared to the bub-
bles forming in equivalent liquid metals due to metal’s non-wetting nature and higher
surface tensions.
Here, we introduce our approach to development of a microbubble generator based
on a rotational shearing mechanism, which could be used to constantly produce micro-
bubbles with sizes within the pre-mentioned range, to enhance metal cleanliness.
Testing of the microbubble generator was carried out in an aqueous medium to study its
performance under different conditions. An empirical model was developed for predict-
ing the sizes of bubbles generated. Factors such as gas flow rate and rotational speed
were investigated as a primary contribution to the development of the model. Similarly,
a numerical CFD simulation was applied to validate the experimental results.
2 Novel Microbubble Generator
The computer-aided design (CAD) software, SolidWorks, was used for design visual-
ization for the fabrication of the microbubble generator. The design of the microbubble
generator consists of three main components. The “sandwich” component at the top
incorporates a sealed gas chamber enclosed by two stainless-steel disks with a gas inlet
on the side to allow gas entrance. The stainless-steel disks are sandwiched between two
smaller brass disks on each side, providing improved stability and better seals at the
connection to the rotating shaft. The shaft connects the upper and lower components
A Novel Experimental Set-up for Generating Microbubbles for Removal of Inclusions 371
Fig. 1. (a) CAD of microbubble generator (b) Water modeling experimental setup.
together and provides a path for gas transfer in between. It also functions as a rotational
axle when the top end is connected to a drill. Since all parts of the microbubble genera-
tor are joined together, they spin at the same speed during the operation. The bottom
end of the shaft is threaded for the connection to the bottom component. The hollow
disk at the bottom is the core of the microbubble generator, which is composed of 2
separate disks of 30 mm diameter, fastened together by screws, and sealed by an O-ring.
The lower disk has four screw nozzles that are equally distributed on the side of the
component for gas injection and bubble generation. The orifices were mechanically
drilled, with 0.2 or 0.6 mm diameter holes. All parts of the microbubble generator, apart
from the brass disks, were made of stainless steel to ensure a tough, rigid, and corrosion-
resistant structure. In addition, a stand was designed to support the microbubble genera-
tor and to fix it to the container (See Fig. 1a). The center of the stand is a large disk
where the upper component of the microbubble generator can be inserted. It is con-
nected to 4 bridges and vertical supports which fasten onto the container’s edges by
screws. Figure 1b shows the fabricated microbubble generator prototype.
2.1 Water Modeling
The water modeling experiments were performed in a transparent container using the
set-up illustrated in Fig. 1b. The container, purchased from Cambro, boasted a 4.5 L
capacity, and was filled with 3 L of water. The upper end of the shaft is connected to a
drill (from MAKITA) with an rpm range of 550–1100, to achieve the range of shearing
velocities we wished to study. A tachometer was aimed at the reflective tape on the bot-
tom disk while the disk is in rotation to accurately measure the speed. Alternatively, for
the liquid metal experiments, the shaft of the pin chuck was used. A flowmeter is con-
nected before the gas inlet, allowing adjustments on gas flow rate. Target flow rates
were in the range of 20–80 mL min−1. A high-speed camera with a frame rate set at
1/3200 s, captured images of the bubbles generated. A LED light was placed on the
back side and covered with a piece of white paper to provide more brightness and better
contrast. ImageJ was used as the analytical tool to define the bubble size and shape. By
manually defining the acceptable range of circularity and particle size using the thresh-
old function, background noise was largely eliminated, leaving only the well-focused
bubbles to be analyzed. The positions of the set-up remained unchanged during the test,
to ensure measurement accuracy.
3 Computational Fluid Modeling
Computational Fluid Dynamics (CFD) provides valuable insight into the properties of
the fluid domain for simulated experimental conditions. In this investigation, a 3D
model was constructed and solved using the ANSYS–FLUENT 19.0.0 commercial
package. Its function was to predict the turbulence defining parameters within the fluid
domain.
3.1 Governing Equations
Continuity and Momentum Equations The continuity and momentum conservation

equations as shown in Eqs. 1 and 2 will be used to solve for the velocity of the continu-

ous phase [9]. ρ and v will represent density and velocity fields, respectively, whereas
p and μ are the pressure fields and dynamic viscosity pertinent to the corresponding
fluid.

v 0, (1)
t

∂v
ρ + ρ ( v • ∇ ) v = −∇p + µ • ∇ 2 v + ρ g . (2)
∂t
Volume fraction transport In the equation below, f will represent the volume fraction
of a species in a finite volume cell [9–11].
∂f
+ ∇ • ( vf ) = 0. (3)
∂t
k-ε transport equations Turbulence in a fluid domain causes arbitrary velocity fluc-
tuations. Thus, the continuity and momentum conservation equations are treated to
form the Reynolds Averaged Navier Stokes (RANS) equations [9].
∂ ( ρk )  µ  
+ ∇ • ( ρ vk ) = ∇  µ + t  ∇k  + Gk − ρε , (4)
∂t  σ k  
∂ ( ρε )  µ   ε ε2
+ ∇ • ( ρ v ε ) = ∇  µ + t  ∇ε  + ρ C1 Gk − C2 ρ. (5)
∂t  σ ε   k k
3.2 Geometry
The Eddy viscosity term can be computed using the following equation [12]:
k2
t C . (6)

Finally, the generation term, Gk, can be expressed using the following equation [9]:
v j
Gk vi v j . (7)
xi
Finally, the standard model comes with a series of constants that are set to the values
recommended by Launder and Spalding [13] where σk, σε, C1, C2, and Cμ are 1.0, 1.3,
1.44, 1.92, and 0.09, respectively.
Using the SpaceClaim application, a 3D design representing 1/4th of the microbub-
ble generator and surrounding fluid domain was made. The adoption of 1/4th geometry
due to rotational symmetry allows for a significant reduction in computational time and
effort. Specifically, as seen in Fig. 2, the submerged section of the rotating shaft and
bottom chamber along with the nozzle features were designed there.
3.3 Meshing
Meshing is the process by which the continuous fluid domain is discretized into finite
volumes in which the governing equations are solved. In this investigation, the geom-
etry presented above is meshed to form 305,753 cells or 67,008 nodes with an average
skewness of 0.238.
3.4 Boundary Conditions and Solver Set-up
To simulate the rotational motion of the microbubble generator, the interior and exterior
walls of the submerged components were selected and then set to a no-slip moving
wall. Rotational motion was specified by inputting the wall velocity in RPM, as 550,
650, 750, 850, and 1100 RPM. Other boundary conditions include setting the inlet as a
wall and the outlet as a pressure outlet at 0-gauge pressure.
The general set-up consists of a pressure-based transient solver, since our flow cor-
responds to the incompressible regime, with gravity enabled on the y axis at −9.81 m s−1.
An explicit VOF model was adopted, as it gives access to the most advanced interface
tracking method offered by FLUENT; this is the Geo-Reconstruct scheme [9]. Air and
Fig. 2. (a) 3D geometry of microbubble generator; (b) Meshed fluid domain.
water were the primary and secondary phases, with surface tension modeling enabled,
using a value of 0.072 N m−1. Finally, the standard k-ε model was also selected. To solve
the selected models, the Semi-Implicit Method for Pressure-Linked Equations
(SIMPLE) scheme was enabled for the pressure velocity coupling. A variable time-
stepping method was also selected with an initial time step set at 10−7 seconds, and the
simulations were allowed to run until the volume average of key parameters such as ε,
stabilized with time.
The aim of the present investigation was to quantitatively study the effect of gas inlet
flow rate and rotational speed on the mean equivalent bubble diameter. Prior to present-
ing the data, it is important to define exactly how the mean diameters were computed.
The image detection plug-in, in ImageJ, detects bubble edges from the threshold images
and gives the user-defined metrics describing those edges. A critical parameter is the
area that is enclosed by the edge. A standardized diameter for each edge can be derived
if the measured area is assumed to represent a circle. Thus, using the area value, an
equivalent diameter sphere is measured using the following equations, where Ai is the
area enclosed by edge ‘i’ and deq, i is the equivalent diameter sphere. Using the nominal
values of equivalent diameters, a linear mean, or D10, mean diameter, can be computed
by simply summing the equivalent diameters and dividing it by the total number of
edges in a sample set. Using this as a measure for mean bubble diameter, a 3D scatter
plot with a superimposed surface captures the mean diameters for all the 42 experi-
ments performed.
4
d eq,i Ai . (8)

It is seen from Fig. 3 that the equivalent mean diameters reduce with an increase in
rotational speed. An increase in bubble diameters can also be observed with increased
gas inlet flow rate within the tested domain. To be specific, the smallest mean diameter,
Fig. 3. 3D scatter plot for rotational speed vs inlet flow rate vs equivalent diameter.
0.490 mm or 490 μm, is observed for a combination of 1110 RPM and 23 mL min−1 air
flow rate. On the other hand, the largest diameter was observed for 550 RPM and
82 mL min−1 gas flow rate, at ~1.69 mm. This agrees with the general conclusion that
microbubble generation, for the target range of ~500 μm, is achieved for experimental
condition characterized by high rotational speeds and low air injection flow rates.
Equally important is to consider the individual effects of rotational speeds and gas inlet
flow rate on the mean bubble diameter. Furthermore, their contributions can be quanti-
fied by performing a regression analysis. A nonlinear regression will also be applied to
the overall scatter data, thereby leading to a predictive model that gives mean bubble
diameter as a function of rotational speed and gas flow rate.
4.1 Rotational Speed
This concept is clarified in the model presented by Marshall et al. [9] where the radius
of the bubble reduces with increasing crossflow velocity. A similar phenomenon is seen
when plotting the rotational speed vs mean bubble diameters at fixed flow rates. In the
figures below, mean bubble diameters are plotted against rotational speeds, for an air-
flow rate of 20 mL min−1.
A general decrease in mean bubble diameter is observed, as expected, with increas-
ing rotational speeds, for all fixed flow rates. As mentioned earlier, the relative motion
between the gas flow and the continuous phase fluid will cause instabilities on the sur-
face of the gas phase. These instabilities are a result of dynamic pressure changes
caused by turbulence-induced velocity fluctuations in the fluid domain. When the
dynamic pressures exert a deformation force on the bubble surface, then, under favor-
able conditions, these surface instabilities will lead to bubble fragmentation. w concept
will be discussed in further detail in the CFD results section. In general, one can expect
increased deformation pressures at larger rotational speeds, owing to the higher relative
velocities between the gas and liquid phases. This will ultimately lead to greater degrees
Fig. 4. Linear and power model curve fitting for (a) RPM vs equivalent diameter at 20 mL min−1. (b)
Air inflow rate vs equivalent diameter at 1100 RPM.
of bubble breakage and fragmentation, thereby explaining the reduction in bubble

diameters with increased rotational speeds.
The effect of rotational speed on the bubble diameter can now be quantified by per-
forming a regression analysis. Both power law and linear models were tested for their
respective quality of fit against the experimental data. While both models present good
data fitting with respect to their r2 values, the power law model provides comparatively
better predictability. This is because it better captures the mean diameter’s asymptotic
approach to a stable value, at higher rotational speed values. This ultimately results in
higher r2 square values over all fixed flow rates, as compared to the linear model fit.
Figure 4 shows the scatter data of rotational speeds vs mean bubble diameter superim-
posed with the fitted curves. R2 values of 0.922 and 0.954 were achieved for the power
law fit at 20 mL min−1 and 80 mL min−1, respectively. The power law model is an equa-
tion of the form, y = axb. Given that x represents the rotational speed, then the mean
value of b averaged over all fixed flow rates is −1.491 ± 0.133.
4.2 Gas Inlet Flow Rate
A similar independent analysis was performed on gas inlet flow rate to quantify its
effect on the mean equivalent diameter, while holding the rotational speed constant.
Generally, an increase in mean equivalent diameter is observed with the gas inlet flow
rate. This can be seen both in the 3D scatter surface flow, as well as in the individual air
flow rate mean diameter plots for a fixed rotational speed of 1100 RPM (see Fig. 4).
The curve-fitting operations performed here with power law and linear fits to deter-
mine a predictive model that will help quantify the independent effect of air flow rate
on mean diameter. Analyzing the two regression statistics, it is concluded that poor fits
are achieved for the gas inlet flow rate vs mean diameter at low rotational speed values.
For instance, both linear and power law models give r2 values of 0.676 and 0.635,
respectively, at 550 RPM. On the other hand, r2 values of 0.871 and 0.901 are achieved
for the linear and power law models, respectively, at 1100 RPM. The increase in model
predictability at higher rotational speeds can be attributed to the difference in the num-
ber of bubbles sampled at different rotational speeds. Typically, the degree of bubble
fragmentation at high rotational speeds significantly exceeds that at lower speeds. That,
in turn, causes finer gas dispersion with a larger number of bubbles. A more rigorous
data collection protocol must be applied to the lower rotational speed testing so that the
significant reduction in bubble number density is compensated for. Using the power law
form of equation described earlier, a mean b value of 0.210 ± 0.051 is reported for air
flow rate vs equivalent diameter. Thus, rotational speed is found to have a greater influ-
ence on the equivalent diameter vs the gas flow rate.
4.3 Overall Predictive Model
Having studied the independent variables individually and having assessed their impacts
on the equivalent mean diameter, it is now possible to apply non-linear regression on
the complete experimental dataset, to determine a predictive model. A model equation
is first proposed, and non-linear regression is then performed to find the coefficients
that will maximize the value of r2.
Figure 3 shows the surface that was used to capture the experimental data. The equa-
tion defining the surface is also provided. The regression analysis helped determine the
overall model with an r2 value of 0.888. This indicates a good predictability with exper-
imental results.
Q 0.171
d eq vrotation ,Qair 9364 air 1.488 ; r 2 0.888 . (9)
vrotation
4.4 Crossflow Model Comparison
Finally, the experimental results can also be compared with the crossflow model pro-
posed by Marshall et al. [8]. This can be done by plotting the experimental data against
the model predicted values for each given set of experimental conditions. In addition, a
1:1 similarity line can also be plotted to assess the degree of similarity. As seen in
Fig. 5, there is good similarity with the model predicted data for the larger bubble diam-
eter region. Crucially, however, there is significant discrepancy in the lower diameter
region. The experimental mean diameters are consistently smaller than the model pre-
dicted values in this regime. A possible reason for this is the occurrence of bubble
break-up. Once a bubble has formed and been released into the liquid, it enters a region
of high turbulence. Surface perturbations caused by dynamic pressure changes will,
under favorable conditions, eventually lead bubble fragmentation [14, 15]. To confirm
the proposed possibility causing deviation from the model predicted diameters, CFD
was implemented to study the possibility of bubble fragmentation, by first quantifying
the dissipation rate of the kinetic energy of turbulence, ϵ, for different rotational speeds.
5 CFD Results and Discussion
Using the ε values and the model presented by Martinez-Bazan et al. and Sevik and
Park [15, 16], critical bubble diameters can be predicted, and this can be used to con-
firm the presence of bubble break-up in the physical experiments performed.
v 2 rb
We . (10)

A critical We number can be defined as the maximum allowable deformation pres-

sure that a bubble in a turbulent field can withstand without undergoing fragmentation.
Under this condition, Martinez-Bazan et al. performed a force balance equation, bal-
ancing the deformation pressure against the restoration pressure, enabling them to
determine thereby, a critical bubble diameter [16]. Sevik and Park state that the critical
We number, for highly turbulent fluid fields characterized by the domination of inertial
forces as opposed to viscous forces, is approximately 1.26 [15]. Thus, a critical bubble
diameter equation is derived as the following equation:
3 2
5 1 5
d critical 1.26 . (11)

Fig. 5. Experimental results comparison with crossflow model.

5.1 1100 RPM Case
For each rotational speed simulated, values of ε in the fluid domain close to the rotating
nozzles were studied, together with the progressive reduction change away from the
nozzle. At 1100 RPM, Fig. 7a displays the contours of dissipation rate with special
focus given to the region adjacent to the nozzle.
Points labeled ‘1,’ ‘2,’ and ‘3’ are arbitrary points that were probed for their respec-
tive ε value so that the critical diameters at those location could be calculated. The CFD
generated results gave ε values of 354 m2 s−3, 5059 m2 s−3, and 276 m2 s3 at points 1, 2,
and 3 in Fig. 6.
These correspond to critical bubble diameters of 0.359 mm, 0.124 mm, and
0.397 mm, respectively. The mean bubble diameter at 1110 RPM and 23 mL min−1 of
air flow was approximately 0.490 mm which is larger than the prescribed critical diam-
eters. Bubble fragmentation can, thus, be expected at this rotational speed. The confir-
mation of bubble break-up once again helps us to be able to account for the deviation in
the current data, from the crossflow model in Fig. 5.
However, an explanation needs to be provided as to why the mean diameters are
significantly larger than the sample calculated critical diameters. To do so, the velocity
vectors of the fluid field must be studied (Fig. 7). It is seen that the gas exits the nozzle
at high speed. Once the bubble has been detached from the nozzle, it is carried rapidly
away from the nozzle wall to a region of significantly lower dissipation rate due to the
large centrifugal force. Therefore, the bubble quickly escapes the region of high turbu-
lence which allows only a partial degree of fragmentation. This can also be further
affirmed by observing the rapid decrease of the ε values away from the nozzle walls in
Fig. 6. Finally, a volume average of the dissipation rates is taken for each rotational
speed simulated. The average value, over this selected volume, increases with the rota-
tional speed which points to a reduction in the critical diameter (see Fig. 7). The
decrease in critical diameter proves that the extent of fragmentation will be lower at
lower rotational speeds.
Fig. 6. Turbulent dissipation rate contours enhanced at the bottom chamber nozzles.
Fig. 7. (a) Velocity vector near the nozzle (1100 RPM). (b) Turbulent dissipation rate vs simulated
RPM.
6 Conclusions
The following conclusions can be drawn from this investigation:

1. A novel microbubble generator design has been proposed, fabricated, and operated,
to investigate the production of microbubbles in rotation-induced turbulent shear.
2. The smallest mean bubble diameter, 0.490 mm, was achieved for 1110 RPM and
23 mL min−1, indicating that higher rotational speeds and low gas inlet flow rates are
necessary to produce microbubbles.
3. Experimental results fall below the crossflow model predicted diameters at lower
diameters due to turbulence-induced bubble breakage after their formation.
4. The average ε value increases with increasing rotational speed, thus, reducing criti-
cal bubble sizes. This is in keeping with a greater degree in bubble fragmentation, at
higher RPM.
Acknowledgments The authors acknowledge the financial support of NSERC, as well member com-
panies of the McGill Metals Processing Centre, Rio Tinto Iron and Titanium, Nippon Steel Corporation,
Japan, as well as Quebec Centre of Aluminum Research and Development.
References
1. Lopez-Urunuela FJ, Fernandez-Diaz B, Pagano F, Lopez-Ortega A, Pinedo B, Bayon R,

Aguirrebeitia J. Broad review of “White Etching Crack” failure in wind turbine gearbox bearings:
Main factors and experimental investigations. Int J Fatigue. 2021;145(106091):106091. https://doi.
org/10.1016/j.ijfatigue.2020.106091.
2. Jean-Pierre B. Steel cleanliness and environmental metallurgy. Metal Res Technol. 2016;113(2):201.
https://doi.org/10.1051/metal/2015050.
3. Satyendra. Steel cleanliness and clean steel technologies; 2020. https://www.ispatguru.com/steel-
cleanliness-and-clean-steel-technologies/. Last accessed 07 Feb 2022.
4. Pretorius EB, Oltmann HG, Schart BT. An overview of steel cleanliness from an industry perspec-
tive. AIST Trans. 2015;12(3)
5. Zhang L, Thomas BG, Wang X, Cai K. Evaluation and control of steel cleanliness review. In: 85th
steelmaking conference. Warrendale: ISS-AIME; 2002. p. 431–52.
6. Chang S, Cao X, Hsin C-H, Zou Z, Isac MM, RlL G. Removal of inclusions using micro-bubble
swarms in a four-strand, full-scale, water model tundish. ISIJ Int. 2016;56(7):1188–97. https://doi.
org/10.2355/isijinternational.ISIJINT-2016-077.
7. Lai Q, Luo Z, Hou Q, Zhang T, Wang X, Zou Z. Numerical study of inclusion removal in steel
continuous casting mold considering interactions between bubbles and inclusions. ISIJ Int.
2018;58(11):2062–70. https://doi.org/10.2355/isijinternational.ISIJINT-2018-319.
8. Marshall S, Chudacek MW, Bagster DF. A model for bubble formation from an orifice with liquid
cross-flow. Chem Eng Sci. 1993;48(11):2049–59.
9. ANSYS Fluent User's Guide. https://www.scribd.com/document/357077969/ANSYS-Fluent-
Users-Guide-pdf. Last accessed 18 Feb 2022.
10. Hirt C, Nichols B. Volume of fluid method for the dynamics of free boundaries. J Comput Phys.
1981;39:201–25.
11. Katopodes ND. Chapter 12 – Volume of fluid method. In: Free-surface flow. Oxford: Butterworth-
Heinemann; 2019. p. 766–802.
12. Chang S, Huang W, Zou Z, Li B, RlL G. Motion behavior of micro-bubbles in a delta shape tundish
using impact pad. Powder Technol. 2020;367:296–304.
13. Launder BE, Spalding DB. The numerical computation of turbulent flows. Comput Methods Appl
Mech Eng. 1974;3:269–89.
14. Lalanne B, Masbernat O, Risso F. A model for drop and bubble breakup frequency based on turbu-
lence spectra. Am Inst Chem Eng. 2019;65(1):347–59.
15. Sevik M, Park SH. The splitting of drops and bubbles by turbulent fluid flow. J Fluids Eng.
1973;95(1):53–60.
16. Martinez-Bazan C, Montanes JL, Lasheras J. On the breakup of an air bubble injected into a fully
developed turbulent flow. I – breakup frequency. J Fluid Mech. 1999;401:157–82.
Asymmetrical Rolling of Aluminum Alloy 6061
Julie Lévesque1 (*) and Alexandra Béland1

1
Québec Metallurgy Center, Trois-Rivières, QC, Canada
Julie.Levesque@cegeptr.qc.ca
Abstract. Aluminum alloy 6061 is a good candidate for automotive sheet components
due to its good mechanical properties. However, its formability is low when compared to
competing steels. In this work, different processing routes, including cross-rolling (CR)
and asymmetrical rolling (ASR), have been used to increase the formability of 6061
sheets. Results show that CR and ASR lead to an increase in the average r-value when
compared to conventional rolling but also increase the anisotropy coefficient.
Keywords: Aluminum alloys · Asymmetrical rolling · Cross-rolling · Texture ·

Formability · Anisotropy
1 Introduction
Aluminum alloy 6061 in sheet form is interesting for automotive applications because
of its good mechanical properties. However, its use is limited by low formability com-
pared to steel sheets developed for the same applications.
In this work, sheets have been produced by different routes, including cross-rolling
(CR) and asymmetrical rolling (AS). These types of rolling can improve the micro-
structure and texture of sheets, leading to a better formability [1, 2]. Cross rolling con-
sists of turning the sheets between rolling passes, while asymmetrical rolling implies
that the speed at the surface of the working rolls is different from one another. This
applies a shear strain through the thickness of the sheet and promotes recrystallization
and texture randomization. Asymmetrical rolling is of particular interest because it is
easier to implement into an industrial manufacturing process.
2 Methodology
A 12.5-mm-thick hot-rolled aluminum alloy 6061 plate has been cut to produce the
samples used in this work. The plate was received in the T6 condition, but a complete
annealing treatment (415 °C for 2 h30 followed by slow cooling at a rate of 30 °C/h to
260 °C) was performed on samples before rolling. Four different processing routes have
then been followed.
https://doi.org/10.1007/978-3-031-17425-4_50
384 J. Lévesque and A. Béland
Fig. 1. Cross-rolling procedure. Samples have been rotated 90° between each rolling pass.
• UD: Unidrectionnal symmetrical rolling

• CR: Cross-rolling (see Fig. 1)
• AS10: Asymmetrical rolling, average thickness reduction per pass = 10%
• AS16: Asymmetrical rolling, average thickness reduction per pass = 16%
The upper roll was 5% smaller in diameter than the lower roll for asymmetrical roll-
ing. The average thickness reduction per pass was 10% for AS10 and 16% for AS16. All
samples have been dry rolled, e.g., without any lubricant, because asymmetrical rolling
requires a high friction coefficient to achieve good results [3]. After rolling, samples
were machined from the sheets, and then solution was heat treated at 530 °C and water
quenched. They have been naturally aged for at least 3 days to stabilize their properties
and obtain the T4 temper. Sheets have been characterized in the T4 condition since, for
automotive components, they are often formed in that temper before being artificially
aged.
The sample’s microstructure and texture have been characterized by electron back-
scattered diffraction (EBSD). Tensile test specimens have been cut at 0°, 45°, and 90°
from the rolling direction. The tensile tests have been performed (two good repetitions
per test condition) on a mechanical testing system using a digital image correlation
(DIC) system (GOM Aramis). The DIC system allows measuring the strains in major
and minor directions during the test and calculating the r-value.
width
r , (1)
( width axial )
where εwidth is the true strain in the specimen width and εaxial is the strain in the axial
direction.
After solution heat treatment and quench, the samples are all well recrystallized, as can
be seen in the EBSD maps (Fig. 2). Table 1 presents the grain sizes obtained for the
different samples in the T4 condition. The average grain size varies from 14.6 to 12.5
microns. The variation is small, showing that the difference in mechanical properties
can mostly be attributed to texture.
Stress–strain curves for all samples are presented in Fig. 3. The yield strength is
approximately the same for all samples, whereas the strain hardening seems to show
Asymmetrical Rolling of Aluminum Alloy 6061 385
Fig. 2. EBSD maps of unidirectionally rolled (UD), cross-rolled (CR) and asymmetrically rolled
samples with 10% (AS10) and 16% (AS16) reduction per pass (horizontal = RD, vertical = ND).
Table 1. Grain size of the rolled samples.

Sample Average grain diameter (microns)
UD 14.6
CR 14.0
AS10 14.1
AS16 12.5
slight differences, but the variation in the maximum stress values is below 3% and,
therefore, not significative.
The main difference between samples is seen in the R-values, as observed in Figs. 4,
5, and 6. There is a striking difference in the R-values of samples asymmetrically rolled
at 10% reduction per pass (AS10) compared to 16% reduction per pass (AS16). AS10
Fig. 3. Stress–strain curves of all samples.
Fig. 4. r-value from tensile tests along the rolling direction (RD).
Fig. 5. r-value from tensile tests at a 45° angle from the rolling direction.
Fig. 6. r-value from tensile tests at a 90° angle from the rolling direction (TD).
samples only show a minor difference with the samples rolled symmetrical (UD).
Therefore, this reduction ratio is insufficient to provide asymmetrical rolling benefits.
The slight asymmetry ratio (1:1.05) used in this work could explain why small passes
improve the R-values. This ratio will be increased in future work to assess its effect on
grain refinement and texture modification. An advantage of asymmetrical rolling is the

formation of shear texture components. In conventional rolling, deformation texture
evolves to cube texture upon recrystallization. In the specimens rolled during this work,
the cube texture is only apparent in the UD and AS10 samples (see Fig. 7), consistent
with the lower R-values obtained at a 45° angle for these samples [3].
The average R-values and anisotropy coefficient shown in Table 2 have been calcu-
lated for all samples at a strain of 20%. The higher average R-values for CR and AS16
samples indicate better drawability [4]. However, the plastic anisotropy parameter (in
absolute value) is higher, which means the earing behavior would be more pronounced
during deep drawing for sheets produced using these rolling conditions.
Fig. 7. Pole figures of the samples (X0 = RD and Z0 = TD).

Table 2 Average r-values and anisotropy parameters of the rolled samples
Sample Average r-value Plastic anisotropy parameter

UD 0.657 0.035
CR 0.767 −0.107
AS10 0.671 −0.0313
AS16 0.740 −0.0655
4 Conclusion
In this work, sheets have been rolled using conventional symmetrical, asymmetrical,
and cross-rolling. The average grain sizes of all resulting samples are close, being
between 14.6 and 12.5 microns. Therefore, no significant improvement in the mechani-
cal properties can be observed. The R-values, however, differ significantly from one
rolling condition to another, primarily due to different texture evolution during defor-
mation and recrystallization. The average r-value is higher for the CR and AS16 sam-
ples, but their anisotropy coefficient is also higher, making it difficult to assess a
practical improvement in deep drawability. Future work will include ASR at higher
asymmetry ratios and drawing tests to evaluate if CR and ASR can effectively improve
the sheet’s drawability.
Acknowledgments The authors would like to thank the Fonds de recherche du Québec – Nature et
Technologies (FRQNT grant 2020-CO-26845) and the MITACS collegial research internship grants
program for financial support.
References
1. Ma C, Hou L, Zhang J, Zhuang L. Effect of deformation routes on the microstructures and

mechanical properties of the asymmetrical rolled 7050 aluminium alloy plates. Mater Sci Eng
A. 2018;733:307–15.
2. Sidor J, Miroux A, Petrov R, Kestens L. Microstructural and crystallographic aspects of conven-
tional ans asymmetric rolling processes. Acta Mater. 2008;56:2495–507.
3. Vincze GF, Simoes FJP, Butuc M. Asymmetrical rolling of aluminium alloys and steels: a review.
Metals. 2020;2020:1126. https://doi.org/10.3390/met10091126.
4. Inoue H, Takasugi T. Improvement in deep drawability by texture control for rolled and annealed
aluminum alloy sheets. Proceedings of the 12th international conference on aluminium alloys,
September 5–9, 2010, Yokohama, Japan, pp. 1201–1206; 2010.
Castability of Al-Fe-Ni Alloys
S. Kotiadis1 and A. Elsayed1 (*)

1
aelsay01@uoguelph.ca
Abstract. The demands of electric and hybrid vehicles have required modern aluminum
alloys that demonstrate high castability, high electrical and thermal conductivity, and
moderate strength. The uses of transition metals Fe and Ni present potential alloy addi-
tions to meet these goals due to their low solubility in solid Al (0.05 wt.%) and the poten-
tial of their intermetallic phases to be modified. One aspect of castability was assessed by
determining the die-soldering behavior of various Al-Fe-Ni alloys using pure Fe wire
submerged in a liquid melt. The potential for Al alloys that takes advantage of the low
Al-Fe eutectic (1.8 wt.%) has shown high castability through low hot tear susceptibility
and low die soldering with high conductivity and moderate strength. These properties
speak to the industrial applicability of this range of alloy compositions for use in hybrid
and electric vehicles, electronic components such as heat sinks, and battery cases and
housings.
Keywords: Aluminum · Castability · Characterization · Conductivity · Light metals

· Die soldering
1 Introduction
Most cast aluminum alloys typically contain a high Si content due to its ability to
improve castability and increase the strength of the alloy [1]. Silicon content improves
castability by increasing fluidity and reducing shrinkage while increasing strength
through solution strengthening [2]. The solid solubility limit (1.65 wt.%) is exceeded
by most traditional cast alloys which contain 5–10 wt.% Si [3]. Unfortunately, the high
solubility of Si in solid Al is also responsible for the reduced electrical conductivity in
traditional cast Al alloys.
The production of highly conductive and castable alloys requires an alternative alloy-
ing system, such as the addition of transition metals such as Fe and Ni [2]. Nickel is
typically added to Al alloys to improve high-temperature stability [4]. Iron is typically
present as an impurity due to the use of secondary material but can be added to increase
die-soldering resistance [1]. The effectiveness of Fe and Ni in maintaining high conduc-
tivity is largely due to their low solid solubility in solid Al (0.05 wt.% for both) [2].
Previous works demonstrated that Fe and Ni additions can provide moderate
strength, low hot tear susceptibility, and high conductivity [5]. Three alloys were
https://doi.org/10.1007/978-3-031-17425-4_51
392 S. Kotiadis and A. Elsayed
assessed for mass loss rate due to die soldering. These compositions were Al 1100 (min.
99 wt.% Al), Al-1.0Fe, Al-1.0Fe-1.0Ni, where units are in weight percent.
The increased cost of die repair and manufacturing, coating use, coating application,
and machining due to geometry defects provide incentive to prevent the washout of fer-
rous dies [7]. Washout occurs due to three mechanisms: soldering, corrosion, and ero-
sion [6].
Erosion is the gradual removal of mold material due to the high velocity in which the
melt is injected into the mold [6]. Die soldering is caused by the diffusion of Fe from the
steel mold to the melt [6]. This also results in gradual corrosion of the mold surface and
can lead to incorrect geometries. The influence of erosion can be mitigated by submerg-
ing the Fe wire in a static melt, and mass loss has been attributed to die soldering.
2.1 Wire Preparation
Pure iron wires (99.99 wt.% Fe) were prepared to assess the die-soldering rates of a
variety of alloy compositions. Three wires were cut to a length of 55 mm and suspended
at a distance of 15 mm between them. Approximately 15 mm of the wire length was
submerged into the melt.
3 Batch Preparation and Melting
A melt weighing a total of 200 g was melted in a graphite crucible. An RDO Easy Melt
induction melter fluctuating between 720 and 730 °C was used to fully melt the alloy.
The melt was prepared using Al 1100 wire, and Al-25Fe, Al-20Ni, and Al-50Si master
alloys. After the Al 1100 was added, fully melted and reached a temperature of 720–
730 °C, the Al-Si master alloy was added. After 20 minutes, the Al-Fe and Al-Ni master
alloys were added to the melt and stirred every 5 minutes for 30 seconds for a total of
20 minutes.
3.1 Die-Soldering Experiments
Once the melt reached a temperature of 720–730 °C, the surface was skimmed with a
boron nitride-coated dross skimmer. The Fe wires were submerged into the melt for
20 minutes before being removed.
Castability of Al-Fe-Ni Alloys 393
Al-1100 Al-1.0Fe Al-1.0Fe-1.0Ni

4.00
Mass Loss Rate 3.00
(mg/mm)
2.00
1.00
0.00
Fig. 1. The mass loss rate of each composition was averaged over two wires. The error bars represent
standard deviation of two wires.
Fig. 2. Optical microscopy shows the severity of damage caused to Fe wire due to die soldering. The
wire was submerged in an Al-1.0 wt.% Fe melt.
3.2 Mass Loss Rate Analysis
Upon removal, a layer of the melt material had adhered to the Fe wire. Two of the three
wires were submerged in a 20 wt.% NaOH solution for 24 hours to dissolve the pure Al.
The raw Fe wires were photographed using a Keyence VHX-5000 digital microscope.
The amount of mass loss was determined by normalizing the loss in mass by the length
of the wire that demonstrated dissolution due to die soldering.

4.1 Mass Loss Rate
The mass loss rates are illustrated in Fig. 1. Despite their elemental similarities, Fe and
Ni have opposite effects on die-soldering severity. While 1 wt.% Fe lowers the mass
loss rate of Al 1100 from 3.13 ± 0.01 mg/mm to 1.39 ± 0.23 mg/mm, the addition of
1 wt.% Ni increases the mass loss rate to 2.80 ± 0.02 mg/mm.
4.2 Optical Microscopy
Optical microscopy (Fig. 2) was conducted to visually assess the effect of die soldering.
Severity was assessed by analyzing the length of damaged wire which has a corroded
appearance. The damaged lengths commonly showed extreme curvature and a decrease
in diameter as the submerged depth increased.
394 S. Kotiadis and A. Elsayed
5 Conclusions
While the addition of Fe to Al reduces die soldering, the addition of Ni and Fe causes
die-soldering severity to increase. Additional microstructural investigation is required
to determine the cause of the detrimental effect of Ni.
Acknowledgments The authors would like to thank NSERC for financial support under the Discovery
program. The authors also acknowledge Nemak Canada for supplying some of the master alloys and
alloy ingots.
References
1. Rana RS, Purohit RK, Das S. Reviews on the influences of alloying elements on the microstructure
and mechanical properties of aluminum alloys and aluminum alloy composites; 2012.
2. Shin J-S, Ko S-H, Kim K-T. Development and characterization of low-silicon cast aluminum
alloys for thermal dissipation. J Alloys Compd. Sep. 2015;644:673–86. https://doi.org/10.1016/j.
jallcom.2015.04.230.
3. Mamala A, Sciężor W. Evaluation of the effect of selected alloying elements on the mechanical
and electrical aluminium properties. Arch Metal Mater. 2014;59(1):1. https://doi.org/10.2478/
amm-2014-0069.
4. Bian Z, et al. Regulating microstructures and mechanical properties of Al–Fe–Ni alloys. Prog Nat
Sci Mater Int. 2020;30(1):1. https://doi.org/10.1016/j.pnsc.2019.12.006.
5. Kotiadis S, Zimmer A, Elsayed A, Vandersluis E, Ravindran C. Effects of Si on the electrical
conductivity, microhardness, microstructure and hot tearing of Al-0.8Fe-0.5Mg-0.4Ni Alloys. In:
Tomsett A, editor. Light metals 2020. The minerals, metals & materials series; 2020. p. 210–5.
6. Fazlalipour F, Naghashnejad M, Niki Nushari M. Evaluation of adhesion and erosion/corrosion
resistance of nano-composite and nano-multilayer thin films in molten aluminum alloy. SN Appl
Sci. 2019;1(10):1308. https://doi.org/10.1007/s42452-019-1299-1.
7. Shankar S, Apelian D. Mechanism and preventive measures for die soldering during Al casting in a
ferrous mold. JOM. 2002;54(8):47–54. https://doi.org/10.1007/BF02711867.
Maintenance Strategy for Reliable Instruments
in Gas Treatment Center (GTC) of Aluminum
Smelter
Stanly Jeyaraj1 (*)

1
Maaden Aluminum Company, Jubail, Saudi Arabia
Abstract. In an aluminum smelter, the metal is separated by electrolysis process, and in

the process, perfluorocarbons gases and dusts are emitted, which are hazardous to the
environment causing corrosion and pollution. These should not escape to the workspace
as well as to the atmosphere. GTC is built with the reduction plant to provide a draft to the
pots, and to remove the chemicals and dust from the emitted gases before venting to the
atmosphere. A smelter may have multiple GTC units that are fully automated and oper-
ated continuously. There are approximately 900 equipment and 2500 instruments. For
reliable operation, economic production, and net zero emission, these instruments should
be maintained optimally. In this paper, a systematic analysis, leading to recommendation
of maintenance strategy for instruments in GTC and implementation of the strategy, is
disseminated. This method is developed for a GTC unit in a smelter that can also be
reused and scaled for similar units.
Keywords: Aluminum smelter · GTC · Preventive Maintenance (PM) · Asset

management · Criticality · Risk · Environment · Sustainable operation · Reliability
Centered Maintenance (RCM) · Intelligent instruments · Automation · Mean Time
Between Failure (MTBF)
1 Introduction
The current era is termed as “Aluminium Age” due to the exponential increase in pro-
duction and use of aluminum [1]. It has been reported about the Aluminum demand
doubling in the next 10 years, more conservatively predicted the tripling by 2050. The
general method of producing primary Aluminum using cryolite (Na3AlF6) for melting
alumina and reduction was established in 1886 by Paul Heroult and Charles Hall inde-
pendently [2]. In electrolysis process of aluminum production, anode effect causes
emission of perfluorocarbons and particles, due to the reaction of carbon anode with
cryolite. Anode effect happens when alumina levels in pot decrease because of poor
process control. This can be minimized by using automatic anode effect suppression, by
multiple points feeding of alumina and by employing advanced intelligent instrumenta-
tion to the electrolytic process. However, it is not possible to completely avoid anode
effect. The emitted gases and particles are contained by GTC, and the spent adsorption
reagent is reused in the production process.
https://doi.org/10.1007/978-3-031-17425-4_52
396 S. J. Jeyaraj
There are many subsystems in GTC that are monitored and controlled, by hundreds
of instruments. Maintenance of these instruments is critical for reliable operation of
smelter. Organizations view maintenance of all instruments in GTC is unnecessary, not
adding considerable value to the assets. But, properly developed and managed mainte-
nance strategies for instruments preserve the valuable assets for reliable operation at
optimal cost. Organizations are rarely aware that, it is possible to optimize maintenance
efforts by RCM methodology [3].
In this paper, a systematic approach to implement a reliable maintenance strategy for
instruments in GTC is presented. The types of failure, Preventive Maintenance (PM)
recommendations, procedure to implement PM activities, instruction for execution of
PM, and inventory analysis of spare instrument are explained. In Sect. 2, the details of
equipment and criticality categorization are furnished. In Sect. 3, classification of
instruments based on functionality, make, and model is furnished. In Sect. 4, various
failure modes and PM strategies are discussed with an example. Implementation of the
recommended PM strategy in ERP system is furnished in Sect. 5, followed by conclu-
sion in Sect. 6.
2 GTC Process, Equipment and Categorization
The principal objective of the GTC is to collect dust and hazardous gases emitted from
the electrolysis process, preventing health hazard and environment pollution. This is
achieved by venting each reduction pot to the scrubber modules, through a flow-
balanced duct system. The emitted gas is absorbed by the reagent and retained in the
scrubber module with emitted dust [4]. The accumulated reagent is partially recircu-
lated to the reactor in a controlled fashion, and remaining reagent is or “enriched alu-
mina” is removed and transported with air slides to storage silos for normal processing
in the electrolysis potline. The other objective of GTC is to provide forced draft through
Forced Convection Network (FCN). During the molten metal extraction and pot main-
tenance operation, air from FCN cools the pot and boosts the draft to prevent escape of
gas and particles to workspace. The function and process of GTC of various manufac-
turers are similar or have minimum variation. The GTC discussed here is part of smelter,
built by joint venture with ALCOA, and specific details could not be presented here due
to limited rights of the author. The technology of the GTC discussed in this paper is
provided by Rio Tinto Alcan. The process, equipment, and their categorization are dis-
cussed in this section.
2.1 Process Description
The process inputs to the GTC: The gases evacuated from potline to be treated, and the
reagent (fresh alumina) and dry-compressed air from the plant network are used for
actuators and filter media cleaning. Fresh air is added to the process by the dilution
dampers when it is required. Ambient air is taken by two blowers for filter hopper flu-
idization and another two blowers for air slide and distribution box fluidization.
Maintenance Strategy for Reliable Instruments in Gas Treatment Center (GTC… 397
Fig. 1. Schematic diagram of GTC.
Primary adsorption takes place inside the reactor and the inlet temperature of gas
affects the scrubbing efficiency with maximum of 137 °C. The reagent forms a cake on
the dirty side of filter bag, adding to the filtering efficiency, where secondary adsorption
takes place. The reagent when no longer could absorb the gases, is considered to be
enriched alumina, partially recirculated to the reactor with fresh reagent.
The process outputs of the GTC: The cleaned potline gas released out to the atmo-
sphere through the large diameter stack. The enriched alumina is partially transported
by air slide and airlifts to storage for use as the process material in the production of
Aluminum. The schematic diagram of GTC is given in Fig.1.
The process input to the FCN system is ambient air to pressurize the cooling air
headers, to provide low-pressure compressed air to force cool the pot shells. The output
is the cooling air available in header dedicated to specific pots.
2.2 System and Subsystem Equipment Description
The GTC is divided into the subsystems, explained in the following passages:
Gas collection subsystem includes the collection ductwork, 8 booster fans, the diverter
gates located at each pot, and the 6 fresh air dilution dampers with fans. During
normal operation, every pot needs an extraction of gas at a flow of 2.6 Nm3/s and
doubles the rate during maintenance, achieved by engaging booster fan. When
required to bring the gas temperature down, pot gas is diluted with ambient air via
four dilution dampers, with the aid of two fans and a backup fan for extreme
conditions.
398 S. J. Jeyaraj
Filter module subsystem is the heart of the GTC that includes the 18 scrubber modules
(operate in parallel) and the fresh alumina distribution box with distribution system.
Each scrubber module is equipped with an integrated venturi reactor with gas inlet
connected to common inlet duct. The reagent (fresh alumina) is injected into the throat
of the venturi reactor, and partially used reagent (enriched alumina) is recycled inter-
nally back to the reactor. The reactor is designed for optimum mixing of inlet gas and
the reagents before it enters the filter module. The dirty side (reactor) is separated from
the clean side (downstream of filter module) by 960 filter bags, arranged in 48 rows.
Gas evacuation subsystem includes 6 Induced Draft (ID) fans, stack, and common inlet
duct and outlet plenums. The ID fans generate the necessary draft at the electrolysis
pots. By modulating the inlet dampers, the draft pressure at the two main inlets to the
GTC can be regulated.
Fluidization subsystem includes 4 fluidization blowers that generate compressed air
required to fluidize the filter hopper beds, the various air slides, and the alumina
distribution box. Two blowers are dedicated to the air slides and distribution box and
two blowers to the filter hopper beds.
FCN system is an independent subsystem that generates the compressed air required to
force cool the pot shells. There are 8 networks (2 per pot room), each with 6 fans,
taking ambient air and providing cooling air to the 8 network headers.
2.3 Equipment Criticality Assessment and Classification
To prioritize the maintenance of various assets in GTC, the assets are categorized based
on operation, maintenance, and safety criticality. The risk is qualitatively calculated,
from severity index of consequence and probability index of a functional failure. The
Risk Assessment (RA) is done by the cross-functional team from Reliability, Inspection,
Technical, Operation, Safety, Engineering, Environment, and Maintenance. Standard
RA procedure is followed to assign criticality index [7].
There are 942 equipment in GTC, and criticality assessment is done for the equip-
ment and categorized into criticality categories AC0, AC1, AC2, AC3, and AC4, and
description is given in Table 1.
The assets classified as AC3 and AC4 are assets that have low or acceptable risk of
failure and have least priority for maintenance [6]. Therefore, the 331 equipment that
are classified under AC0, AC1, and AC2, would be analyzed.
2.4 Generic Classification of Equipment in GTC
When we analyze further closely, it is observed that the generic type of equipment repeat-
ing in the group classified based on criticality. For example, in Table 1, 41 equipment are
categorized as AC1 and the equipment is enumerated in Table 2. It is found that only 7
generic equipment of same configuration and instrumentation supplied by 7 respective
vendors constitute 41 equipment of varying functions. So detailed strategy analysis
needs to be done for 7 generic equipment only, instead of 41 AC1-classified group [5].
Based on the generic equipment classification, 331 functional equipment classified
as AC0, AC1, and AC2 are represented by 37 generic equipment, as shown in Table 3.
Table 1. Equipment criticality classification and number of instrument attached.

Criticality index Risk level Total equipment Attached instruments
AC0 EHS critical 175 1268
AC1 High critical 41 85
AC2 Moderate critical 115 322
AC3 Low critical 608 947
AC4 Acceptable 3 3
Total 942 2625
Table 2. Detail of AC1-categorized equipment.

Group# Generic equipment Numbers
1 Airlift – Fluorinated Alumina 1
2 High- Dense Phase System (HDPS) Cover Material 4
3 Dense phase Vessel 8
4 Air Gravity Conveyer (AGC) 2
5 AGC Aerator Fans 2
6 Main HDPS Feed System Pot Distribution Fan 12
7 Main Fans 12
Total 41
Table 3. Criticality classification and generic equipment.
Equipment criticality Total equipment Generic

AC0 175 22
AC1 41 7
AC2 115 8
Total 331 37
Therefore, in this reported work, extensive analysis was done on instruments attached
to very a smaller number of generic equipment, comparing 942 equipment initially
enumerated in Table 1.
3 Instrument Classification
With reference to Table 1 column 4, there are 2625 instruments in GTC. The failure
modes and failure data depend on the instrument type, service, and environment. So the
instruments are classified based on the measurement principle, type of transducer used,
and signal conditioning features. In this section, the critical functions are performed by
these instruments in GTC and the classification is presented.
Inlet temperature to the GTC is measured to control and maintain the same at a
maximum of 130 °C, for optimum adsorption. The fresh alumina flow is monitored, to
ensure a continuous supply of reagent to the reactor. The differential pressure across the
clean and dirty sides of the filter module is monitored. The flow to the filter is measured
for each filter module, and the inlet dampers are controlled, to have balanced process.
400 S. J. Jeyaraj
At the outlet side of the filter module, the presence of particulate in the gas stream is
monitored, giving an early indication of bag failure and long-term performance of the
filter bags. This differential pressure between inlet and outlet of filter is proportional to
thickness of the reagent cake forming on the filter bags. After a preset differential pres-
sure limit is reached, 2 of the 48 rows (filter bags) are automatically cleaned.
Level of alumina in 18 distribution boxes are monitored for replenishing the fresh
alumina. The distribution box is equipped with 18 outlet ports, having an orifice plate
that is used to control the flow at each outlet port that feeds the siphon box. By fluid-
izing or not fluidizing the siphon box, the flow of alumina is enabled or disabled for the
respective filter reactor.
The draft pressure at the two main inlets to the GTC is measured and controlled by
modulating the dampers of ID fans. Outlet temperature, perfluorocarbons content, and
gas flow are measured at a sampling platform, located on the stack.
Fluidizing blower is equipped with an intake air filter, and differential pressure
across the filter is monitored, which gives an indication when the filter element needs
replacement.
The blower motors are equipped with built-in electric heaters to prevent interior
condensation. The blower motors are protected by motor protection relays, monitoring
the motor winding temperature, motor bearings temperature, blower bearings tempera-
ture, and blower bearing vibration. FCN fan motors are equipped by similar protection
as blowers.
The specification and name plate details of instruments are collected from instru-
ment index database and verified by site inspection. The instruments in GTC are classi-
fied into 57 types based on functionality, make, and model [8]. A sample listing is
furnished in Table 4.
Table 4. Generic instrument types based on function, make, and model (sample list).
Type of instrument Number of instruments Make Model
Limit Switch- Damper 112 Schneider XCKM115
DP Switch 16 Dwyer DHII-008
DP Transmitter- Type-2 133 Endress+Hauser Deltabar FMD77
Flow Element (Multiphase) 144 Solios 532-040-0340
Flow Transmitter- Gas 72 Endress+Hauser PMD55
Fluoride Analyzer 4 Unisearch Not available
Gas Particle Analyzer 12 Auburn Systems TRIBO.dsp U3600
Level Switch 127 Endress+Hauser FTM 50/51
Particle Analyzer Probe 72 Solios 532-040-0370
Position Transmitter 152 Centork YT-2500
Pressure Switch- Type-1 216 Endress & Hauser PTP31
Rope Switch 6 Listenow IG50-2A
Speed Switch 64 IFM Electronic DI0001
Temperature Transmitter 100 Endress & Hauser TR62
Valve Position Switch 20 Pepperl+Fuchs NCB8-18GM50-Z4
More instrument types… 1375 … …
Total (57 types) 2625 … …
4 Instrument Maintenance Strategies and Recommendations
Instrument failure modes are sensor (electromechanical) failure, transmitter (random/

electronics) failure, and measurement deterioration (calibration failure) [9]. The PM
strategies to eliminate these failures, RCM analysis, and recommendation are presented
in this section.
4.1 Visual Inspection and Maintenance Activities
Electromechanical failures are due to environmental condition and operation lifetime of

sensor components that can be prevented by visual inspection, maintenance, and peri-
odic replacement. Instruments that cannot be reached or physically inspected while in
operation or do not have isolation or bypass facilities require visual inspection sched-
uled during Shutdown.
Maintenance activities include replacing minor components such as seals, gaskets,
run to fail components, setting, aligning, and adjusting components, tightening fasten-
ings to the required torque, removing instruments, and re-connecting instrumentation
with impulse line or power supply, labeling/marking of components, checking signal
transmission is satisfactory, replacing or repairing damaged or defective electrical and
mechanical components and backup batteries [10].
4.2 Function Testing Activities [12]
Random or electronics failures in instruments can be diagnosed by the automation sys-

tem, using error codes and Namur signal standards. Intelligent instruments that employ
communication protocol such as HART, FF, and MODBUS with failure diagnostics and
alarm function also are used to conduct function test and diagnose instrument faults.
Hidden failures can be detected and rectified by function test PM, testing the power
supply, signal sources, pressure sources, analog or digital meters, manometers, and
digital pressure indicators. Function testing can be Instrument function test or interlock
function test, as discussed below:
Instrument Function Check in meant for instruments that do not require calibration and
require only testing the instrument device functionality. The instruments covered are
hand switches, push buttons, lamps, hooters, buzzers, level position/limit/proximity
contact switch, etc.
Interlock Function Check refers testing instruments that are part of Safety Instrumented
Function (SIF) or interlock functions. Entire safety loop, i.e., initiator, logic solver,
and final control elements are tested by physically creating the process condition and
verifying the intended function is performed by the SIF [11]. Trip or abnormal con-
dition is created in a controlled situation and confirming that the functionality at the
final output devices, such as trip relay, valves, dampers, and actuators, uncovers hid-
den failure in any of the components of safety loop [12].
402 S. J. Jeyaraj
4.3 Calibration
The deterioration of accuracy and precision of instruments are due to continuous usage
of instruments beyond calibration period, recommended by the OEM or calibration
standard of the organization. By following the calibration PM as per the strategy, these
failures can be avoided [13].
Calibration PM is recommended for all transmitters used for indication and control.
Instruments that need be removed and calibrated should have isolation or bypass facil-
ity (field mounted switch) or the equipment should have redundant or stand-by instru-
ments. Inventory analysis should be done if an instrument needs to be removed and
calibrated from standard laboratory with considerable lead time.
4.4 RCM Analysis and Instrument PM Strategy Recommendation
In this subsection, the RCM analysis and instrument PM strategy recommendations are
discussed. The 57 generic instruments categorized in Table 4 of Sect. 3 were analyzed
referring the Operation and Maintenance (O&M) documents. The failure modes and
failure data from the vendor records and documents have been used in RCM analysis to
select inspection, function test, or calibration strategy and to determine the cycle time.
The recommended strategies through RCM analysis (a sample is in Table 5) are inte-
grated inspection and function test at frequency 91 days, 182 days, 364 days, etc. and
calibration at site or local lab or offsite lab at frequency of 364 days, 728 days, etc.
The validity of above-mentioned recommendations was discussion in detail for 37
generic equipment that were considered critical (AC0, AC1, and AC2) in Table 3 of
Sect. 2. Final recommendation for attached instruments to the respective equipment
was generated.
For example, DUST COLLECTOR–FRESH ALUMINA SILO, a generic equip-
ment, is categorized as AC0. There are 2 such DUST COLLECTORs that are in GTC,
Table 5. Generic PM strategy recommendation.

Name Primary function Cause Strategy recommendation
Position Switch – Positioning truck Environment, 91 Days – Cleaning,
LISTENOW(Make) and loading sensor random, Inspection and Function
E-20 (Model) electronic Test of Positioning truck
failure loading sensor
Position Transmitter – Positioners Environment, 364 Days – Cleaning,
CENTORK(Make) and electromechanical random, Inspection, and Function
YT-2500 (Model) electronic Test of Positioners
failure electromechanical
Pressure Switch- Type-1 – Intelligent pressure Environment, 182 Days – Calibration
ENDRESS & switch with random, at Local/ Workshop of
HAUSER(Make) and PTP31 piezoresistive sensor, electronic Pressure Switch
(Model) overload resistance failure
Pressure Switch- Type-2 – Pressure transmitter Environment, 364 Days – Calibration
IFM ELECTRONIC(Make) with trip setting random, Local/ Workshop of
and PN2224 (Model) electronic Pressure transmitter with
failure trip setting
supplied by the same vendor with exactly same configuration. This Dust Collector has
one DP Transmitter Type-1 and two, Level Probe (Silo). The following instrument PM
strategies were recommended, based on the RCM analysis:
• Calibration at Local/ Workshop of DP transmitter (Flow application) in cycle of
364 days
• Cleaning, Inspection, and Function Test of Level switch in cycle of 726 days
The following number of recommendations is generated based on the analysis:
• Inspection and Function test (varying cycle) – 282 Recommendations
• Calibration at site (varying cycle) – 90 Recommendations
• Calibration at own facility onsite (varying cycle) – 238 Recommendations
• Calibration at standard lab offsite (varying cycle) – 492 Recommendations
5 Implementation of Instrument Maintenance Strategies
In this section, the details of implementation of recommended PM strategies for instru-

ments in GTC are discussed.
5.1 ERP System Implementation
The instrument asset items were created in the ERP system and linked to the parent
assets. The failure modes were programmed as dropdown menu for each asset, to record
the failure mode in maintenance log. This information will be useful for calculation of
failure rate and MTBF, to review and update RCM analysis. Standard work instructions
were developed for each PM activity.
5.2 Instrument Inventory Analysis
Before scheduling the calibration PM, the spare instruments availability was studied.
The instruments in the material management system were reviewed and items were cre-
ated if not available. The instrument tag, type code, OEM description, make model,
instrument parameter, unit, span, and output were populated in the database. The
installed quantity and available spare were considered before implementation. Purchase
request for the spare instruments for calibration is placed.
5.3 Resource Assessment
Since the execution of the recommended PM activities requires additional skilled

resources, the staff requirement was analyzed in advance. Requirement of 2268 person-
hours per annum was predicted to execute 567 inspection and function tests PM. Annual
requirement of 5076 person-hours was predicted for calibration PM. Accordingly, the
personnel was mobilized and trained before scheduled execution.
404 S. J. Jeyaraj
6 Conclusion
It seemed the analysis, recommendation, and implementation would take long time and
extra resources, because of the substantial number of assets and instruments, and so it
was planned to outsource this project. When the idea and method (presented here) were
proposed by the management, the organization approved the project to utilize the inter-
nal resources. The RCM analysis was done for 57 instrument types, and detailed valida-
tion workshop was conducted for 37 generic equipment. It took 12 weeks (about
3 months) to complete the analysis and recommendation (time for data collection,
equipment criticality assessment, and implementation are not included) using available
resources. The same method can be used for other sections of smelter like Billot section
of cast house and baking furnace of carbon.
References
1. The Age of Aluminium. Out of acorns grow big trees. Alum Int Today. 2017:2–3.
2. UC Rusal. The history of aluminium industry.
3. https://www.aluminiumleader.com/history/industry_history/. Accessed 1 March 2022.
4. Letot C, Dehombreux P. Actualized reliability model based on degradation data and maintenance
optimization. 2nd IFAC Workshop on Advanced Maintenance Engineering, Spain. November
22–23, 2012, pp. 211–6.
5. Brough D, Jouhara H. The Aluminium industry: a review on state-of-the-art technologies, envi-
ronmental impacts and possibilities for waste heat recovery. Int J Thermofluid. 2020;1–2:100007.
https://doi.org/10.1016/j.ijft.2019.100007.
6. Storm K. Industrial piping and equipment estimating manual. Elsevier Inc.; 2017. ISBN:
978-0-12-813946-2
7. Smith DJ. Reliability, maintainability and risk. Elsevier Inc.; 2017. ISBN 978-0-08-096902-2
8. Kelly A. Maintenance systems and documentation. Elsevier Inc.; 2006. ISBN 13: 978 0 75 66995 5
9. Smith AM, Hinchcliffe GR. RCM-gateway to world class maintenance. Elsevier Butterworth–
Heinemann; 2004. ISBN 0-7506-7461-X
10. Goettsche LD. Maintenance of instruments & systems. International Society of Automation; 2005.
11. Smith R, Mobley K. Rules of thumb for maintenance and reliability engineers. Butterworth-
Heinemann, ISBN 0750678623; 2007.
12. Jeyaraj SJ, Al Shehri AI. Framework for Life Cycle Management of Safety Instrumented System
(SIS) in manufacturing facilities of large corporation. 2022 4th International Conference on
Applied Automation and Industrial Diagnostics (ICAAID), March 29–31, 2022, Saudi Arabia.
https://doi.org/10.1109/icaaid51067.2022.9799490.
13. Ahola R. Ultimate calibration. 2nd ed. Beamex Oy Ab; 2009.
Experimental and Numerical Study of 3003-H14
Aluminum Alloy-Welded Joints with Different
Clamping Conditions
M. Hernández1, R. R. Ambriz1, C. García1 (*), C. Gómora1, and D. Jaramillo1

1
Instituto Politécnico Nacional CIITEC-IPN, Mexico City, Mexico
cjgarcia@ipn.mx
Abstract. The present work analyzes thermal and mechanical processes induced by
GMAW welded joints of Al alloy 3003-H14, under restricted and unrestricted clamping
conditions. An ANSYS FE model was developed, where a moving double ellipse heat
source was implemented to capture the temperature distribution induced by the welding.
Residual stresses were resolved by the FE model by coupling the thermal analysis with a
structural one. A temperature error of less than 10% was obtained between the experi-
mental measurements and FE results. In the case of the residual stresses, the best correla-
tion between the experimental and numerical results were for the longitudinal component
in the unrestricted condition with a maximum root mean square error (RMSE) of 1.95
MPa. The maximum RMSE in the transverse component was for the restricted condition
with a value of 22.74 MPa. In general, a good correlation was observed between the FE
model and the experimental measurements.
Keywords: GMAW process · FE model · Residual stress · Weld thermal cycles
1 Introduction
Gas Metal Arc Welding (GMAW) process can be used to weld aluminum alloys by
means of an intense power source and a filler wire material. The heat-moving source
induces large thermal gradients and different cooling rates in the welded plate. Residual
stresses are formed as a result of these gradients. The first analytical models for the
heat-moving source were proposed by Rosenthal [1]. Goldak et al. [2] analyzed the
heat-moving source by using a double-ellipsoidal geometry model. Lu et al. [3] reported
a 3D finite-element model for welded joints by GMAW.
The present work analyzed weld thermal cycles and residual stresses produced in
3003-H14 aluminum alloys plates welded by GMAW process with a fully and free
restriction condition. A thermo-mechanical finite-element model was developed and a
double-ellipsoidal model was used to define the heat flux during the welding process.
https://doi.org/10.1007/978-3-031-17425-4_53
406 M. Hernández et al.
Fig. 1. Heat flux distribution according to the double-ellipse model [4].
2.1 GMAW Process
Aluminum alloy plates of 3003-H14 with dimensions of 300 × 100 × 6.6 mm were
welded by GMAW. ER4043 filler material and direct current positive electrode were
used. Welding procedure specifications have been previously reported in the literature
[4]. Type-K thermocouples were used to measure the weld thermal cycles, and strain
gage rosettes were used to determine the post-welding residual stress.
2.2 Finite-Element Model
A thermal transient analysis coupled with a structural one was set in ANSYS APDL
(academic version 19.0). A double-ellipse model was used to determine the heat flux in
the finite-element model. Figure 1 shows the heat flux set in the finite-element model as
a function of distance from the center of the weld pool.
3 Results
The finite-element (FE) model was able to capture the weld thermal cycle, with just a
difference of 2% for the peak temperature with respect to experimental measurements.
Figure 2 presents the uniaxial distribution for the normalized transverse residual stresses
along three different paths (a–b, c–d, and e–f) as determined by the FE model.
Experimental and Numerical Study of 3003-H14 Aluminum Alloy-Welded Joints… 407
Fig. 2. Transverse residual stresses distribution and experimental data value. (a) Unrestricted and
(b) restricted clamping conditions [4].
4 Conclusions
The double-ellipsoidal model was implemented for the analysis of the heat-moving
source in GMAW process. A good correlation was found between the finite-element
results and experimental measurements. Transverse residual stresses were larger in the
fully restricted conditions, but they changed along the weld bead.
Acknowledgments The authors thank to CONACyT-México (project A1-S-27474) and SIP-IPN for
supporting this research.
References
1. Rosenthal D. The theory of moving sources of heat and its application to metal treatments. Trasn
ASME. 1946;68:849–66.
2. Goldak J, Chakravarti A, Bibby M. A new finite element model for welding heat sources. Metall
Mater Trans A. 1984;15:299–305. https://doi.org/10.1007/bf02667333.
3. Lu Y, Zhu S, Zhao Z, Chen T, Zeng J. Numerical simulation of residual stresses in aluminum alloy
welded joints. J Manuf Process. 2020;50:380–93. https://doi.org/10.1016/j.jmapro.2019.12.056.
4. Hernández M, Ambriz R, García C, Jaramillo D. The thermomechanical finite element analysis of
3003-H14 plates joined by the GMAW process. Metals. 2020;10(6):786. https://doi.org/10.3390/
met10060708.
Out-of-Plane Bending Assessment of Heat-Treated
Aluminum Alloys Subjected to Free-Fall Impact
Loads
J. A. Frutos1, E. R. Barragán1, R. R. Ambriz1 (*), C. García1, C. Gómora1,

and D. Jaramillo1
1
Instituto Politécnico Nacional CIITEC-IPN, Mexico City, Mexico
rrambriz@ipn.mx
Abstract. Energy absorption capacity under dynamic impact events is a relevant prop-
erty for structural materials used in the transport industry. Fractures produced by impact
events can result in catastrophic failures. The analysis of the shape, speed, and impact
angle of projectiles has been reported in the literature with respect to ballistic tests [1, 2].
Also, the impact phenomena on target plates have been studied in the literature [3, 4].
Keywords: Heat-treated aluminum alloys · Impact load · Out-of-plane bending
1 Introduction
Energy absorption capacity under dynamic impact events is a relevant property for
structural materials used in the transport industry. Fractures produced by impact events
can result in catastrophic failures. The analysis of the shape, speed, and impact angle of
projectiles has been reported in the literature with respect to ballistic tests [1, 2]. Also,
the impact phenomena on target plates have been studied in the literature [3, 4].
The present work analyzed the out-of-plane bending of 6061-T651 and 7075-T651
aluminum alloy plates subjected to dynamic impact loads. A property impact drop
tower was used, which has a capacity to apply impact loads of up to 1600 J with a maxi-
mum velocity of 10 m/s.
Square plates of the aluminum alloys 6061-T651 and 7075-T651 were fully clamped in
the target system of the used impact drop tower (Fig. 1).
The used projectile was a semi-spherical indenter with a 16.0 mm in diameter. A
measurement system was used to determine the force–displacement curves experienced
https://doi.org/10.1007/978-3-031-17425-4_54
410 J. A. Frutos et al.
Fig. 1. Aluminum alloy plate fully clamped in the target system of the drop tower.
Fig. 2. Force–displacement curves for different impact velocities at 200 J impact energy: (a) 6061-
T651, and (b) 7075-T651 aluminum alloys.
by the aluminum alloys plates during the impact tests. Impact tests were performed
under different velocities, but with the same energy impact value.
3 Results
Figure 2 presents the experimental determined force–displacement curves for the 6061-
T651 and 7075-T651 aluminum alloys. In general, larger forces were supported by the
7075-T651 aluminum alloy but with a limited capacity of deformation. The alloy 6061-
T651 exhibited a larger capacity to absorb out-of-plane bending, especially for the low-
impact velocities. For instance, the out-of-plane bending was of 6 mm for the 6061-T651
Al alloy, while that for the 7075-T651 was of around a 2 mm. This resulted in a larger
Out-of-Plane Bending Assessment of Heat-Treated Aluminum Alloys Subjected… 411
energy absorption capacity of the 6061-T651 aluminum alloy when compared to the
6061-T651. For reference, it is also presented the quasistatic behavior of the aluminum
plate subjected to out-of-plane bending loading.
4 Conclusions
The 6061-T651 exhibited a superior capacity to support out-of-plane bending without

fracture of the material under quasistatic and dynamic applied loads. This contrast with
the nominal superior mechanical properties of the 7075-T651 aluminum alloy when
compared with the 6061-T651 alloy.
Acknowledgments The authors thank to CONACyT-México (project A1-S-27474) and SIP-IPN for
supporting this research. Alvaro Frutos and Edú Barragán are thankful to CONACYT for the scholar-
ship provided for their doctorate studies.
References
1. Iqbal MA, Khan SH, Ansari R, Gupta NK. Experimental and numerical studies of double-nosed
projectile impact on aluminum plates. Int J Impact Eng. 2013;54:232–45. https://doi.org/10.1016/j.
ijimpeng.2012.11.007.
2. Senthil K, Iqbal MA. Prediction of superior target layer configuration of armour steel, mild steel
and aluminum 7075-T651 alloy against 7.62 AP projectile. Structure. 2021;29:2106–19. https://
doi.org/10.1016/j.istruc.2020.06.010.
3. Jones N. Dynamic inelastic response of strain rate sensitive ductile plates due to large impact,
dynamic pressure and explosive loadings. Int J Impact Eng. 2014;74:3–15. https://doi.org/10.1016/j.
ijimpeng.2013.05.003.
4. Zhang Y, Li Y, Guo K, Zhu L. Dynamic mechanical behavior and energy absorption of aluminum
honeycomb sandwich panels under repeated impact loads. Ocean Eng. 2021;219:108344. https://
doi.org/10.1016/j.oceaneng.2020.108344.
A Study on Aluminum-Based Lightweight Entropic
Alloys with High Strength at Elevated Temperature
Liying Cui1 (*), Zhan Zhang1, Dilip K. Sarkar1, Duygu Kocaefe1,
and X. -Grant Chen1
1
Department of Applied Science, University of Quebec at Chicoutimi, QC, Canada
liying.cui1@uqac.ca
Abstract. Entropic alloys are a category of emerging advanced materials derived from
the novel design concept of high-entropy alloys, which are also known as multicompo-
nent alloys or complex concentrated alloys. In this study, two Al-based lightweight entro-
pic alloys, Al74Cu16Cr1Zn7Mg2 and Al78Cu18Cr2Zn1V1, were designed using a strategy
based on physical thermodynamic parameters and Thermo-Calc simulations. The micro-
structures of the as-cast alloys revealed a multiphase feature, including fcc-Al solid solu-
tion and several intermetallic compounds. The maximum compressive fracture strengths
at 20 °C of both alloys exceeded 620 MPa, while the compressive yield strengths at
300 °C were found to be >200 MPa; these values are two to three times higher than those
of conventional Al alloys. The strengthening mechanism was mainly arising from solid
solution hardening and intermetallic network hardening. After the thermal exposure at
300 °C for 100 h, an obvious change in appearance and composition of the high Mg-Zn
containing phase occurred in Al74Cu16Cr1Zn7Mg2 alloy, which deteriorated the yield
strength at 300 °C. In contrast, the Al78Cu18Cr2Zn1V1 alloy exhibited barely changed yield
strength at 300 °C with a value of as high as 201 MPa after the thermal exposure, indicat-
ing its excellent thermal stability and great potential for high-temperature applications.
Keywords: Aluminum-based entropic alloys · Intermetallic network · Compressive

properties · Thermal stability
1 Introduction
Traditional aluminum alloys are based on one principal element with the minor addition
of alloying elements to obtain desirable mechanical and material properties. This con-
ventional strategy limits the possibilities for full-range alloy development. In 2004, a
brand-new classification of alloys named high-entropy alloys (HEAs) was developed
by Yeh et al. [1] and Cantor et al. [2]. HEAs are equiatomic multicomponent alloys that
include five or more equimolar or near-equimolar elements, thereby, improving the
entropy of alloys and promoting the formation of solid solution phases. Many works
have been devoted to developing various HEAs with simple phase formation and excel-
lent properties [3]. Hereafter, using the concept of entropy, studies have been gradually
https://doi.org/10.1007/978-3-031-17425-4_55
414 L. Cui et al.
extended to new alloys named complex concentrated alloys, complex multicomponent

alloys, entropic alloys, etc. [4].
Developing lightweight and low-cost structural alloys with outstanding high-
temperature performance has attracted a great attention in recent years. Lightweight
entropic alloys with a vast alloy design space have offered researchers’ new avenues to
explore novel materials. Many investigators have studied lightweight entropic alloys
[5], and aluminum becomes the first choice of necessary elements. Due to the big dif-
ference in chemical properties among light elements, the lightweight entropic alloys
usually have a more complex phase constitution than the high-entropy alloys [5].
Nevertheless, excellent mechanical properties such as high strength and good plasticity
were achieved in lightweight entropic alloys. The Al20Be20Fe10Si15Ti35 alloy, with a den-
sity of 3.91 g/cm3, showed an extremely high hardness value of 911 HV [6]. The com-
pressive strength of the Al80Li5Mg5Zn5Cu5 alloy reached 800 MPa, with a plastic strain
up to 17% [7]. The Al80Zn14Li2Mg2Cu2 alloy exhibited high compressive strengths of
more than 1 GPa and high plasticity of more than 20% [8]. However, until now there are
scarce reports focusing on the elevated-temperature mechanical properties of Al-based
lightweight entropic alloys and their thermal stability.
In this study, two Al-based lightweight entropic alloys, Al74Cu16Cr1Zn7Mg2 and
Al78Cu18Cr2Zn1V1, were designed and studied. The microstructure characteristics,
phase formation, and mechanical behavior at room and elevated temperatures were
investigated. In addition, the thermal stability of the alloys was verified by prolonged
thermal exposure at elevated temperature. Differences of the mechanical behavior
between two alloys are elucidated, and the mechanism behind is discussed herein.
In the alloy design of Al-based entropic alloys, Thermo-Calc software and TCAL7
database were used to calculate the equilibrium phases as a function of temperature. Cu
was selected to form eutectic microstructure with Al in the alloys, because the eutectic
microstructure can help improve the castability that is relatively poor in most entropic
alloys [9, 10]. The type and proportion of elements were adjusted based on the physical
thermodynamic parameters, e.g., mixing entropy and mixing enthalpy [11]. In the pres-
ent study, two alloys, Al74Cu16Cr1Zn7Mg2 (Alloy-1) and Al78Cu18Cr2Zn1V1 (Alloy-2),
were designed and investigated. They were melted in an induction melting furnace
under an argon atmosphere and cast into a permanent copper mold. To ensure the chem-
ical homogeneity, each ingot was remelted at least three times.
The microstructure of the alloys was examined by an optical microscope (OM,
Nikon Eclipse ME600) and a scanning electron microscope (SEM, JEOL JSM-6480LV)
equipped with an energy dispersive X-ray spectrometer (EDS). The phase identification
was performed by an X-ray diffractometer (XRD, D8 Discover) using Cu Kα radiation,
with a diffraction angle (2θ) range from 10° to 80°, a step size of 0.05°, and a step time
of 0.5 s/step.
The mechanical properties were studied at room temperature (RT) and 300 °C by
conducting compression tests in a Gleeble 3800 thermomechanical simulator unit.
A Study on Aluminum-Based Lightweight Entropic Alloys with High Strength… 415
Cylindrical samples were used for the compression tests with a diameter of 8 mm and
length of 12 mm. A part of the as-cast samples was thermally exposed at 300 °C for
100 h. At least two samples at the same condition were tested to verify the repeatability
of the results.
3.1 Predicted Phase Formation of the Alloys
According to the simulation of Thermo-Calc in Fig. 1, Alloy-1 consisted of five differ-

ent phases. Al2Cu was the main phase, followed by fcc-Al. Al45Cr7 was predicted to
form at 600 °C and remain stable afterward. Two kinds of Mg-Zn phases (Mg2Zn11 and
MgZn2) also appeared. The predicted phases in Alloy-2 were simpler than in Alloy-1.
Al2Cu was the predominant phase, followed by fcc-Al, Al45Cr7, and a very small amount
of hcp-Zn. The Thermo-Calc simulation indicated a multicomponent microstructure of
both alloys.
3.2 As-Cast Microstructure and Phase Constituents
The XRD patterns of as-cast Alloy-1 and Alloy-2 were very similar, as shown in Fig. 2.
Only three kinds of phases were detected, which were indexed as Al2Cu, fcc-Al, and
Al45Cr7, respectively. These three phases were the most abundant phases in both alloys
predicted by Thermo-Calc in Fig. 1. According to the intensity of the diffraction peak,
Al2Cu was the major phase, followed by fcc-Al, then Al45Cr7, which is in accordance
with the simulation results in Fig. 1. The alloys also contained other minor phases,
Fig. 1. Phase predictions of the two designed alloys calculated by Thermo-Calc: (a) Alloy-1
(Al74Cu16Cr1Zn7Mg2), (b) Alloy-2 (Al78Cu18Cr2Zn1V1).
416 L. Cui et al.
Fig. 2. XRD patterns of the as-cast Alloy-1 and Alloy-2.
Fig. 3. SEM images of as-cast microstructure, (a) Alloy-1 and (b) Alloy-2.
which were not detected by XRD due to the low content. Therefore, SEM was further
used to characterize the phase features.
Figure 3 showed the as-cast microstructure of two alloys. Based on morphology and
brightness, four kinds of phases were identified in as-cast Alloy-1 in Fig. 3a, denoted as
A, B, C, and D, respectively. According to the SEM-EDS results listed in Table 1, phase
A contained 90 at.% Al, 5 at.% Cu, 3 at.% Zn, and 2 at.% Mg, which was determined
as the fcc-Al matrix. The high content of Cu, Zn, and Mg in the matrix indicates a high
solid solution. Correspondingly, phases B and C were identified as Al2Cu and Al45Cr7.
At the same time, a very small amount of phase D was detected, which contained simi-
Table 1. Chemical composition and volume fraction of different phases in the as-cast alloys.
Chemical composition
Density (g/ Region in (at.%) Phase Volume fraction
Alloy cm3) Fig. 3 Al Cu Cr Zn Mg V name (vol.%)
Alloy-1 3.75 A 90 5 – 3 2 – fcc-Al 33.45
B 70 29 – 1 – – Al2Cu 58.33
C 79 9 10 2 – – Al45Cr7 7.67
D 74 15 – 5 6 – Mg-Zn 0.55
Alloy-2 3.73 A 95 4 – 1 – – fcc-Al 30.11
B 69 31 – – – – Al2Cu 54.53
C 82 5 8 1 – 4 Al45Cr7 15.36
lar amount of Zn and Mg content. The volume fraction of each phase was quantified by
image analyzing on at least 5 OM images. It can be found that phase D in Alloy-1 only
took up 0.55 vol.%, much less than either of the MgZn2 or Mg2Zn11 predicted in Fig. 1,
which could not be detected by XRD. Therefore, the phase D was temporarily referred
as Mg-Zn in this work.
As-cast Alloy-2 in Fig. 3b contained three phases, dark phase A – fcc-Al, bright
phase B – Al2Cu, and gray phase C – Al45Cr7. Compared with Alloy-1, Alloy-2 showed
a more obvious eutectic microstructure consisting of fcc-Al and Al2Cu. The content of
Al45Cr7 in Alloy-2 was 15.36 vol.%, higher than that in Alloy-1 (7.67 vol.%), which was
well supported by the Thermo-Calc simulation in Fig. 1.
From Table 1, both Alloy-1 and Alloy-2 were composed of the three main phases:
Al2Cu followed by fcc-Al and Al45Cr7. And both alloys showed a lightweight property,
with a low density of 3.75 and 3.73 g/cm3, respectively. It is apparent that results of
Thermo-Calc simulation, XRD pattern, and SEM were comparable and supportive to
each other.
3.3 Mechanical Properties and Thermal Stability
Compression tests of as-cast Alloy-1 and Alloy-2 were first conducted at room tempera-
ture, and the results are listed in Table 2. Alloy-1 and Alloy-2 exhibited a yield strength
(YS) σ0.2 of 588 and 387 MPa, respectively. Alloy-1 and Alloy-2 displayed similar frac-
ture strength σmax of 624 and 621 MPa, which are 2–3 times higher than that of conven-
tional Al cast alloys. The low-fracture strain εp of both alloys indicates a poor ductility.
The high strengths of as-cast Alloy-1 and Alloy-2 can be attributed to two reasons: one
is the high solid solution (SS) strengthening, and the other is the strong intermetallic
(IC) network in the two alloys. The large difference between YS values of the two
alloys can be ascribed to the different SS strengthening and distinct IC network in the
two alloys.
The SS strengthening (∆σSS) at room temperature can be calculated according to the
following equation [12, 13]
SS HC , (1)
418 L. Cui et al.
Table 2. Compressive mechanical properties of the as-cast alloys.

Alloy σ0.2 (MPa) σmax (MPa) εp (%)
Alloy-1 588 624 0.72
Alloy-2 387 621 1.05
Fig. 4. Yield strength of as-cast alloys and thermal-exposed alloys at 300 °C.
wherein C is the concentration of solute atoms, and H and α are positive constants for a
certain element. From Table 1, Alloy-1 contained more Cu, Zn, and Mg solute atoms in
fcc-Al matrix than Alloy-2 did. Therefore, the solid solution strengthening of Alloy-1
was higher than Alloy-2. In addition, Al2Cu in Alloy-1 was mainly in compact block
form, compared with the thin lamellar Al2Cu in Alloy-2, and the compact block form of
Al2Cu in Alloy-1 made the IC network stronger. It is reported that the block ICs are
typically the source of cracking [14]. Therefore, Alloy-1 showed a lower εp than
Alloy-2.
To study the mechanical properties of the alloys at high temperature, compression
tests of the as-cast alloys were conducted at 300 °C and the results are shown in Fig. 4.
The YS values of both as-cast alloys at 300 °C exceeded 200 MPa, showing excellent
mechanical strength at high temperature. It is reported that Al2Cu and Al45Cr7 can be
present at high temperature up to 520 °C [15]. The chemical composition of Al2Cu and
Al45Cr7 in the alloys after being exposed at 300 °C was checked by EDS, and it showed
the same results as listed in Table 1, which confirmed the stability of these two ICs.
Therefore, the high YSs, which far exceeded that of commercial Al cast alloys, was
believed largely due to the high contribution of the strong IC networks of Al2Cu and
Al45Cr7 at 300 °C.
In addition to elevated-temperature YS, the thermal stability is one important aspect
of the alloy applicable at elevated temperature. Thermal exposure experiment was car-
ried out to the as-cast alloys at 300 °C for 100 h, followed by the compression test at
300 °C, and the results are shown in Fig. 4. Alloy-2 exhibited a high YS of 201 MPa
after thermal exposure, which was barely changed compared with the as-cast alloy,
indicating its excellent thermal stability. However, Alloy-1 showed a low YS of
188 MPa, deceased largely from the YS at as-cast condition (225 MPa). To explore the
mechanism behind, the microstructure of thermal-exposed alloys was further observed,
as shown in Fig. 5.
OM images of Fig. 5 showed the microstructure after thermal exposure for both
alloys. Compared with the as-cast alloys whose OM images are embedded in the top-
right corners of Fig. 5a and b, a “new” phase E occurred in Alloy-1 after thermal expo-
sure (white arrows in Fig. 5a), while the main phases in thermal-exposed Alloy-2
remained unchanged and they exhibited the same chemical composition as the as-cast
alloy as indicated by EDS results. The “new” phase E in Alloy-1 had a volume fraction
of 2.6 vol.%, larger than phase D in the as-cast alloy. Moreover, it can be found that
phase E always existed around Al2Cu. Phase E displayed a bright color under backscat-
ter SEM mode (red arrows in Fig. 6), and the SEM-EDS result also revealed that phase
E had a high content of Zn and Mg, especially Zn, much higher than phase D in Table 1.
In addition, the Al2Cu phase in as-cast Alloy-1 contained Zn (1 at.%), while no Zn was
detected in the Al2Cu of thermal-exposed Alloy-1, indicating the Zn diffusion from
Al2Cu to phase E. In the meantime, Mg was barely detected in the matrix in the thermal-
exposed Alloy-1, which also diffused to the phase E. Phase E may be a result of phase
transformation from phase D, according to the Thermo-Calc simulation in Fig. 1, a
phase transformation from MgZn2 to Mg2Zn11 may occur after a long thermal exposure
at 300 °C; on the other side, phase E may just be the product of a change in the chemical
composition of phase D. Either of them indicates the poor thermal stability of the high
Mg-Zn containing phase D. Further experiments regarding this transformation, either
Fig. 5. OM images after thermal exposure at 300 °C for 100 h, (a) Alloy-1 and (b) Alloy-2.
420 L. Cui et al.
Fig. 6. SEM-EDS of thermal-exposed Alloy-1.
chemical compositionally or crystal structurally, will be conducted to give hints to

future research on the development of lightweight entropic alloys.
Phase E mainly located in the interfaces between Al2Cu and fcc-Al, which might
weak the original strong IC network structure. According to SEM-EDS results, Mg and
Zn solute atoms in the fcc-Al matrix partially diffused to phase E after thermal expo-
sure, causing a decrease of the SS strengthening in the matrix. These two factors
resulted in a big drop-off in YS of Alloy-1 after thermal exposure. Meanwhile, the main
phases in Alloy-2 were thermally stable, ensuring its high YS after thermal exposure.
This suggests that the simultaneous addition of Mg and Zn with a high content induced
an unsatisfactory high-temperature property and a low thermal stability, which should
be avoided in the lightweight entropic alloys for applications at elevated temperature.
To compare the elevated-temperature strength of different Al alloys, a parameter R
is introduced in this work:
0.2
R , (2)

wherein σ0.2 is the yield strength at 300 °C of alloys at the thermal exposure condition
of 300 °C/100 h, and ρ is the alloy density either published or estimated by chemical
composition [16]. It can be seen that with a certain limitation of the alloy density (for
example, 4.5 g/cm3) [17], the larger the R value is, the more the alloy conforms to the
lightweight and high strength, and the greater the potential of its application at elevated
temperature in the transportation (automobile and aerospace) industries.
Recently, several studies have been focused on the improvements of elevated-
temperature mechanical properties and the thermal stability of Al alloys [18–23].
Figure 7 shows the comparison of the R value and YS value of several laboratory
improved Al alloys with Alloy-1 and Alloy-2 in the present work. For example, the heat
treatment improved 3004 alloy exhibited a YS of 78 MPa and a R value of 29 kN⋅m/kg
Fig. 7. Yield strength at 300 °C after thermal exposure of 300 °C/100 h vs R of the alloys in this work
and several Al alloys in references.
after thermal exposure at 300 °C/100 h [20], the modified low Si/Mn concentration
206-L alloy had a YS of 93 MPa and a R value of 33 kN⋅m/kg [19], and the Mg-micro
alloyed 224 alloy displayed a high yield strength of 142 MPa and a good R of 51 kN⋅m/
kg [23]. Alloy-2 in this work showed a YS up to 201 MPa as well as a high R of 54
kN⋅m/kg, indicating its beneficial potential for applications at high temperatures.
4 Conclusions
1. Two Al-based lightweight entropic alloys, Al74Cu16Cr1Zn7Mg2 and Al78Cu18Cr2Zn1V1,

were designed and studied in this work. The as-cast microstructure of the two
entropic alloys revealed a multiphase feature. The phase prediction of the Thermo-
Calc simulation was in good agreement with XRD and metallographic results.
2. Al74Cu16Cr1Zn7Mg2 and Al78Cu18Cr2Zn1V1 alloys showed YS values of 588 MPa and
387 MPa at room temperature, and high YS values of more than 200 MPa at 300 °C,
respectively. The strengthening of both alloys was mainly contributed by the solid
solution strengthening and the intermetallic network hardening.
422 L. Cui et al.
3. After thermal exposure at 300 °C for 100 h, the YS at 300 °C of Al74Cu16Cr1Zn7Mg2

alloys deteriorated from 225 to 188 MPa, owing to the decreased SS strengthening
in the matrix as well as the poor thermal stability of the high Mg-Zn containing
phase. Therefore, the simultaneous addition of Mg and Zn with a high content is not
recommended in the lightweight entropic alloys for applications at elevated
temperature.
4. Compared with the as-cast alloy, the thermal-exposed Al78Cu18Cr2Zn1V1 alloy
showed a barely changed YS of 201 MPa at 300 °C, indicating its excellent thermal
stability due to the heat resistance of the main phases. Although the alloy density of
Al78Cu18Cr2Zn1V1 is moderately higher than conventional Al alloy, the high R value
of this alloy suggests its great potential for applications at high temperature.
Acknowledgments The authors gratefully acknowledge the financial support of the Natural Sciences
and Engineering Research Council of Canada (NSERC) under the Grant No. CRDPJ 514651-17 and
Rio Tinto Aluminum through the Research Chair in the Metallurgy of Aluminum Transformation at
University of Quebec at Chicoutimi.
References
1. Yeh JW, Chen SK, Lin SJ, et al. Nanostructured high-entropy alloys with multiple principal ele-
ments: novel alloy design concepts and outcomes. Adv Eng Mater. 2004;6:299–303. https://doi.
org/10.1002/adem.200300567.
2. Cantor B, Chang ITH, Knight P, et al. Microstructural development in equiatomic multicomponent
alloys. Mater Sci Eng A. 2004;375:213–8. https://doi.org/10.1016/j.msea.2003.10.257.
3. Zhang Y, Zuo TT, Tang Z, et al. Microstructures and properties of high-entropy alloys. Prog Mater
Sci. 2014;61:1–93. https://doi.org/10.1016/j.pmatsci.2013.10.001.
4. Miracle DB, Senkov ON. A critical review of high entropy alloys and related concepts. Acta Mater.
2017;122:448–511. https://doi.org/10.1016/j.actamat.2016.08.081.
5. Li Y, Zhang Y. Light-weight and flexible high-entropy alloys. In: Sharma A, Duriagina Z, Kumar
S, editors. Engineering steels and high entropy-alloys. London: IntechOpen; 2019. https://doi.
org/10.5772/intechopen.88332.
6. Tseng K, Yang Y, Juan C, et al. A light-weight high-entropy alloy Al20Be20Fe10Si15Ti35. Sci
China Technol Sci. 2018;61:184–8. https://doi.org/10.1007/s11431-017-9073-0.
7. Yang X, Chen SY, Cotton JD, et al. Phase stability of low-density, multiprincipal component alloys
containing aluminum, magnesium, and lithium. JOM. 2014;66:2009–20. https://doi.org/10.1007/
s11837-014-1059-z.
8. Li R, Ren Z, Wu Y, et al. Mechanical behaviors and precipitation transformation of the light-
weight high-Zn-content Al–Zn–Li–Mg–Cu alloy. Mater Sci Eng A. 2021;802:140637. https://doi.
org/10.1016/j.msea.2020.140637.
9. Lu Y, Gao X, Jiang L, et al. Directly cast bulk eutectic and near-eutectic high entropy alloys with
balanced strength and ductility in a wide temperature range. Acta Mater. 2017;124:143–50. https://
doi.org/10.1016/j.actamat.2016.11.016.
10. Shao L, Zhang T, Li L, et al. A low-cost lightweight entropic alloy with high strength. J Mater Eng
Perform. 2018;27:6648–56. https://doi.org/10.1007/s11665-018-3720-0.
11. Zhang Y, Yang X, Liaw PK. Alloy design and properties optimization of high-entropy alloys.
JOM. 2012;64:830–8. https://doi.org/10.1007/s11837-012-0366-5.
12. Ryen Ø, Holmedal B, Nijs O, et al. Strengthening mechanisms in solid solution aluminum alloys.
Metall Mater Trans A Phys Metall Mater Sci. 2006;37:1999–2006. https://doi.org/10.1007/
s11661-006-0142-7.
13. Li Z, Zhang Z, Chen XG. Improvement in the mechanical properties and creep resistance of
Al-Mn-Mg 3004 alloy with Sc and Zr addition. Mater Sci Eng A. 2018;729:196–207. https://doi.
org/10.1016/j.msea.2018.05.055.
14. Singh SS, Guo E, Xie H, et al. Mechanical properties of intermetallic inclusions in Al 7075
alloys by micropillar compression. Intermetallics. 2015;62:69–75. https://doi.org/10.1016/j.
intermet.2015.03.008.
15. Glazoff MV, Khvan A, Zolotorevsky VS, et al. Casting aluminum alloys: their physical and
mechanical metallurgy. 2nd ed. Butterworth-Heinemann; 2018.
16. Chauhan P, Yebaji S, Nadakuduru VN, et al. Development of a novel light weight
Al35Cr14Mg6Ti35V10 high entropy alloy using mechanical alloying and spark plasma sintering.
J Alloys Compd. 2020;820:153367. https://doi.org/10.1016/j.jallcom.2019.153367.
17. Campbell FC. Lightweight materials: understanding the basics, 4. ASM International; 2012.
https://doi.org/10.31399/asm.tb.lmub.t53550001.
18. Lin B, Li H, Xu R, et al. Thermal exposure of Al-Si-Cu-Mn-Fe alloys and its contribution to
high temperature mechanical properties. J Mater Res Technol. 2020;9:1856–65. https://doi.
org/10.1016/j.jmrt.2019.12.018.
19. Rakhmonov J, Liu K, Chen GX. Effects of compositional variation on the thermal stability of
θ’-Al2Cu precipitates and elevated-temperature strengths in Al-Cu 206 alloys. J Mater Eng
Perform. 2020;29:7221–30. https://doi.org/10.1007/s11665-020-05227-5.
20. Liu K, Chen XG. Development of Al–Mn–Mg 3004 alloy for applications at elevated tem-
perature via dispersoid strengthening. Mater Des. 2015;84:340–50. https://doi.org/10.1016/j.
matdes.2015.06.140.
21. Jin L, Liu K, Chen XG. Evolution of dispersoids and their effects on elevated-temperature strength
and creep resistance in Al-Si-Cu 319 cast alloys with Mn and Mo additions. Mater Sci Eng
A. 2020;770:138554. https://doi.org/10.1016/j.msea.2019.138554.
22. Farkoosh AR, Chen XG, Pekguleryuz M. Dispersoid strengthening of a high temperature Al–
Si–Cu–Mg alloy via Mo addition. Mater Sci Eng A. 2015;620:181–9. https://doi.org/10.1016/j.
msea.2014.10.004.
23. Rakhmonov J, Liu K, Pan L, et al. Enhanced mechanical properties of high-temperature-resistant
Al–Cu cast alloy by microalloying with Mg. J Alloys Compd. 2020;827:154305. https://doi.
org/10.1016/j.jallcom.2020.154305.
The Development of the HSBC (Horizontal
Single-Belt Casting) Process From Inception
to Present Day
D. R. Gonzalez-Morales1, M. M. Isac1, and R. I. L. Guthrie1 (*)
1
Roderick.Guthrie@mcgill.ca
Abstract. Horizontal Single-Belt Casting (HSBC) has been developed and studied for
the last thirty-five years, proving to be a viable alternative to the traditional casting pro-
cesses, e.g., Direct Chill (DC) and Conventional Continuous Casting (CCC), for alumi-
num and steel, respectively. This paper will summarize developments made in Canada
and Australia, ever since the 1980s to the present day. Theoretical and experimental work
will be summarized to provide the necessary processing parameters to cast a wide range
of alloys, in pilot plant and in industrial-scale operations. It is anticipated that HSBC will
prove to be a far more versatile, economic, and environmentally friendly option, versus
conventional casting technologies, for a world desperately needing to reduce carbon diox-
ide emissions from the metallurgical industries.
Keywords: CFD modeling · Horizontal Single-Belt Casting (HSBC) · Meniscus ·

Thin strips/sheets
1 Introduction
Horizontal Single-Belt Casting (HSBC), as a Near-Net-Shape Casting Process (NSSC),

has been under development for the last thirty-five years, as a viable alternative to con-
ventional processes traditionally used for producing thin strips for a wide range of
alloys. HSBC was one of two emerging NSSC technologies surviving from the last half
of the past century, for steel casting, the other being Twin Roll Casting (TRC) [1]. TRC,
or Bessemer Casting, has been practiced in the non-ferrous industry, at relatively low
rolling speeds, but never for ferrous sheet products, despite many attempts by steel
companies over the previous one hundred and fifty years since Bessemer’s patent in
1857. Finally, in 2008, TRC proved to be industrially successful for NUCOR for pro-
ducing NNSC steel sheet products [1]. On the HSBC side, this process was commer-
cialized by Salzgitter GmbH, in 2012, at their steel plant in Peine, Germany, under the
name “Belt Cast Technology.” More recently, the HSBC has also been used commer-
cially for producing thin strips of copper alloys, by Materion Brush, USA, following
earlier successful demonstrations in Canada, at the pilot scale, by MMPC/MetSim Inc.,
casting crystalline metals and alloys such as aluminum, steel, and copper alloys, and
https://doi.org/10.1007/978-3-031-17425-4_56
bulk amorphous sheet products. These two commercial successes illustrate that the two
processes, HSBC and TRC, are viable future candidates for replacing conventional
casting options.
The HSBC was independently conceived by R. Guthrie and J. Herbertson in 1988,
in Canada, and by the University of Clausthal research team lead by K Schwerdtfeger,
also in 1988. Significant technical support in the development of the HSBC by the
MMPC and MetSim was provided by M. Isac, since the return of the caster to North
America in 1999, from Australia. Similarly, K-H Spitzer made significant contributions
to the (Horizontal) Belt Cast Technology, at their pilot-scale laboratory of Clausthal
University, prior to commercialization by Salzgitter GmbH, in 2012.
The HSBC research at the MMPC McGill process metallurgy laboratories, followed
by continuing research and development work at its off-campus MetSim Inc. high tem-
perature melting casting and solidification laboratory, focused on processing ferrous
and non-ferrous, crystalline, and amorphous, 2–10 mm thick strips. The Single-Belt
Casting research in Germany focused on processing 10–15 mm thin strips of various
grades of steel that required electro-magnetic damping. Once the Hazelett single-belt
caster had been returned to North America, the McGill Metals Processing Center
(MMPC) took up experimental research with that pilot-scale caster, processing a large
variety of thin strips of steel (High Strength High Ductility steels, electrical transformer
steels, TRIP and TWIP steels), various grades of aluminum (AA6111, AA2024,
AA5182 etc.), and copper-Sn-Ni alloys.
The experimental work at the laboratory scale was validated by experimental work
at the pilot-scale level, always accompanied by Computational Fluid Dynamics (CFD)
studies, using Ansys Fluent software. The computational work was carried out on a
dedicated 288 core High-Performance Computer clusters at MetSim, enabling high-
speed computations. Figure 1 shows the casting of thin strips of AA2024 alloy on the
HSBC pilot-scale system, as an example of supporting experimental work. Such trials
Fig. 1. The HSBC pilot-scale caster operating at MetSim Inc., Montreal.

The Development of the HSBC (Horizontal Single-Belt Casting) Process… 427
at the pilot-scale level were conducted for a wide range of metal alloy systems, always
supported by extensive mathematical modeling.
The main advances in NNSC processes up until 2014 have been presented in two
reviews [2, 3]. The present publication focuses on the processing details required to cast
lightweight aluminum alloys, and lightweight bulk amorphous metals, since 2014 to the
present. The main processing parameters, including type of liquid metal-feeding sys-
tem type, belt speed, surface tension, contact angle interactions, and belt surface rough-
ness, all of which can be applied to full-scale industrial processes, will be discussed.
Experimental observations and mathematical modeling will also be discussed to pro-
vide the optimal parameters to process good-quality thin strip via the HSBC process.
2 Experimental Results Obtained at the Simulator, and Pilot-

Scale, Levels
The MMPC research group studied the possibility of casting three different Ca-Mg-Al
based, bulk metallic glass (BMG), and amorphous alloys, using the HSBC process [4].
The purpose was to take advantage of its sustained high cooling rate capabilities, and
horizontal output of BMG sheet. The authors used the HSBC simulator with a perfectly
flat copper substrate (Fig. 2), to produce 70 mm wide strips of BMG’s. Helium was
injected to enhance the heat transfer to obtain thicker amorphous strips, as helium was
expected to displace the air and offer a higher gaseous thermal conductivity. Customized
thermocouples were manufactured and included in the HSBC simulator’s cooling sub-
strate to measure instantaneous heat fluxes during solidification of these “glassy” alloy
systems. It was found that the helium only increased the maximum heat flux by around
40 to 70%, versus ~500%, since some air was evidently still entrapped within the gas
mixture contained within the peaks and valleys, on which the liquid metal was
freezing.
Fig. 2. HSBC simulator set-up used by Guthrie et al. [4] & Ge et al. [5].
To detect recrystallization, XRD (X-Ray Diffraction) was used for all the samples
obtained. For the ternary alloy, 60Ca-20Mg-20Al, an amorphous matrix was formed
only in the middle region of the strip of 0.8 mm thickness, while for a 1.2 mm thickness,
no amorphous matrix was observed at all. The high cooling rates (800 K/s) required to
produce a completely amorphous matrix for the complete strip could not be reached
with the HSBC simulator, and it was, therefore, concluded that the casting of a ternary
alloy was not viable via HSCB. However, with a quaternary alloy, 60Ca-15Mg-10Al-
15Zn, strips of 0.9 and 1.1 mm thickness were successfully cast, and they exhibited
completely amorphous matrices. For a 1.8–mm-thickness strip, some recrystallization
was detected. It was concluded that the possible thickness threshold for the studying
quaternary alloys was ~1.1 mm. By adding 5 atomic% copper, for a quinary alloy,
55Ca-15Mg-10Al-15Zn-5Cu, the maximum thickness to assure a completely amor-
phous matrix was increased to 2.0–3.0 mm. Additionally, it was discovered that impuri-
ties within the quinary alloy (chlorides from alloy preparation) needed to be carefully
controlled, to avoid the formation of undesired crystals at the grains’ edges, compro-
mising strip quality.
Significant research was carried out at MetSim Inc, the MMPC’s off-campus labora-
tory, on the experimental and mathematical modeling of a novel 5000 series aluminum
alloy, Al-Mg-Sc-Zr alloy strip [5]. This is an alloy that offers good corrosion resistance,
high ductility, and good weldability. For the experimental part, the versatile HSBC
simulator was again used. It included a flat, 100 mm long, copper slab substrate, with
large square copper inserts located along its length, to allow for instantaneous heat flux
evaluations for a variety of surface roughness modifications to be made. These heat flux
curves were made possible by using customized thermocouples with ~10 ms response
times.
The set-up for the HSBC simulator is shown in Fig. 2. Additionally, a mathematical
model was developed to validate the HSBC simulator results. Details of the mathemati-
cal model are discussed in the next section. From this work, it was concluded that
accurate dimensional predictions can be done, based on material properties and pro-
cessing parameters. Also, owing to the non-wetting solidification of aluminum alloys
on the copper substrate, rounded edges along with the length of the strip allowed one to
cast without the use of side dams. The air entrained between the moving surface and the
molten metal was determined to contribute significantly to the thermal resistance of the
system, and to the solidification process. Additionally, a uniform heat transfer was
observed, and higher initial heat fluxes (~10 to 20 MW/m2) were determined, as com-
pared with more typical values measured for conventional casting (~1 to 3 MW/m2).
These higher heat transfer rates and associated high cooling rates improved the micro-
structure, by generating smaller grains. The high heat fluxes of the HSBC process, and
associated high cooling rates, experimental data presented in Fig. 3, potentially offer
many advantages to the as-cast alloys strip’s qualities, in terms of as-cast microstruc-
ture and mechanical properties and provide for a better process, and provided dimen-
sional equivalence is obtained versus DC-cast strip material [6].
The mechanical properties of the alloy were evaluated: a detailed comparison is
made between the properties of the strips obtained on the HSBC simulator versus those
for an ingot-cast product after heat treatment, for a Sc and Zr-containing aluminum
alloy, elements that promote precipitate hardening after heat treatment [6]. The chemical
Fig. 3. Transient initial interfacial heat fluxes from the melt to the substrate, as predicted by the model
(along the centerline and 35 mm displaced sideways from the centerline), versus experimental data.
Substrate speed during casting was 0.8 m/s, and experimental data carry 10 pct error bars [5].
Fig. 4. Optical microscopy images of an ingot-cast sample (left) and an HSBC-cast sample [6].
composition of both products differed slightly in the content of Sc (0.39 and 0.60%wt
for the strip and ingot products, respectively) and Zr (0.05 and 0.10%wt for the strip
and ingot products, respectively), but with practically same content of Mg, the main
alloying element. It was found that the high cooling rates of the HSBC process pro-
moted finer grain sizes (~17 μm) versus an equivalent ingot-cast product (~50 μm) and
that the cast strip product required less heat treatment time (45 minutes versus 24 hours)
to obtain the equivalent mechanical properties. Since the rapid solidification promoted
a higher Sc saturation in the matrix, less bulk Sc content was necessary to achieve the
same mechanical properties. This also promoted higher resistance to over-aging. An
optical microscopy comparison between both samples is presented in Fig. 4, where the
HSBC sample presented less polygonal precipitates which are detrimental to the mate-
rial properties.
The extensive research demonstrated that the new Al-Mg-Sc-Zr Al-light alloy can be
successfully processed using HSBC technology [5, 6]. Various metal delivery systems
have been researched with the HSBC simulator, as well as at the pilot-scale level.
Among the first metal delivery systems, it researched for A356 aluminum alloy was the
vertical, or the single-impingement system (see Fig. 5a). The pilot-scale HSBC system
was operated at the MMPC since 1999, used a 600 lb steel capacity induction furnace,
with the melt being displaced upward into the metal delivery system, using a downward
motion of a refractory piston to displace the molten metal. The metal is poured onto a
2.6-m-long, specifically textured steel, or copper, belt, and intensively water cooled.
The metal subsequently solidifies into an as-cast thin strip, 3–10 mm thick, according
to the mass flow rate of liquid metal onto the belt. The pilot-scale caster system was
used for extensive experimental and mathematical modeling research for casting the
aluminum alloy AA6111, using two types of metal-feeding systems, inclined or double-
impingement metal-feeding system, and the vertical or single-impingement feeding
system [2].
For double-impingement system, the stream of falling molten metal collides twice,
first with the inclined refractory to lose kinetic energy and promote a more stable sur-
face, and second with the moving water-cooled substrate. A depiction of the two metal-
feeding systems is shown in Fig. 5.The use of an inclined feeding system was found to
present some intricacies related to the double collision encountered by the falling metal
stream, namely oscillations in the first impingement zone (shown in Fig. 6). These
oscillations can promote air entrainment plus instabilities on the meniscus generated in
the back-gap zone. It was concluded that further research was necessary to optimize the
distance between the inlet nozzle and the inclined wall, “free fall distance,” to promote
a more stable process and avoid air entrainment [8].
The HSBC technology at the pilot scale was successfully used for processing thin
strips of AA6111 alloy, 250 mm wide, using an inclined feeding system. Previously
observed flow instabilities were present when using the single-impingement metal
delivery system, and they were absent when using a double-impingement metal-feeding
system for casting thin strips of AA6111 alloy. Figures 7 presents the pilot-scale HSBC
system used to process thin strips of AA6111 aluminum alloy.
Successful HSBC pilot-scale experimental work and CFD work have been per-
formed at the MetSim off-campus laboratory to process thin strips from another light-
weight alloy, the AA2024 aluminum alloy [9]. The final strip, Fig. 8a, presented similar
Fig. 5. Main feeding systems in the HSCB process. (a) Vertical or single impingement. (b) Inclined
or double impingement.
Fig. 6. Experimental instabilities found with a double-impingement feeding system by Xu et al. [7].
Fig. 7. Pilot-scale HSBC system during casting of AA6111 aluminum alloy, (a) view toward the feed-
ing system at the beginning of casting, and (b) view toward the pinch roll during casting.
Fig. 8. AA2024 alloy strip obtained via pilot-scale HSCB system. (a) Final strip with horizontal
streaks on the edges (b) resulted microstructure with optical microscopy at 50× [9].
average roughness values on both top and bottom strip’s surfaces. A fine, uniform
equiaxed microstructure, Fig. 8b, was obtained across the thickness of as-cast thin-strip
product.
Significant research work was performed at the MetSim lab for processing thin strips
of the AA5182 aluminum alloy on the HSBC pilot-scale system, again using the double-
impingement feeding system. The resulting 5.0-mm-thick strips presented acceptable
levels of porosity and an average grain size of 63.1 μm, practically half the size of the
123 μm found for a DC (Direct Chill) cast sample. Additionally, desired secondary
phases were found within the grains, rather than at grain boundaries, due to the uni-
formly high cooling rates across the strip, for the HSBC process. This translates into an
improved strength and hardness compared to the ingot cast product. Regarding the use
of the inclined feeding system, it was commented that preferably an isokinetic (inlet
velocity = belt velocity) is desired to reduce the surface disturbances during casting.
A sample of the strip and its corresponding microstructure are presented in Fig. 9
[10]. More recent HSBC pilot-scale experimental and CFD work was performed to
compare the thin-strip product quality when casting the AA6111 alloy (used for the
Ford F150 truck), using a single- and double-impingement feeding systems, with no
side dams. The resulting strips again presented similar roughness values on both sur-
faces, without any surface indentations on the bottom surface. The strip microstructure
presented an average grain size of 85 μm with intermetallic phases distributed at the
grain boundaries. Regarding the double-impingement system, it was concluded that it
moderates the impact of the incoming melt stream and decreases the velocity of the
stream when impinging with the moving belt, generating less surface perturbances and
no backflow through the back gap is generated. Figure 10a presents the final surface
quality of the AA6111 strips obtained, while (b) presents the respective micrograph
showing fine-equiaxed grains, but some porosities (dark spots) were also detected, sim-
ilarly to a DC-cast product [11, 12, 16].
The most recent work on HSBC process development relates to the extensive math-
ematical modeling of the back-meniscus stability during HCBC casting, using a single-
impingement feeding system, for processing a AA2024 alloy. The CFD work focused
on analyzing the effect of the belt speed, clearance, and contact angle, on the backflow
and meniscus behavior [13, 14]. Back-meniscus stability is of paramount importance to
obtain a good surface strip quality, as presented in Fig. 11. Optimal parameter combina-
tions were predicted to promote a stable meniscus and finally a stable process, avoiding
product defects. The importance of the contact angle and its impact on the process sta-
bility was additionally studied. It was concluded that a non-wetting contact angle (>90°)
is needed to avoid backflow and to promote a stable meniscus. Therefore, surface
Fig. 9. AA5182 alloy strip obtained via pilot-scale HSBC system. (a) Final obtained strip (b) resulted
microstructure with optical microscopy at 50× [10].
Fig. 10. AA611 alloy strip obtained via pilot-scale HSBC process [11]. (a) Final cast strip surface, (b)
resulting microstructure with optical microscopy at 50×.
Fig. 11. Influence of the back-meniscus behavior on the strip surface quality [13].
modification of the moving belt, or of interfacial gas composition, will be necessary for
contact angle modifications in the processing of some metals, using HSBC.
A mathematical study of the back-meniscus stability with a double-impingement
system for a AA2024 will be published soon [15], where the optimal process parame-
ters to promote meniscus and process stability have been defined. After comparison
with previous results for a double-impingement system, it was found that the “free fall
distance,” played an essential role in precluding oscillations and air entrainment [8].
The “free fall distance” was changed from 6 [7, 8, 11, 12] to 3 mm, as depicted in
Fig. 12, resulting in enhanced stability on the falling stream and no air entrainment,
according to calculated predictions.
Fig. 12. Comparison of double-impingement feeding system’s “free fall distance” and its correspond-
ing calculated melt volume fraction contours.
Table 1. Chemical composition (%wt.) and dimensions of light metals alloys successfully cast with
the HSBC process, at pilot-scale and/or simulator levels.
Thickness Wide
Al Cu Mg Mn Si Zn Sc Zr Ca (mm) (mm)
Ca-Mg-Al 20 – 20 – – – – – 60 0.8 70
Ca-Mg- 10 – 15 – – 15 – – 60 0.9–1.1 70
Al-Zn
Ca-Mg-Al- 10 5 15 – – 15 – – 55 0.5–2.0 70
Zn-Cu
Al-Mg- Balance – 3.93 – – – 0.39 0.5 – 5.0–7.0 70,100
Sc-Zr
AA5182 Balance – 4.0– 0.2– <0.2 – – – – 5.0 200,
5.0 0.5 250
AA6111 Balance 0.790 0.879 0.287 0.903 0.0007 – – – 6.0 250
AA2024 Balance 3.8– 1.2– 0.3– <0.5 – – – – 8.3 183
4.9 1.8 0.9
3 Conclusions
This paper summarizes the comprehensive research and development work undertaken
at the McGill Metals Processing Centre and later at MetSim, its off-campus laboratory,
aimed at directly casting thin strips via the HSBC process. It has presented some of the
process intricacies of a simpler metal-processing operation, versus DC or CCC casting,
for high-quality cast strips to be obtained more directly from the molten metal.
Table 1 presents a summary of the light metal alloys successfully cast on the HSBC
pilot-scale caster, to the present day. These data prove the versatility and great potential
of the HSBC process to produce high-quality lightweight metals thin strips, for wide
industrial applications.
Several light metal alloys have been produced with the HSBC pilot scale, proving
their advantages compared to DC-cast products. Important equipment development and
process parameter improvements have been done to increase the HSBC productivity,
and its product quality. It is similar, for ferrous and copper-based alloys. These findings
prove process viability and show its readiness for future implementation. Given the
world’s urgent needs for reducing metal-processing costs, and reduced contaminant
emissions, industrial large-scale HSBC systems can be envisaged.
Processing Centre, for this work. The authors would also like to acknowledge the support in software
licensing received from ANSYS Inc. to facilitate this research. Daniel R. Gonzalez-Morales would also
like to thank the Mexican National Council of Science and Technology (CONACyT) for their financial
support provided during his graduate studies (CVU:1036046).
References
1. Ge S, Isac M, Guthrie RIL. Progress of strip casting technology for steel; historical developments.
ISIJ Int. 2012;52(12):2109–22. https://doi.org/10.2355/isijinternational.52.2109.
2. Ge S, Isac M, Guthrie RIL. Progress in strip casting technologies for steel; technical developments.
ISIJ Int. 2013;53(5):729–42. https://doi.org/10.2355/isijinternational.53.729.
3. Guthrie RIL, Isac M. Horizontal single belt casting of aluminum and steel. Steel Res Int.
2014;85(8):12091–1302. https://doi.org/10.1002/srin.201300105.
4. Guthrie RIL, Isac M, Li D, Calzado L. Horizontal Single Belt Casting (HSBC) of Ca-based bulk
metallic glass (BMG) strips. Proceedings of the extraction and processing division symposium,
TMS. 2014; https://doi.org/10.1007/978-3-319-48234-7_8.
5. Ge S, Isac M, Guthrie RIL. The computational fluid dynamic (CFD) modeling of the hori-
zontal single belt casting (HSBC) processing of Al-Mg-Sc-Zr alloy strips. Metall Mater. Tran.
B. 2015;47B:2264–77. https://doi.org/10.1007/s11663-015-0375-1.
6. Ge S, Celikin M, Isac M, Guthrie RIL. Analysis and evaluation of novel Al-Mg-Sc-Zr aerospace
alloy strip produced using the Horizontal Single Belt Casting (HSBC) process. Metall Mater Tran
B. 2015;46B:1035–42. https://doi.org/10.1007/s11663-014-0284-8.
7. Xu M, Isac M, Guthrie RIL. A numerical simulation of transport phenomena during the horizontal
single belt casting process using an inclined feeding system. Metall Mater Tran B. 2018;49:1003–13.
https://doi.org/10.1007/s11663-018-1237-4.
8. Xu M, Isac M, Guthrie RIL. Flow instabilities in the Horizontal Single Belt Casting Process with
an inclined feeding system. ISIJ Int. 2018;59(1):69.75. https://doi.org/10.2355/isijinternational.
ISIJINT-2018-514.
9. Lee J, Isac M, Guthrie R. Computational study and microstructural analysis of AA2024 strips
processed via the horizontal single belt casting technique, COM, MS&T, 2018.
10. Hsin CH. Production of automotive aluminum alloys AA5182 thin sheet material using the
Horizontal Single Belt Casting (HSBC) process. MSc thesis, McGill University, Montreal,
Canada, 2019.
11. Niaz U, Isac M, Guthrie RIL. Numerical modeling of transport phenomena in the horizontal single
belt casting (HSBC) process for the production of AA6111 aluminum alloy strip. PRO. 2020;8:529.
https://doi.org/10.3390/pr8050529.
12. Niaz U, Isac M, Guthrie RIL. The casting of AA611 aluminum alloy strips using single and double
impingement types of liquid metal feeding system. J Miner Sci Mater. 2020;1:1018.
13. Morales DG, Riviere C, Isac MM, Guthrie RIL. Predicted back-meniscus stability for the horizon-
tal single belt casting (HSBC) process. Paper presented at the 60th COM Annual Meeting, virtual,
17–19 August 2021.
14. Gonzalez-Morales DR, Isac MM, Guthrie RIL. Predicted Back-Meniscus stability for Horizontal
Single Belt Casting (HSBC) using a single impingement feeding system. Light Metals 2022
TMS. 2022; https://doi.org/10.1007/978-3-030-92529-1_97.
15. Gonzalez-Morales DR, Isac MM, Guthrie RIL. HSBC back-meniscus stability for a double
impingement liquid metal feeding system COM 2022. In Press; 2022.
16. Niaz U. Numerical modeling & experimental casting of AA6111 aluminum alloy and Advanced
High Strength Steels (AHSS) produced through Horizontal Single Belt Casting (HSBC) process.
PhD thesis, McGill University, Montreal, Canada, 2020.
Scanning Electron Microscopy Analyses of Horizontal
Single Belt Casting Thin Strips of AA6111 Alloy
C. Riviere1, D. Lu1, M. M. Isac1 (*), and R. I. L. Guthrie1
1
McGill Metals Processing Center, McGill University, Montreal, QC, Canada
Mihaiela.isac@mcgill.ca
Abstract. Horizontal Single Belt Casting (HSBC) is a promising Near-Net-Shape

Casting (NNSC) technology for the future production of thin metal strips, given its out-
standing economical and environmental benefits, compared with present day commercial
Direct Chill (DC) technology. In this research, thin strips of AA6111 aluminum alloy
were obtained via HSBC, wherein the molten metal was solidified on a moving, water-
cooled steel belt. Pilot-scale HSBC experiments were used to validate the experimental
results of thin strips processed via the HSBC simulator. Scanning Electron Microscopy
(SEM) was used to study the cast microstructures of the thin strip products. Electron
Back-Scattered Diffraction (EBSD) was used to measure the average grain size, identify
the chemical elements, and determine the composition of intermetallic compounds
(IMCs) inside the grains and at the grain boundaries. The microstructural and SEM analy-
ses confirmed the capability of HSBC in producing high-quality, as-cast microstructures,
suitable for minimal rolling and annealing operations to produce AA6111 aluminum
strips.
Keywords: Aluminum Thin Strips · Grain Size · Horizontal Single Belt Casting ·
Intermetallic Compounds · Microstructure · Scanning Electron Microscopy
1 Introduction
Near-Net-Shape Casting (NNSC) technology has been gaining increasing popularity in

the metal processing industry since it produces thin metallic strips and plates with
dimensions close to the final products, thereby requiring considerably less processing
steps compared to conventional processes [1, 2]. Horizontal Single Belt Casting
(HSBC), as a novel NNSC process, offers exciting possibilities and potential to further
reduce operating costs, energy consumption, and greenhouse gas emissions, in produc-
ing thin metal sheets. In the context of environmental protection, energy sustainability,
and climate change, the role of HSBC deserves more emphasis and study since the
HSBC process does not require multi-hot reduction passes. Thus, the repeated re-
heating and hot rolling steps in hot-strip production for aluminum are eliminated by
integrating casting and rolling into one single process [3]. The initial fast-cooling rates
of up to 500 °C/s for the HSBC process lead to improved homogeneity of microstruc-
ture, less segregation, and finer grain sizes within the cast products [4].
https://doi.org/10.1007/978-3-031-17425-4_57
438 C. Riviere et al.
Figure 1 shows the configuration of a pilot-scale HSBC system operating at the

MetSim off-campus laboratory of the McGill Metal Processing Centre (MMPC). It
consists of an upstream delivery system, a midstream belt caster, and downstream roll-
ing and coiling. During the casting process, the molten metal empties from a tundish
through a refractory slot nozzle, onto a horizontal, single, moving belt. The molten
metal flow rate is adjustable by controlling the metal head inside the metal delivery
tundish [5]. The moving belt is water-cooled to maintain it close to room temperature
and thereby act as a cooling substrate/mold for the solidification of molten metals.
Finally, the downstream equipment setup enables the immediate rolling and coiling
process of the final product, ready for batch annealing.
The world’s first commercial-scale Single Belt Casting plant was commercialized by
Salzgitter Gmbh, in 2012, at their steel plant in Peine, Germany, under the name “Belt
Cast Technology.” It has since been successfully producing various grades of steel and
selling them to Salzgitter’s customers [3, 6]. More recently, the HSBC has also been
used commercially for producing thin strips of Cu–Sn–Ni alloys, by Materion Brush,
USA, following earlier successful demonstrations in Canada, at the pilot-scale, by
MetSim Inc./MMPC. The capability of HSBC in producing high-quality, non-ferrous,
and ferrous alloy strips has also been confirmed by many other studies [7–9]. Aluminum
alloy 6111 (AA6611), which is famous for its unique combination of formability and
paint bake strengthening [10], has been used extensively across many industries.
Keeping in view AA6111’s inherent advantages and its potential in various applica-
tions, the objective of the present research was to produce AA6111 thin strips using the
researcher’s HSBC simulator, and then study the microstructure of the product using
SEM. The researchers also validated the HSBC simulator results with pilot-scale exper-
imental work.
Fig. 1. Schematic of the pilot-scale HSBC system operated at the MetSim/MMPC laboratory [1].
Scanning Electron Microscopy Analyses of Horizontal Single Belt Casting Thin Strips… 439
2 Materials and Manufacturing Procedure
An induction furnace was used to prepare and melt 300 g of AA6111 alloy in a ceramic
crucible. The chemical composition, in weight percent, of the AA6111 used, is pro-
vided in Table 1.
Argon gas was injected through a shroud to displace oxygen and mitigate surface
oxidation of the melt. The AA6111 cast strips were produced via the HSBC process
using the simulator described in Fig. 2.
The AA6111 alloy melt was poured into a small, steel-fabricated, tundish, with a
stopping block that temporarily pauses the flow of the melt from it. The tundish is lined
internally with an alumina-based refractory called METAL KAST 90. When the metal
was seen to reach the desired level, it was allowed to settle for a couple of minutes, to
allow for the dissipation of any turbulence. The refractory block under the tundish acted
as an exchangeable slot nozzle, which allowed the liquid metal to exit onto the belt. The
slot-nozzle had a dimension of 75 mm in slot length and 3 mm in slot width. A movable
copper substrate was fixed to a horizontal shaft. This substrate measures 110 mm wide,
500 mm long and has a thickness of 12.7 mm. The copper substrate acts like a chill and
can be propelled by a spring mechanism that allows for two set speeds; a high speed of
0.896 m/s and a low speed of 0.66 m/s. The release of the spring automatically opens
the nozzle, allowing the liquid metal alloy to deposit and solidify onto the copper sub-
strate, travelling below.
Two K-type grounded thermocouples were placed inside the pure copper substrate
for the measurement of temperature evolution within the strip during casting. They
were imbedded at depths of 3.7 and 4.1 mm from the top surface of the substrate,
respectively. Temperature data were produced, filtered, and measured at a rate of 16 Hz
using DAISY Lab data acquisition software and an InstruNeti555 system. An inverse
heat conduction problem approach (IHCP) was used to derive the interfacial heat fluxes
between the melt and the substrate. The IHCP method uses the internal temperatures
acquired by the embedded thermocouples, to establish a boundary condition, and com-
pute instantaneous surface heat fluxes, and heat transfer coefficients.
The AA6111 strip used for microscopic analyses were cast from 730 °C (70 °C
superheat) and 0.66 m/s belt speed. The maximum heat flux was measured to be
1.02 MW m−2 during the casting process. An average thickness of 2.63 mm was obtained
for the cast strip.
A 2 cm by 2 cm square sample was cut off from the strip as illustrated in Fig. 3 for
further analyses. The side along the direction of casting, called the longitudinal side,
was prepared for microscopic characterization. Grinding, polishing, and electro-etching
were performed to reveal the product’s microstructure. The sample was prepared using
silicon carbide papers from 600 grits to 1200 grits, followed by polishing with 6, 3, and
1 μm of diamond alcohol suspensions, sequentially. The reagent used for electro-
polishing was a solution of 58% sulfuric acid, 22% glycerin, and 20% water, according
Table 1. Chemical composition of the AA6111 alloy used in the experiment.

Element Al Cu Fe Mg Mn Si Ti Zn
Weight percent (wt.%) Bal. 0.790 0.177 0.879 0.287 0.903 0.004 0.007
Fig. 2. Side view schematic of the HSBC simulator used for producing AA6111 cast strips.
Fig. 3. Sample cut from the cast strip.
to the Buehler guidebook [11]. The microstructure of the sample surface was analyzed
using SEM (Hitachi S-3400 N).
3 Microscopic Characterization
3.1 Grain Size
EBSD is a SEM-based technique that gives crystallographic information about the

microstructure of a sample. It can be used as an analytical tool for grain size measure-
ments. The measuring technique used was the mean linear intercept method. The aver-
age grain size was determined by laying several randomly positioned test lines on the
image and counting the number of times that grain boundaries are intercepted.
Mathematically, the average grain size is calculated according to
Average Grain Size = number of intercepts / length of test line (1)
The average grain size of the HSBC as-cast sample was 72.1 μm. As such, the grain
size obtained was much smaller than 231 μm for a DC cast AA6111 alloy [12].
3.2 Mapping of Distribution of Chemical Elements
Elemental mapping is a common method used to characterize the element distribution

across a selected area on an exposed sample surface. Specific element compositions
were determined using EDS. It detects the X-rays emitted from the sample which is
bombarded by the electron beam of SEM. The energy of these X-rays is unique to the
elements present in the sample, and thus can be used for elemental mapping of the
sample.
Figure 4 shows the SEM image and its corresponding element mappings taken for a
specific area. Intermetallics can be observed as the elongated and round compounds at
the grain boundary and inside the grains, respectively. It was found that the grain
Fig. 4. (a) SEM image of a selected area, and its corresponding elemental mapping for the following
elements: (b) manganese (c) silicon, (d) copper, (e) iron, and (f) magnesium.
b oundary intermetallics are largely irregular in shape, whereas the intermetallics found
inside grains were more rounded and spheroidal in shape.
Silicon and Cu are found distributed at both the grain boundary and inside the grains
where the intermetallic compounds containing them are located. But Cu is richer within
the intra-granular intermetallics rather than at the grain boundaries. Fe and Mg are more
distributed in intermetallics at the grain boundaries. Traces of Mn are also found at the
grain boundaries, but in smaller proportion compared with Fe and Mg. The haze-like
distributions of Mn, Cu, and Fe mapping images are identified, as these elements are
also dissolved in the solid solution, that form the matrix of the alloy.
3.3 Intermetallic Phase Analysis
As mentioned, two types of intermetallic phases were found based on their location
relative to the grain structure analyzed. The first type of intermetallic was found at the
grain boundaries and was of elongated, irregular shape. The second type of intermetal-
lic was small and spheroidal in shape and was contained within the grain
(intra-granular).
Figure 5 shows both the band constant and EBSD layered images taken from the
sample surface on the longitudinal side of the cast strip. Intermetallic phases can be
easily observed inside the grains.
Figure 6 shows the diffraction spectra associated with the above-mentioned interme-
tallic compounds. For instance, spectrums 6 and 7 are irregular-shaped intermetallic
phases at the grain boundaries, while spectrums 9 and 10 are spherical intermetallic
phases inside the grains. The spectra were used to determine the chemical composition
and further deduce the stoichiometry of the intermetallic phases identified using ele-
mental line scans in the SEM. The resulting element compositions at these spectra are
summarized in Table 2.
The comparison of element compositions shown in Table 2 provides a better insight
for making a correlation between the intermetallic composition and their location. The
grain boundary intermetallics are generally richer in Si and Mg, but are leaner in Cu,
Fig. 5. Intermetallic characterization using SEM, (a) band contrast image, (b) EBSD layered image.
Fig. 6. Spectra showing different sample points for which line scanning of intermetallic phases is
conducted.
Table 2. Element compositions for intermetallics at grain boundaries and within grains.
Intra-granular At grain Intra-granular in the At grain boundaries in the
Element (wt.%) boundaries (wt.%) reference (wt.%) reference (wt.%)
Al 67–70 63–77 65–72 78–79
Mg 0.6–3.7 0.4–11 2.2–11 0.4–4
Si 4–12 8–15 5–16 2–8
Mn 0.09–3 1–3 0.07–0.3 0–0.86
Fe 0.1–6.2 3–6 0.1–0.5 3–8
Cu 7–23 1–9 7–20 1–2
compared to the intra-granular intermetallics. The solute elements tend to preferentially

precipitate at the grain boundaries than within the grains. This preferential tendency is
seen to be the most pronounced for magnesium, and the ratio of magnesium to alumi-
num is markedly higher in intermetallics at the grain boundaries.
Notably, the elemental analysis for both the grain boundary and intra-granular inter-
metallic compounds was compared with prior research [13], in which the work was
performed on the same alloy which had been subjected to the same thermal cycling at
cooling rates similar to those of the HSBC process. Comparing the observations of the
present study with the research reported in literature [13], the range of atomic percent-
ages of various elements present in the intra-granular intermetallic compounds shows a
good positive correlation. A notable difference is the upper limit of iron present in the
intermetallic found in this study. The intra-granular intermetallic compounds found in
this study were more iron-rich than in the prior research. A similar comparison for
intermetallic compounds found at the grain boundaries, again, for the most part, show
a good one-to-one correlation between the proportions of the various elements reported
Table 3. Ratio of aluminum to various elements in the intermetallic compounds.

Location Al/Mg ratio Al/Si ratio Al/Fe ratio Al/Cu ratio
Grain Boundary 5 4–10 21–49 19
Intra-granular 17 6–21 22 7–23
Table 4. Proposed empirical formulae for the intermetallics.

Present study Previous study
Element Proposed Proposed
location present stoichiometry location stoichiometry
Grain Boundary 1 Al, Mg, Si, Al5MgSi4Fe21 Grain Al5Mg0.4Si0.4Cu1.5
(spectrum 6) Fe Boundary 1
Grain Boundary2 Al, Si, Cu Al10SiCu9 Grain Al10Si2.5Cu7.5
(spectrum 7) Boundary 2
Intra-granular1 Al, Si, Cu Al21SiCu7 Intra-granular Al21Si1.6Cu6.5Mg1.6
(spectrum 9) 1
Intra-granular2 Al, Si, Fe, Al17Si6Fe22Cu23 Intra-granular Al17Si8.5Cu33
(spectrum 10) Cu 2
by the two bodies of work. Some differences are the relatively lower quantities of man-
ganese and the richer amounts of silicon reported in this study.
Table 3 gives the proportion of aluminum to other elements in the intermetallic com-
pounds, for both intra-granular intermetallics as well as those located at the grain
boundaries.
Utilizing the atomic ratios, the appropriate empirical formula can be suggested for
the intermetallics. This is done in Table 4 for both the grain boundary and intra-granular
intermetallics. Furthermore, the proposed empirical formulae were compared with the
ones proposed in the previous study [13]. To make the comparison starker, the base of
the aluminum was taken as the same, this allows a better comparison of the elemental
atomic ratios that are proposed by the present study versus that reported in the refer-
ence [13].
4 Conclusions
Some of the conclusions of this study are enumerated below:

1. HSBC resulted in an approximate grain size of 72 μm versus 231 μm obtained by the
DC chill casting process, extrapolated for the same operating temperature.
2. Fine intermetallic phases were found to be present in the portions of the strip ana-
lyzed. Two types of intermetallic phases were found at the grain boundaries and
inside the grains.
3. The size, shape, and composition of these IMCs agreed well with the previously
reported literature, and specifically literature which used similar thermal cycling
conditions.
4. Grain boundary IMCs were, in general, found to be more solute-rich compared to

their intra-granular counterparts.
Processing Centre, for this work.
References
1. Ge S, Isac MM, Guthrie RL. The computational fluid dynamic (CFD) modeling of the hori-
zontal single belt casting (HSBC) processing of Al-Mg-Sc-Zr alloy strips. Metall Mater Trans
B. 2015;46(5):2264–77. https://doi.org/10.1007/s11663-015-0375-1.
2. Ge S, Celikin M, Isac MM, Guthrie RL. Analysis and evaluation of novel Al-Mg-Sc-Zr aerospace
alloy strip produced using the horizontal single belt casting (HSBC) process. Metall Mater Trans
B. 2015;46:1035–43. https://doi.org/10.1007/s11663-014-0284-8.
3. Ge S, Isac MM, Guthrie RL. Review of the history and recent development of strip casting technol-
ogy. In: Roderick Guthrie Honorary Symposium on Process Metallurgy. Montreal; 2011.
4. Niaz U, RlL G, Isac MM. A numerical simulation of transport phenomenon for the casting of
AA6111 plate using horizontal single belt pilot caster. In: The materials science and technology
2018. Columbus; 2018.
5. Niaz U, Isac MM, Guthrie RL. Numerical modeling of transport phenomena in the horizontal
single belt casting (HSBC) process for the production of AA6111 aluminum alloy strip. Processes.
2020;8(5):529. https://doi.org/10.3390/pr8050529.
6. Guthrie RL, Isac MM. Conventional and near net shape casting options. Ironmak Steelmak.
2016;43(9):650–8. https://doi.org/10.1080/03019233.2016.1216510.
7. Guthrie RL, Isac MM, Li D, Calzado L. Horizontal single belt casting (HSBC) of Ca-based bulk
metallic glass (BMG) strips. In: Celebrating the megascale: proceedings ofthe extraction and
processing division symposium on pyrometallurgy in honor of David G.c. Robertson. Cham/San
Diego: Springer; 2014. p. 105–17. https://doi.org/10.1007/978-3-319-48234-7_8.
8. Guthrie RL, Isac MM. Horizontal single belt casting of aluminum and steel. Steel Res Int.
2014;85(8):1291–302. https://doi.org/10.1002/srin.201300105.
9. Lee J. Horizontal single belt casting process of AA2024. A numerical and physical modelling
study. Graduate Thesis. Montreal: McGill University; 2019.
10. Gupta A, Lloyd D, Bull M. The properties and characteristics of two new aluminum automotive
closure panel materials. SAE Trans. 1996;105(5):180–5.
11. Buehler. Operation and Maintenance Instructions Electromet 4 Polisher/Etcher. 1998.
12. Al-Helal KW, Lazaro-Nebreda J, Patel JB, Scamans GM, Fan Z. Melt conditioned direct chill
(MC-DC) casting of AA-6111 aluminium alloy formulated from incinerator bottom ash (IBA).
Recycling. 2019;4(3):37. https://doi.org/10.3390/recycling4030037.
13. Shabestari S, Li D, Isac MM, Guthrie RL. The role of solidification conditions on the microstruc-
tural features of AA6111 aluminum strips cast on a horizontal single belt caster. Int J Mater Res.
2008;99(12):1384–92. https://doi.org/10.3139/146.101780.
The Effect of Hydrogen Embrittlement on Aluminum
Alloys Used for Hydrogen-Fueled Internal
Combustion Engine
Rashiga Walallawita1 (*) , Joshua Stroh1 , and Dimitry Sediako1

1
High Performance Powertrain Materials Laboratory, School of Engineering, The University of
British Columbia–Okanagan, Kelowna, BC, Canada
rashiga.walallawita@ubc.ca
Abstract. Due to hydrogen’s unique characteristics, such as high energy density, low
mass, and the capability of using it as a zero/low emission fuel, it has become a viable
contender towards fulfilling current and future emission goals. With the recent techno-
logical development, there has been an increasing interest in hydrogen internal combus-
tion engines (HICE). The increased hydrogen concentration and the high working
temperatures inside the engine’s combustion chamber will create an environment that can
increase the rate of hydrogen diffusion into the engine components, leading to hydrogen
embrittlement (HE). Therefore, the authors have focused on studying the feasibility of
using current cast powertrain aluminum alloys for HICE. From the comprehensive litera-
ture survey, it has been found that there is limited published literature on HE in cast alu-
minum alloys for high pressure and high-temperature charging conditions. The majority
of the literature was focused on low-temperature high-pressure charging of wrought alu-
minum alloys. Keeping this in mind, an economically attainable 2-step experimental pro-
cedure has been proposed to identify if cast powertrain aluminum alloys can withstand
the cycling load, high pressure, and high-temperature conditions of a HICE.
Keywords: Aluminum alloys · Hydrogen embrittlement · Hydrogen internal

combustion engine
1 Introduction
To meet future emission goals, hydrogen plays a major role as an energy carrier. This
has been evident from the role that hydrogen has in reaching the target of global net-
zero emissions by the year 2050. As an example, the Canadian government has already
developed a “hydrogen strategy for Canada” to decarbonize the economic sectors.
The transportation sector has the foremost contribution to greenhouse emissions. By
using hydrogen instead of fossil fuels, it is possible to reach zero/low emissions. Apart
from fuel cell electric vehicles (FCEV), hydrogen internal combustion engine (HICE)
vehicles use hydrogen as their primary energy source. In terms of operation, the internal
combustion engine (ICE) and HICE have the same layout apart from the usage of spe-
cialized hydrogen injectors and hydrogen storage tanks.
https://doi.org/10.1007/978-3-031-17425-4_58
448 R. Walallawita et al.
For ICEs to meet the current and near-future emission goals, light-weighting is a key
factor, and aluminum (Al) alloys are used to replace cast iron and steel components.
One such instance is the usage of aluminum alloys (A319, A356, etc.) for engine blocks.
With the introduction of concentrated hydrogen (relative to cylinder pressure of
~20 MPa) [1], high temperature (200–300 °C) [1], and the moisture from the combus-
tion, hydrogen diffusion will increase, which will result in hydrogen embrittlement
(HE) near the cylinder walls and cylinder head surfaces. The present article reviews the
available literature on HE of Al alloys. Due to the nature of the hydrogen-induced dam-
age expected in the HICE, the study is focused on gaseous hydrogen charging of sample
materials.
2 The Available Literature on HE in Al Alloys
Before reviewing HE in aluminum alloys, it is important to study the HE in pure Al. In

most gas phase environments, Al forms a stable oxide layer. Due to its slow reaction
rate, the depth of this oxide layer will be limited. Hence, when Al is exposed to dry
hydrogen it first interacts with the oxide layer. Due to the high energy barrier in chemi-
sorption, limited hydrogen is chemically absorbed into the oxide layer. However, in the
presence of water vapor and at high temperatures, the solubility of hydrogen tends to
increase which will result in HE [2].
Campbell [3] has stated that Al and Al alloys do not show any sort of embrittlement
at ambient temperature. Tests were conducted for Al alloys 1100-O, 6061-T6, and
7075-T73 using notched and unnotched tensile specimens at gas (He vs H2) pressure of
34.5 and 69.0 MPa at room temperature. No significant change in mechanical proper-
ties was observed.
San Marchi and Somerday [4] have conducted fatigue crack growth and fracture
toughness tests on Al 7475-T7351 in ambient conditions and with gaseous hydrogen at
a pressure of 103 MPa at room temperature. The researchers have concluded that the
fracture toughness and fatigue crack growth rate do not reduce significantly with hydro-
gen exposure. Similar results were obtained for Al 6061 alloy with 45 MPa of gas pres-
sure at room temperature by Ohmiya and Fuji [5].
Sprowls and Brown [6] have used pre-cracked specimens of high-strength Al alloys
were exposed to dry hydrogen (pressure of 0.1 MPa) for 47 days at room temperature
to measure the crack growth rate. Within the first 47 days, the researchers have not
observed any crack growth. Next, the same specimen was exposed to wet hydrogen and
after changing the environment from dry to wet there was a steady increase in crack
growth rate. This implicates how wet hydrogen can contribute to HE in Al alloys.
The majority of the work that is already available for HE is on wrought Al alloys
which are used in fuel cell storage and the aircraft/aerospace industry. Therefore, gas
charging is done at room temperature to suit real-world charging conditions. Even
though there has been no reported significant HE at these conditions, the authors believe
the increased diffusivity at elevated temperatures may show effects of HE as compared
to room temperature. Moreover, when testing materials with casting defects there may
be a higher tendency of hydrogen trapping which may contribute to increased HE.
The Effect of Hydrogen Embrittlement on Aluminum Alloys Used for Hydrogen-Fueled… 449
In this study, HE in automotive Al alloys is studied in two steps. The first step is to
identify if the embrittlement process is reversible or irreversible (explained in the sub-
sequent paragraph) for the given charging conditions. The second step is to identify the
degree of damage done by the identified type of embrittlement.
The importance of identifying the type of HE damage is twofold, if the embrittle-
ment is irreversible this means the damage is permanent. It is important to measure the
amount of permanent damage to see how it will affect the life of the HICE components.
Furthermore, if the damage is reversible this means the damage will be recovered when
the hydrogen is removed from the environment, e.g., between the engine operation
cycles. However, it is important to characterize how quickly the mechanical properties
will regain and the amount of embrittlement for a given amount of time.
Thermal pre-charging will be used to identify the type of damage and to quantify the
degree of irreversible damage. Similar to the conditions in a HICE, the specimen pre-
charging conditions are set at 20 MPa hydrogen pressure and 200 °C. The pre-charging
duration is 200 h, following which the specimen is kept in ambient condition for differ-
ent periods of time. Tensile tests are to be conducted at ambient conditions for each
storage period; the results obtained are to be compared with the tensile results obtained
from the virgin specimens.
The reversible damage is assessed in in situ slow strain rate tensile [7] testing. The
typical setup involves a modified pressure vessel [8]which is expensive and complex.
To bypass these shortcomings, hollow tensile specimen [9] is used in this study and,
instead of using a gaseous environment, the hollow tensile specimen has been designed
to act as the pressure vessel. Several researchers [10, 11] have used hollow tensile
specimens to conduct in situ tensile testing. One drawback of this method is the addi-
tional radial load applied due to the internal gas pressure. Therefore, to reduce the
biaxial loading the applied gas pressure should be kept to a designated maximum. In
this study, the internal pressure is set at 10 MPa at 200 °C which allows keeping the
hoop stress below 10% of the applied axial load. Furthermore, to compare the results
with a no-hydrogen reference, another set of hollow tensile specimens are to be tested
using an internal pressure of 10 MPa with nitrogen gas.
All the crack surfaces are examined using optical microscopy (OM) and scanning
electron microscopy (SEM) to identify the mechanism of embrittlement and to observe
the unique features [12] which occur due to embrittlement. Energy-dispersive spectros-
copy (EDS) is employed to indetify and assess the traces of hydrogen in the sample
materials.
4 Conclusion
Eventhough there has been extensive research conducted on Al alloys that are used in
different industries, there have been no reported studies done on cast Al alloys, espe-
cially the Al alloys that are used in the manufacture of automotive engine components.
450 R. Walallawita et al.
To add to this, the elevated temperature and pressure conditions of the gaseous environ-
ments are also not considered. It is therefore of paramount interest that the material
science and engineering community has begun addressing this lack of published knowl-
edge, by testing Al powertrain alloys at engine operating conditions while exposed to
representable hydrogen environments.
References
1. Welch A, Mumford D, Munshi S, Holbery J, Boyer B, et al. Challenges in developing hydrogen

direct injection technology for internal combustion engines. SAE Tech Pap. 2008;724. https://doi.
org/10.4271/2008-01-2379.
2. Scully JR, Young GA, Smith SW. Hydrogen embrittlement of aluminum and aluminum-based
alloys. In: Gaseous hydrogen embrittlement materials in energy technoligies. Woodhead
Publishing; 2012. p. 707–68.
3. Campbell JE. Aluminum and aluminum alloys. In: Effects of hydrogen gas on metals at ambient
temperature, vol. 1970. Defense Metals Information Center; 1970. p. 9–12.
4. San Marchi C, Somerday BP. Fatigue crack growth of structural metals for hydrogen service. In:
Pressure vessels and piping conference; 2011. p. 851–7. https://doi.org/10.1115/PVP2011-57701.
5. Ohmiya S, Fujii H. Fatigue properties of liner materials used for 35MPa-class on-board hydrogen
fuel tanks. In: Pressure vessels and piping conference; 2005. p. 499–503. https://doi.org/10.1115/
PVP2005-71735.
6. Sprowls DO, Brown RH. What every engineer should know about stress corrosion of aluminum.
Met Progr. 1962;841:79.
7. ASTM standard. ASTM G142-98 standard test method for determination of susceptibility of met-
als to embrittlement in hydrogen containing environments at high pressure, high temperature, or
both. E-rea. 2011;2016:1–8.
8. Somerday BP, Campbell JA, Lee KL, Ronevich JA, San Marchi C. Enhancing safety of hydro-
gen containment components through materials testing under in-service conditions. Int J Hydrog
Energy. 2017;42:7314–21. https://doi.org/10.1016/j.ijhydene.2016.04.189.
9. Chandler W, Walter R. Testing to determine the effect of high-pressure hydrogen environments
on the mechanical properties of metals. In: Hydrogen embrittlement testing, vol. 1974. West
Conshohocken: ASTM International; 1974. p. 178–9.
10. Boot T, Riemslag TAC, Reinton ETE, Liu P, Walters CL, et al. In-situ hollow sample setup design
for mechanical characterisation of gaseous hydrogen embrittlement of pipeline steels and welds.
Metals (Basel). 2021;11(8):1242. https://doi.org/10.3390/met11081242.
11. Ogata T. Influence of high pressure hydrogen environment on tensile and fatigue proper-
ties of stainless steels at low temperatures. AIP Conf Proc. 2012;2012:39–46. https://doi.
org/10.1063/1.4712078.
12. Lynch SP. Metallographic and fractographic techniques for characterising and understanding
hydrogen-assisted cracking of metals. In: Gaseous hydrog embrittlement mater energy technolo-
gies, the Problem its characterisation effects on particular alloy classes. Woodhead Publishing;
2012. p. 274–346.
Development of LiMCA (Liquid Metal Cleanliness
Analyzer) Since Its Invention to Date
G. D. Di Silvestro1, R. Tiwari1, M. M. Isac1 (*), and R. I. L. Guthrie1

1
Mihaiela.Isac@McGill.ca
Abstract. Non-metallic inclusions less than 50 μm remain harmful to liquid steel qual-
ity. These deleterious inclusions are difficult to be monitored and even more difficult to be
removed during steel processing. LiMCA (Liquid Metal Cleanliness Analyzer) is one
method to monitor such inclusions. Its development since its inception to date is now
reviewed. This will include the successful contributions of the McGill Metals Processing
Centre, Alcan, Bomem, and ABB in the development of LiMCA for the detection of
inclusions in aluminum, magnesium, or copper melts, as well as the contribution of
Heraeus Electro-Nite and Sumitomo Metals in the development of the (LiMCA) ESZpas
for the detection and monitoring of inclusions in liquid steel.
Keywords: Electric Sensing Zone–Particle Analyzer (ESZpas) · Inclusions ·

LiMCA · Aqueous Particle Sensor (APS III) System · Microbubbles
1 Technical Development of Liquid Metal Cleanliness Analyzer

(LiMCA and ESZpas)
In establishing methods for measuring and/or detecting inclusions in liquid metals, the
LiMCA technique is the only commercial method available for the in-situ, online detec-
tion of inclusions [1, 2]. The technique analyzes both the number concentration and size
distributions of inclusions in molten metals. Developed at McGill University in the
early 1980s, it saw rapid development in collaboration with ALCAN and BOMEM,
ABB, and Heraeus Electro-Nite over the past three decades [2]. The initial research
initiated by two inventors Dr. R. I. L. Guthrie, with his student, now Dr. D. Doutre,
related to analyzing soluble salt particles in frozen aluminum samples by employing the
Coulter Counter Technology, previously used for aqueous solutions [3]. Initial tests
performed at McGill utilized room temperature liquid gallium and a mixture of flour as
inclusions. Due to the higher conductive medium of all liquid metals, a larger DC cur-
rent was required. More specifically, whereas a DC current of 20 mA was used for an
aqueous solution, here 60 A would be required. As shown in Fig. 1, inclusions were
able to be detected as voltage peaks or signals. As a result of using liquid metal, unde-
sired signal interferences or “white noise” were subsequently corrected via the “Faraday
Cage” effect. This permitted legible generation of micro-signals within the liquid metal,
as made visible by a standard oscilloscope [1].
https://doi.org/10.1007/978-3-031-17425-4_59
452 G. D. Di Silvestro et al.
Fig. 1. Schematic of the ESZ (electric sensing zone) principle of LiMCA [1].
Ten years after the initial development work started at McGill and continued with
Alcan, the arrival of LiMCA II with its integrated, compact, and improved performance,
developed for commercial uses by ALCAN and BOMEM, opened the door for the wide
use of this technology by the metallurgical industry [4].
Following the commercial application of the LiMCA technique to aluminum, other
liquid metals such as zinc, lead solders, gallium, copper, and magnesium were also
tested, and appropriate sensors were developed. The development of an inclusion sen-
sor for liquid steel represented the ultimate materials challenge. LiMCA II has now
been successfully used in plants producing high-quality thin gauge or surface critical
applications (aluminum can body, and can end stock, foils, or anodizing or lithographic
sheet material) [4].
Following this successful evolution, LiMCA III was developed by ABB for use
under harsh industrial production conditions to provide an accurate indicator of the
metal cleanliness of aluminum alloys. The LiMCA III took advantage of a lighter mea-
suring head for more versatile and flexible positioning within the molten metal, vs the
LiMCA II analyzers. Its lower center of gravity also made it more stable and easier to
maneuver. Furthermore, the LiMCA III came with a laser level sensor that enabled
automatic raising and lowering of the measurement head to adapt to the varying levels
of molten metal during a casting [5].
Since then, ABB has developed another successful in-line sensor, called LiMCA CM
©, shown in Fig. 2, for the continuous monitoring of inclusions to ensure the quality of
aluminum in real-time, as measured by the total concentration and size distribution of
inclusions. This new system was designed in keeping with the need for it to be an auto-
mated, in-situ instrument for continuous, maintenance-free operation [6].
Identification and classification of inclusions are essential aspects of establishing
clean steel practices. As such, methods of evaluation are essential in determining popu-
lation distributions, sizes, compositions, and shapes of inclusions, for every process.
Development of LiMCA (Liquid Metal Cleanliness Analyzer) Since Its Invention to Date 453
Fig. 2. Liquid metal cleanliness analyzer for continuous monitoring of melt quality (LiMCA CM ©)
[6].
Fig. 3. Instrumentation and sensor of LiMCA for liquid steel [2].
Some 30 years later since LiMCA’s invention for moderate temperature melts,
researchers at McGill University’s MMPC, in collaboration with Heraeus Electro-Nite
and Sumitomo Metals Industries, were instrumental in developing a one-shot probe for
liquid steel. Nonetheless, the challenge remains in use, and proper design of a LiMCA
probe, with a sampling probe suitable for operating continuously in liquid metal envi-
ronments at temperatures of 1500 °C or more. Past designs for liquid steel included
probes of pure silica but were met with varying degrees of success. This was due to their
extensive softening and erosion during operation. Such a probe is shown in Fig. 3. As
seen in Fig. 4, it was for this reason that the current industrial design for the LiMCA
“ESZpas” probe is a “single-grab” design for steel melts [2] which proved to be
successful.
This version was achieved, exhibiting minimal erosion of the sampling orifice (with
a maximum diameter of 500 μm) while liquid steel flowed through the ESZ. The exact
degree of erosion depended upon the steel grade and was compensated for in the moni-
toring software [2].
Users’ feedback and analysis have demonstrated that the LiMCA ESZpas provide
satisfactory results for medium to high carbon heats. However, the performance of the
Fig. 4. Internal probe construction before and after a sampling sequence; “single-grab” design [2].
LiMCA ESZpas has not been as successful for ultra-low carbon (ULC) steel grades,
owing to their lower superheat temperatures needed for casting. Nonetheless, LiMCA
ESZpas continues to be a field of interest and continued research [2].
In fact, to address these latter factors, including orifice blockage issues at the ori-
fices, owing to the accumulation of inclusions forming a “raft” at their entrances,
McGill research also focused on modeling various orifice shapes of LiMCA probes for
various metals applications [7–9]. In keeping with the prevailing principles that have
dominated the LiMCA ESZpas design, it has been important to evaluate various geo-
metric orifice designs, combined with resistant material construction, capable of with-
standing the more abrasive and “chemically corrosive” nature of liquid steel melts. For
this reason, a more tapered or “fluted” design is used in the “single-shot” steel probe,
while a parabolic orifice design is used for the continuous use of Al-LiMCA sensors, as
shown in Fig. 5a, b, respectively [7].
2 Operating Principles for LiMCA Analyzers
LiMCA is primarily based on the ESZ principle, enunciated in the fundamental work of
J.C. Maxwell. Thus, for measuring the size of non-conducting (or non-metallic) inclu-
sions, the ESZ method is often illustrated and represented schematically in Fig. 6,
shown below:
The essential theoretical problem of the sensor is to record the resistive variation of
the ESZ caused by the passage of inclusion through the orifice. It provides metrics on
the volume concentration of inclusions within a melt and the size distribution of these
inclusions within a melt. The LiMCA technique, by using a signal amplifier, can detect
micro-ohmic changes in resistance during the flow of inclusions through the ESZ in the
presence of large DC currents, to measure non-conducting inclusions.
Fig. 5. (a) Fluted ESZ for steel-LiMCA (b) Centered parabolic ESZ for Al-LiMCA [7].
Fig. 6. Working principle of the electric sensing zone principle [1].
When inclusions flow through the orifice, they displace their volume of the conduct-
ing fluid, causing a momentary rise in the electrical resistance of the orifice. This resis-
tance variation, in the presence of an applied DC current, causes a μV pulse of duration
approximately equal to the transit time of the particle as it passes through the orifice.
The amplitude of the voltage change is, to a first approximation, proportional to the
electric current and the volume/diameter of the particle.
Thus, by counting the number of pulses and measuring their respective amplitudes,
the number, concentration, and size distribution of the inclusion population within a
melt can be inferred. As mentioned, on-line, in-situ, LiMCA is currently used success-
fully in quality and process control operations by leading aluminum-producers such as
Rio Tinto-Alcan, Alusuisse, Alcoa, Reynolds, and Pechiney. Some hundreds of units
are now in commercial use.
According to the ESZ principle, every particle generates a voltage pulse when pass-
ing through the ESZ, in the presence of an electric current. Non-conducting particles, of
the same size, but of a different type, can generate a voltage pulse of the same magni-
tude. This would appear to preclude the discrimination of hard deleterious solid parti-
cles, from harmless microbubbles, and/or liquid droplets, but this is not the case, as
discovered by further research work at McGill.
3 Metal-Based LiMCA Systems
Thus, the principal issue in developing a LiMCA system for molten metals, compared
with aqueous electrolytes, lies in the very low electrical resistivity values vs. aqueous
systems (≈1,000,000 times lower). This renders the changes in electrical resistance
when a non-conducting inclusion displaces liquid metal changes when passing through
the ESZ in the micro-ohm region. The changes in the microvolt pulses are correspond-
ingly small and are given by the following equation:
4r Id 3
V f d / D, (1)
D 4
where f(d/D) is a correction factor referenced and defined in Refs. [1, 7, 10].
This factor is needed when the inclusion size approaches that of the orifice. So, to
obtain substantial and measurable voltage pulses, the DC electrical current needs to be
increased substantially, from ≈ 1 mA, for an aqueous-ESZ, to ≈60 A for an Al-ESZ and
≈24 A for an ESZpas steel system. Similarly, the initial pulse height for particle detec-
tion is greatly lowered into the microvolt range. Fortunately, extraneous background
ambient electrical noise is eliminated in liquid metals, owing to their highly mobile
clouds of electrons. In fact, a LiMCA probe submerged in liquid metal is equivalent to
it being enclosed within a Faraday cage or Faraday shield. So, the critical problem is
to amplify these signals into the millivolt range for recording and classification.
4 Review of Previous Research at McGill’s MMPC
Various materials have been used for the fabrication of the LiMCA probe to date based
on the specific melt. For instance, alumino-silicate tubes are used for aqueous and alu-
minum media. The Type II ESZ LiMCAprobes were manufactured by a glass-blowing
method. Other probe materials used for melts of magnesium alloys have used insulated
concentric steel cylinders, separated by a boron nitride ESZ. More recently, silica tubes
have been used for the type III ESZ probe for steel melts. As any proposed new probe
for steel melts must be used for continuous, in-situ, applications, a material resistant to
flow rates ranging from 1–5 m/s, and resistant to a relatively corrosive environment
needs to be found. For this, any validation of a new LiMCA probe for steel must include
further industrial steel plant tests.
Previous work presented by Mei Li and R.I.L. Guthrie on an Al-LiMCA probe has
attempted to establish an optimal geometric design for the ESZ orifice configuration.
The research addressed problems with possible blockage of the orifice with clusters of
inclusions, and the accumulation of inclusions on the sidewall of the extension of the
Al-LiMCA. It was determined that with the applied DC current, the separation of larger
from smaller inclusions is affected by orifice size, orifice length, orifice geometry,
extension length and radius, as well as the mean fluid velocity through the orifice, and
the DC current. Increasing the DC current, or decreasing the orifice size, lowers the
flow-through fraction of inclusions of any size. A longer orifice gives rise to a lower
flow-through fraction, and this will in-turn decrease as the length increases up to
500 μm (or 0.5 mm). It has been found that the geometry of the orifice has little effect
on the flow-through fraction of the inclusions. However, it is critical to have a smooth
inlet to avoid creating a recirculation zone within the ESZ orifice. It was found that by
increasing the extension radius, the flow-through fraction of small inclusions increased,
while larger ones decreased. Furthermore, the flow-through fraction also increased for
both increases in the extension length, and the mean fluid velocity through the orifice
[11, 12].
Further research has been conducted in analyzing the effects of particle conductivity,
density, and size. M. Li and R.I.L. Guthriealso demonstrated that non-conductive inclu-
sions follow a trajectory through the ESZ closer to the sidewall, while inclusions that
are more conductive than the molten steel follow a trajectory closer to the central axis.
It was also shown that flow rates are affected by the density and size of the inclusions.
For instance, microbubbles tend to lead flow in the axial direction and travel faster than
Al2O3 particles, which are denser than the fluid and lag the metal flow. These impact
particle transit times through the ESZ, as larger particles have longer transit times than
smaller particles. Microbubbles demonstrate even shorter transit times versus alumina
particles of a similar size, with the difference in transit times more significant for larger
particles [11, 12].
Wang, Isac, and Guthrie have provided numerical studies to evaluate a new sen-
sor inlet geometry for the in-situ measurement of inclusions in liquid steel using the
LiMCA-ESZ technique. In that study, a theoretical approach of a two-phase fluid
flow model was used to measure the effects within the ESZ by estimating the elec-
tromagnetic fields and forces and their impact on the flow fields and particle
motions. This study confirmed the results of previous studies by Mei Li and RIL
Guthrie, in that given the introduction of relatively high currents, these can be used
to reverse the flow of liquid aluminum from the central region of the ESZ, thereby
breaking up any inclusion agglomerations at the entrance to the ESZ. As such, these
strong electrical currents can combine with varying entrance geometric configura-
tions of typical ESZ’s. More specifically, for fluted and parabolic orifice designs,
these can cause metal flow recirculation of inclusion, and variable flow-through
fractions [8, 9].
Relatively recent research in 2009 by Wang, Isac, and Guthrie, further highlighted a
preferred “fluted” geometric design of the ESZ for the Steel-LiMCA and its influence
on the electromagnetic field. There, it was shown that a design that had a non-divergent
outlet from the ESZ would avoid recirculation of the fluid flow back into the orifice. In
addition, improved flow parameters have provided a relatively wider resistive pulse
peak, enabling improved inclusion discrimination. Optimal design parameters for the
ESZ include (i) a fluted orifice, with an entrance to throat radii ratio of approximately
1.6:1, (ii) a configuration that consists of a slight parabolic inlet and rectangular cylin-
der to avoid recirculating flows into the ESZ, and (iii) n optimal operating DC current
that should be about 20 A, with an orifice conditioning current of 120 A [8, 9].
5 Experimental Measurement Results
The control of non-metallic inclusions represents an essential step in the production

process of aluminum alloys, including furnace preparation, alloying practice, feedstock
mix, settling time, and their effect on melt cleanliness. Figure 7 below shows LiMCA
III developed by ABB [5].
LiMCA technology has been widely used to characterize and improve alloy prepara-
tion techniques, in-line treatment units, and casting practices. The size distribution and
total concentration of inclusions are displayed in real-time on the computer screen.
LiMCA proved essential in monitoring the effects of settling on cleanliness, or filter
efficiency or measuring inclusion release from a sudden metal level change while cast-
ing [4, 5].
As demonstrated in the following Figs. 8, 9, and 10, the need to report specific infor-
mation regarding the time to settle, the number of inclusions present or released at each
casting change, and information related to the efficacy of the in-situ filtration system
are important parameters that have been made available by some of the features devel-
oped for the use of ABB’s LiMCA CM system [6].
Fig. 7. ABB LiMCA III [5].

Fig. 8. LiMCA Showing the effects of settling on cleanliness [5].
Fig. 9. LiMCA results showing filter efficiency [5].
5.1 Measurements in a Steelmaking Tundish
The population of large inclusion through the casting process could also be visualized
by using the distribution charts shown in Fig. 11. The chart displays the particles sized
greater than 20 μm in 1 kg of steel. However, due to the limitation of the measurement
system, the small particles sized below 35 μm could not be well resolved [13].
5.2 Measurements in Ladle
As indicated in Fig. 12, the inclusions in the ladle are larger than that in tundish; and the
inclusion size varies by the process step, such as wire injection and time.
Also, the number of inclusions is strongly influenced by the degree of argon stirring,
which drives the inclusions to float up [13].
Fig. 10. LiMCA measuring inclusion release from a sudden metal level change while casting [5].
Fig. 11. Size distribution determined from various Heats (resolution set at below 30 μm) [13].
Index: 15 Index: 17
180 180
160 160
140 140
Size, micron
120
Size, micron
120
100 100
80 80
60 60
40 40
20 20
1 2 3 4 5 1 2 3 4 5
Time, second Time, second
Fig. 12. Inclusion scatters for various measurements at ladle station [13].
6 Conclusions and Future Development
Several variations of LiMCA have come to pass over the past four decades. From
LiMCA’s initial research and work developed at McGill’s MMPC using the principles
developed by JC Maxwell, and the Coulter Counter Technology, to efforts developed by
ALCAN, Heraeus Electro-Nite, BOMEM, and later ABB, a marketable system has
been manufactured and successfully used for the Aluminium industry. As discussed, a
multitude of disciplines come into play for the successful development and use of
LiMCA for liquid metals. Successful research at the MMPC has shown that the theory
behind LiMCA can be successfully used to distinguish between conventional solid
inclusions, liquid droplets, and air microbubbles [14].
However, applications to molten steel metal at high temperatures have yet to be
materialized, and this remains an interesting and challenging area for future research.
Acknowledgments The authors acknowledge the financial support of NSERC, as well member com-
panies of the McGill Metals Processing Centre, Rio Tinto Iron and Titanium, Nippon Steel Corporation,
Japan, as well as Quebec Centre of Aluminum Research and Development.
References
1. Guthrie RIL, Isac M. In-situ sensors for liquid metal quality. J High Temp Mater Processes.
2012;31(4/5):633–43.
2. Isac MM, Guthrie RIL. On-line sensors for monitoring steel quality. 5th ICS 2012 Proceedings,
Dresden (Germany).
3. Doutre D. The development and application of a rapid method of evaluating molten metal cleanli-
ness. Doctorate Thesis, Department of Mining and Metallurgical Engineering, McGill University,
May 1984.
4. BOMEM Hartmann & Braun Brochure: LiMCA II system – description and specifications.
5. ABB Measurement & Analytics Brochure: LiMCA III – liquid metal cleanliness analyzer.
6. ABB Measurement & Analytics Brochure: LiMCA CM – liquid cleanliness analyzer for continu-
ous monitoring.
7. Guthrie RIL, Mei L. In situ detection of inclusions in liquid metals: part I. mathematical mod-
elling of behaviour of particles traversing the electric sensing zone. Metall Mater Trans
B. 2001;32B:1067–79.
8. Wang X, Isac M, Guthrie RIL. Numerical studies on the in-situ measurement of inclusions in liquid
steel using the E.S.Z. or LiMCA technique. ISIJ Int. 2009;49(7):975–84.
9. Wang X, Isac M, Guthrie RIL. Modeling investigation of a new sensor for detecting inclusions in
liquid steel. Dresden, EMP 2009, 19–23 Oct 2009.
10. Guthrie RIL, Isac M. In-situ sensors for the direct detection of inclusions in liquid steel.
International Conference on Clean Steel 2012, Budapest.
11. Li M, Guthrie RIL. On the detection and selective separation of inclusions in liquid metal cleanli-
ness analyzer (LiMCA) systems. Metall Mater Trans B. 2000;31B:767–77.
12. Li M, Guthrie RIL. Liquid metal cleanliness analyzer (LiMCA) in molten aluminum. ISIJ Int.
2001;41(2):101–10.
13. Liu CC, Conti RF. Inclusion counter-an on-line measurement system for steelmaking. USA:
Heraeus Electro-Nite Co.
14. Guthrie RIL, Wang X, Isac M, Conti R, Stone R. An overview of recent developments in experi-
mental and numerical studies for the in-situ Measurement of large inclusions (≈50 – 400 microns)
in liquid steel processing vessels. Mater Sci Technol. 2009;2:1163–74.
Cr-Oxide: Why Should You Have It in Your
Refractory?
Dean Gregurek1 (*), Philip W. Schantl1, Jürgen Schmidl2, and Alfred Spanring2
1
RHI Magnesita Technology Center Leoben, Leoben, Austria
dean.gregurek@rhimagnesita.com
2
RHI Magnesita, Vienna, Austria
Abstract. Chromium oxide is one of the oxides within chromite (Fe2+, Mg2+)(Cr3+, Al3+,
Fe3+)2O4 that belongs to the spinel group of minerals. It is an important strategic raw mate-
rial particularly in the production of refractories, widely used in non-ferrous metal ves-
sels. In the non-ferrous furnaces (e.g., copper, nickel, lead) the refractory linings are
subjected to either continuous wear by hot erosion or discontinuous wear caused due to
batch-wise processes and frequent thermal shocks. The refractory wear will begin usually
by weakening of the brick microstructure after infiltration by foreign substances (e.g.,
slag, sulfate, matte, and copper/lead) and corrosion of the brick-inherent components. The
continuous wear is caused by corrosion of refractory due to diffusion-controlled dissolu-
tion in slag or by hot erosion, triggered by fluid motion at the interface between refractory
and slag. The main advantage of the chromite within the magnesia-chromite bricks is high
corrosion resistance against acidic slags (e.g., fayalite slag). The lower chromite solubility
in such a slag can be explained by phase diagrams (i.e., miscibility gap between chro-
mium and silica in the Cr2O3-SiO2 phase diagram) and FactSage™ calculations. In case
of discontinuous furnace operation, the chromite addition reduces the structural brittle-
ness and it improves the thermal shock resistance.
Keywords: Refractories · Chromite · Non-ferrous Metal Vessels · Slag Corrosion
1 Introduction
Chromite (Fe2+, Mg2+)(Cr3+, Al3+, Fe3+)2O4 is an oxide mineral belonging to the spinel
group of minerals. The spinels include a class of minerals with general formulae AB2O4
which crystallize in the cubic (isometric) crystal system, with oxygen arranged in a
cubic close-packed lattice and the cations A and B occupying some or all of the octahe-
dral and tetrahedral sites in the lattice [1]. Although the charges of A and B in the pro-
totypical spinel structure are (+2) and (+3), respectively, other combinations
incorporating divalent, trivalent, or tetravalent cations, including magnesium, zinc,
iron, manganese, aluminium, chromium, titanium, and silicon, are also possible [2, 3].
The annual production of chromite is estimated atapproximately ~15 Mio. tons [4].
It is mainly used for the production of ferrochrome (FeCr), which is an iron-chromium
alloy used instainless steel [5]. The chromite is also used in the refractory industry,
namely for the production of magnesia-chromite bricks and monolithics. However,
https://doi.org/10.1007/978-3-031-17425-4_60
464 D. Gregurek et al.
demand of chromite in the refractory industry is estimated to approximately ~1–2% of

the world’s production. Magnesia-chromite bricks are installed in the non-ferrous (cop-
per, nickel, lead) furnaces and in the iron and steel industry (e.g., pig iron mixers, AOD
converters, and steel degassing units) [6].
The present paper gives an overview of the main properties of the magnesia-chromite
bricks. It also highlights the importance of chromite addition to the refractories in terms
of corrosion resistance and thermomechanical properties. The knowledge was collected
from the post-mortem investigations on spent refractories from the non-ferrous fur-
naces supported bythermomecanical considerations and FactSage™ calculations. The
type of furnace operation (i.e., continuous or batch-wise) as well as the slag type used
(i.e., mainly fayalite slags) has a significant influence on the dominating refractory wear
mechanism.
2.1 Properties of the Magnesia-Chromite Bricks
The main chrome ore types used in the refractory production are shown in Table 1. The
ores in group A represent concentrate ores in which small and homogeneously distrib-
uted chromite crystals are embedded in enstatite (Fig. 1a). Those in group B and C are
lump ores with large chromite crystals.
For the refractory production it is necessary to classify the chrome ores into three
groups which differ specifically in their chemical analysis (i.e., Cr2O3/Al2O3ratio, Fe2O3,
and SiO2 content) [6]. For the refractory applications, the SiO2-content should be low
and the ratio Cr2O3/Al2O3, as high as possible. Additionally, the grain size of chrome ore
plays an important role. Basically, there is distinction between lump ore (originates
from massive or disseminated chrome ore; no upper grain size limit) or concentrates
(separation by oscillating tables; grain size <2 mm, with the main proportion <1 mm).
2.2 Chemical Corrosion of Magnesia-Chromite Bricks
Mineralogical Investigation The corrosion of the refractories particularly magnesia-

chromite bricks by slag attack manifests itself mainly by [7, 8]
Table 1. Chemical composition of chrome ores (wt.%).

Group MgO Al2O3 SiO2 CaO Cr2O3 Fe2O3 Cr2O3/Al2O3
A 12–14 12–16 0.1–0.9 <0.3 44–48 24–28 2.8–4
B 16–18 8–13 0.5–5 0.1–0.5 47–57 16–19 3.1–7
C 18–20 25–29 3–6 0.5–0.8 31–35 14–17 1.1–1.4
Modified after [6]
Cr-Oxide: Why Should You Have It in Your Refractory? 465
Fig. 1. Microstructural detail of an used magnesia-chromite brick. Reflected light microscopy.

Corroded magnesia (MgO). Not corroded chromite. Formation of forsterite . Copper (Cu).
• Dissolution reaction occurring at the immediate brick hot face

• Dissolution and chemical reaction within the refractory microstructure
• Kinetics of slag infiltration that depends on several parameters like temperature,
viscosity, pore size distribution and wetting angle
Figure 1 shows at the microscopic level corrosion of the magnesia-chromite brick in
contact with fayalite slag. There is a severe dissolution of the magnesia brick compo-
nent frequently leaving relics of chromite. Due to corrosion of the brick-inherent mag-
nesia (coarse grains and matrix fines), the main reaction products include magnesium
silicate such as forsterite (Mg2SiO4). In general, chromite is not corroded compared to
the magnesia but at rims enriched with iron and copper oxide due to diffusion phenom-
ena. The reason for this is the fact that periclase (MgO) is more basic than chromite, and
therefore more susceptible to acidic corrosion.
The fayalite slag is the most common slag type in the non-ferrous furnaces. However,
in some metallurgical processes Ca-Fe-oxide (calcium-ferrite) slag is preffered.
Due to a much lower viscosity compared to fayalite slag, the calcium-ferrite slag is
able to penetrate much deeper into the refractory. Within a deeply infiltrated brick
microstructure there is severe corrosion of both brick components, i.e., magnesia and
chromite. Neverthless, the chromite corrosion compared to the magnesia is on signifi-
cantly lower level. Additionally, a high supply of iron oxide from the slag results in the
formation of low melting Mg-Fe-oxide (magnesio-wustite, (Mg,Fe)O).
Dietzel’s Field Strength and Phase Diagrams The solubility of refractory compo-
nents in a melt was explained by Adolf Dietzel [9], who observed an increasing ten-
dency of compound formation and species dissolution with increasing difference in
basicity/acidity between species to be dissolved and slag compounds. In order to quan-
tify the basicity/acidity of a single species, Dietzel’s field strength (Fs) was defined. It
is expressed through the formula FS = ZC/a2, whereas ZC is the valence of the cation and
a is the distance [Å] between the cation and the bonded oxygen.
In the present study, the corrosive resistance of magnesia-chromite bricks against

silica-dominated melts (e.g., fayalite slag) were considered and discussed in terms of
solubility under the application of Dietzel’s field strength. The FS for the respective
cations are Mg2+ = 0.51 [Å−2], Cr3+ = 0.94 [Å−2], and Si4+ = 1.56 [Å−2]. Thus, a pure
magnesia brick in contact with a silica dominated slag implies a high ΔFs of 1.01.
[Å−2] between Mg2+ and Si4+, which results in a high tendency of compound forma-
tion and a high solubility and corrosion of magnesia within a silica melt. By contrast, a
magnesia-chromite brick would show a much higher resistance against a silica-
dominated slag attack. This is due to Cr3+ addition and the much lower ΔFs of 0.42
between Cr3+ and Si4+, representing their low affinity to each other and high resistance
of Cr against dissolution in silica slags. This can be also demonstrated by the large
stable miscibility gap in the binary phase diagram Cr2O3-SiO2 (see Fig. 2).
For demonstrating the dependency between MgO solubility and the Cr2O3 content in
a magnesia-chromite brick that interacts with a pure silica melt, the ternary MgO–
Cr2O3–SiO2 phase diagram was considered (see Fig. 3a, b). At a temperature of 1750 °C
Fig. 2. Binary phase diagram SiO2-Cr2O3 after [10]. The diagram shows the large stable miscibility
gap between SiO2 and Cr2O3.
Fig. 3. Ternary phase diagram MgO–Cr2O3–SiO2 after [11]. The diagram illustrates the dependence
between MgO solubility within the liquid phase and the Cr2O3 of a magnesia-chromite product. At a
temperature of 1750 °C (point 1 in (a)), the liquid phase is in equilibrium with picrochromite and dis-
solves 28.7 mass % of MgO. If the Cr2O3 concentration exceeds 20 mass % in the magnesia-chromite
brick, the MgO solubility in the liquid phase decreases dramatically and lies at maximum 0.18 mass.
% (point 2 in (b)).
the liquid phase is in equilibrium with picrochromite and dissolves 28.7 mass % of
MgO (point 1 in Fig. 3a) when pure SiO2 is in contact with a magnesia-chromite brick
including 20 mass. % Cr2O3. If the Cr2O3 concentration in the magnesia-chromite prod-
uct decreases, the solubility of MgO within the liquid phase increases. In contrary, if the
Cr2O3 concentration exceeds 20 mass % in the magnesia-chromite brick, the MgO solu-
bility in the liquid phase decreases dramatically and lies at maximum 0.18 mass. %
(point 2 in Fig. 3b). A further increase of the Cr2O3 amount in the magnesia-chromite
brick would only lower the MgO solubility insignificantly within the liquid phase.
Thermo-chemical calculations by FactSage™ The solubility of the refractory compo-
nents (e.g., MgO, Al2O3, and Cr2O3) in the CaO–SiO2–FeOx–O2 system was calculated
with FactSage™ software by using databases FactPS and FToxide [12]. For both slag
types—fayalite/olivine and calcium-ferrite slag the results were summarized as follows:
• The solubility of refractory components increases in the order of

Cr2O3–MgO–Al2O3
• MgO and Al2O3 is sensitive to temperature increases while the Cr2O3 solubility is
rather stable
• Solubilities of Cr2O3 and MgO are not sensitive to any changes in the Fe/SiO2 ratio,
i.e., solubility increased solely as a function of temperature
• With increasing Fe/SiO2, the Al2O3 solubility decreases
• Al2O3 is highly sensitive to temperature increases while the MgO and Cr2O3 solubil-
ity increases and the MgO comes closer to Cr2O3 at higher temperatures. Also the
maximum solubility of Al2O3 is a multiple higher than for MgO and Cr2O3
2.3 Thermo-Mechanical Behavior of Magnesia-Chromite Bricks
In case of batch-wise processes, the refractory lining is suffering from high thermo-
mechanical load. The infiltration of the refractory microstructure increases the struc-
tural brittleness. Temperature changes due to discontinuous furnace operation create
stresses in the brick lining which can only be absorbed in limited extent, thus leading to
crack formation, as soon as microstructural strength is exceeded.
Chromite addition reduces the structural brittleness and it improves the thermal
shock resistance. This can be explained by initiation of microcracks particularly during
burning process. Microcracks in the magnesia-chromite bricks are usually generated by
differences in the thermal expansion between chromite and the matrix (i.e., thermal
misfit). The thermal misfit is caused by lower thermal expansion of chromite in com-
parison to the matrix (i.e., thermal expansion coefficients ~8.5·10−6/K and ~13.5·10−6/
K, respectively) [13].
3 Conclusions
In the non-ferrous furnaces, the refractory linings are subjected to either continuous
wear by hot erosion or discontinuous wear caused due to batch-wise processes and
frequent thermal shocks. The refractory wear will begin usually by weakening of the
brick microstructure after infiltration by foreign substances (e.g., slag, sulfate, matte,
and copper/lead) and corrosion of the brick-inherent components. This results in soft-
ening of the brick microstructure, loss of flexibility, and brick strength. The fayalite slag
mainly corrodes the magnesia component of the brick, but not the chromite. The chro-
mite shows generally a higher corrosion resistance against both fayalite slag and sulfur
attack. The lower chromite solubility in such slag can be explained by phase diagrams
(i.e., miscibility gap between chromium and silica in the Cr2O3-SiO2 phase diagram)
and FactSage™ calculations. Thus, the primary role of chromite addition is increased
corrosion resistance of the refractory particularly in the case of slags ranging from
acidic to slightly basic.
In case of thermal shock, the changed thermo-mechanical properties (i.e., increased

refractory brittleness) yield to crack formation, primarily at the interface between the
infiltrated and non-infiltrated brick area and finally to discontinuous wear by spalling.
Chromite addition reduces the structural brittleness and it improves the thermal shock
resistance.
References
1. Deer WA, Howie RA, Zussman J. An introduction to the rock-forming minerals. 2nd ed. London:
Longman Publishing Co.; 1992.
2. Naumann RJ. Introduction to the physics and chemistry of materials. Boca Raton: CRC Press; 2008.
3. Meyer H-J. Riedel moderne anorganische chemie. Walter de Gruyter; 2012.
4. World mining data: information. https://www.world-mining-data.info. 2021. Accessed 28
Nov 2021.
5. Gasik M. Handbook of ferroalloys – theory and technology. 1st ed. Oxford: Butterworth-
Heinemann; 2013.
6. Routschka G, Wuthnow H. Handbook of refractory materials. 4th ed. Essen: Vulkan-Verlag; 2012.
7. Gregurek D, Reinharter K, Majcenovic C, Wenzl C, Spanring A. Overview of wear phenomena in
lead processing furnaces. J Eur Ceram Soc. 2015;35:1683–98.
8. Harmuth H, Vollmann S. Refractory corrosion by dissolution in slags – challenges and trends of
present fundamental research. Iron Steel Rev. 2014;58(4):157–70.
9. Dietzel. Glastech. Ther Ber. 1948;22:41.
10. Levin EM, Robbins CR, McMurdie HF. Phase diagrams for ceramist. Columbus: American
Ceramic Society; 1964.
11. Keith ML. System Cr2O3-SiO2. J Am Ceram Soc. 1954;37(10):490–6.
12. Bale CW, Chartrand P, Decterov SA, Eriksson G, Hack K, Mahfoud B, Melançon KR, Pelton J,
Petersen S. FactSage thermochemical software and databases. Calphad. 2002;62:189–228.
13. Harmuth H, Tschegg EK. A fracture mechanics approach for the development of refractory materi-
als with reduced brittleness. Fatigue Fract Eng Mater Struct. 1997;20(11):1585–603.
Microstructural Study of Containerless Solidification
of Al-20wt%Ce Alloys
Hani Henein1 (*), Akankshya Sahoo1, and Jonas Valloton1
1
Advanced Materials and Processing Laboratory, Department of Chemical and Materials
Engineering, University of Alberta, Edmonton, AB, Canada
hani.henein@ualberta.ca
Abstract. Aluminum alloys are heavily used within the transportation sector due to their
excellent mechanical properties, high strength to weight ratio, corrosion resistance, and
high thermal conductivity. Cerium-containing aluminum alloys have become a focus of
interest for high-temperature alloys. Al-Ce-based alloys consist primarily of fcc alumi-
num and Al11Ce3 intermetallic. Due to the near-zero solubility of cerium in fcc aluminum
(less than 50 ppm at the eutectic temperature), the strengthening Al11Ce3 intermetallic is
very stable against high-temperature dissolution. Furthermore, while cerium is the most
abundant of the rare earth elements, it has yet to have any substantial high-volume use. It
currently remains relatively inexpensive as an alloying element. In this work, container-
less solidification of hypereutectic Al-20wt%Ce is carried out using ElectroMagnetic
Levitation and Impulse Atomization.
1 Introduction
Aluminum alloys are heavily used within the transportation sector due to their excellent
mechanical properties, high strength to weight ratio, corrosion resistance, and high
thermal conductivity. Cerium-containing aluminum alloys have become a focus of
interest for high-temperature alloys. Al-Ce-based alloys consist primarily of fcc alumi-
num and Al11Ce3 intermetallic. Due to the near-zero solubility of cerium in fcc alumi-
num (less than 50 ppm at the eutectic temperature), the strengthening Al11Ce3
intermetallic is very stable against high-temperature dissolution. Furthermore, while
cerium is the most abundant of the rare earth elements, it has yet to have any substantial
high-volume use. It currently remains relatively inexpensive as an alloying element. In
this work, containerless solidification of hypereutectic Al-20wt%Ce is carried out using
ElectroMagnetic Levitation and Impulse Atomization.
https://doi.org/10.1007/978-3-031-17425-4_61
472 H. Henein et al.
2 Research Methodology and Results
Al-20wt%Ce is of hyper-eutectic composition (Fig. 1 left). The solidification path as

predicted by Scheil is as follows (Fig. 1 right): upon cooling, solidification of
Al-20wt%Ce starts at TL = 813.5 °C with the crystallization of the Al11Ce3 intermetallic.
Then, a eutectic reaction yields α-Al and Al11Ce3 at TE = 640.5 °C. No other phases are
expected to form.
Al-20wt%Ce samples were solidified using two containerless solidification tech-
niques: Electromagnetic levitation (EML) [1] and impulse atomization (IA), a type of
drop tube technique [2]. EML allows for the solidification of one 6–7-mm-diameter
sample per run with a direct measurement of the cooling rate and undercoolings. One
single run of IA yields a large quantity of droplets ranging from ~100 to 1000 μm. The
samples experience large cooling rates which increase with decreasing diameter and
variable undercoolings. While cooling rates and undercoolings cannot be directly mea-
sured in IA, the former can be estimated with a thermal model of the process [3, 4].
Figure 2 shows the microstructure of a sample solidified in EML. The white phase is
the Al11Ce3 intermetallic while the dark gray phase is the α-Al fcc solid solution.
Numerous dendrites of pro-eutectic Al11Ce3 are observed throughout the samples. Their
dispersion and seemingly random orientation suggest multiple nucleation events. A fine
eutectic surrounds the primary phase. The eutectic Al11Ce3 phase takes different mor-
phologies. A typical lamellar eutectic is observed, as well as more complex C- and
U-shaped (or so-called Chinese script) eutectic intermetallics. There is a halo of α-Al
surrounding the primary phase as shown in Fig. 2 (right).
The microstructures of two droplets solidified in IA are presented in Fig. 3. The
larger one, with a diameter between 850 and 1000 μm (Fig. 3 left), exhibits a micro-
structure qualitatively similar to that of the EML sample: numerous dendrites of pro-
eutectic Al11Ce3 surrounded by the α-Al-Al11Ce3 eutectic. However, due to the higher
Fig. 1. Al-Ce phase diagram and the Gulliver–Scheil solidification path calculated using Thermo-Calc
TCAL7 database [5].
Microstructural Study of Containerless Solidification of Al-20wt%Ce Alloys 473
Fig. 2. EML sample. Left: full sample. Right: close up.
Fig. 3. IA samples from different size ranges. Left: 850–1000 μm. Right: 150–212 μm.
cooling rates involved in IA, the microstructural features are much finer than in EML.
The smaller droplet (150–212 μm diameter, Fig. 3 right) presents a drastically different
microstructure. A very fine eutectic structure seems to develop first, followed by elon-
gated Al11Ce3 lamella which are then surrounded by a regular eutectic. The undercool-
ing experienced by the smaller particles is likely large enough to bypass primary
nucleation of Al11Ce3. The temperature of the liquid prior to nucleation is below the
eutectic temperature and growth of a coupled zone is seen instead. During recalescence,
it is supposed that the droplet temperature increases over TE, which leads to the nucle-
ation of “primary” Al11Ce3. Finally, as the droplet further cools, the remaining liquid
solidifies as a coarser eutectic. These experimental observations are currently being
quantitatively analyzed with in-depth.
Acknowledgments The financial support of the Natural Sciences and Engineering Research Council
of Canada (NSERC) and Equispheres is gratefully acknowledged.
474 H. Henein et al.
References
1. Herlach DM. Containerless undercooling of drops and droplets. In: Solidification of containerless
undercooled melts. Hoboken: John Wiley & Sons, Ltd; 2012. p. 1–30.
2. Henein H. Single fluid atomization through the application of impulses to a melt. Mater Sci Eng
A. 2002;326(1):92–100.
3. Wiskel JB, Henein H, Maire E. Solidification study of aluminum alloys using impulse atomization:
part I: heat transfer analysis of an atomized droplet. Can Metall Q. 2002;41(1):97–110.
4. Valloton J, Bogno A-A, Rappaz M, Henein H. Numerical model of rapidly solidified droplets of
Al–33 wt pct Cu eutectic growth. Metall Mater Trans A. 2022;53(2):460–9.
5. Andersson J-O, Helander T, Höglund L, Shi P, Sundman B. Thermo-Calc & DICTRA, computa-
tional tools for materials science. Calphad. 2002;26(2):273–312.
Processing, Properties, and Applications
of Lightweight Al and Mg Alloys
P. Emadi1, B. Andilab1, A. Belcastro1, and C. Ravindran1 (*)

1
Centre for Near-net-shape Processing of Materials, Toronto Metropolitan University,
Toronto, ON, Canada
rravindr@ryerson.ca
Abstract. The future of materials engineering will be focused upon environmental sus-
tainability. Improvements in material processing and properties have the potential to
reduce greenhouse gas emissions. With the ongoing development of next generation
vehicles, lightweighting of components must be promoted, property enhancement
becomes critical, and sustainability challenges will need to be addressed. At the Center for
Near-net-shape processing of materials, future focused research is being carried out. The
current article summarizes our recent progress towards improving the properties of cast
magnesium and aluminum alloys in the fields of grain refinement, thermal conductivity
enhancement, melt cleanliness, and processing using modern techniques. Examples of
our present research as well as the outlook of the materials science field will be
presented.
Keywords: Aluminum · Magnesium · Thermal Conductivity · Mechanical

Properties · Grain Refinement · Sustainability
1 Introduction
With the ongoing crisis of global warming, increasing GHG emissions, and climate
change, scientists from across the world have been directing their efforts toward light-
weighting and high-strength alloys. The common goal is to replace higher density
materials such as iron (Fe) and steel with lightweight alloys. However, despite the
necessity to develop towards a sustainable future, carbon emissions are constantly
increasing.
To achieve lightweighting, high specific-strength alloys such as aluminum (Al) or
magnesium (Mg) are well-known candidates. For example, Mg alloys can be 70%
lighter than steel and 30% lighter than Al [1]. However, despite the benefits of using Mg
or Al alloys for lightweighting, they are not as predominant as Fe-based alloys, which
is largely due to their low mechanical properties. Consequently, to promote the use of
lightweight and high-strength alloys in the aerospace, transportation, and consumer
electronics industries, their strength and ductility must be improved.
https://doi.org/10.1007/978-3-031-17425-4_62
476 P. Emadi et al.
For many years, researchers have been exploring alternative methods for improving
the properties of light alloys. Grain refinement has been identified as a potential method
to achieve this goal [2]. It can be achieved through various methods, such as nucleation
from particle substrates or grain growth restriction by alloying. Decreasing grain size
can simultaneously improve the strength and ductility of alloys. Moreover, grain refine-
ment helps reduce casting defects. Thus, grain refinement is an effective method for
enhancing the properties of light alloys and promoting their increased use in industry.
The thermal properties of cast alloys must be improved as well. This is especially
important for conventional internal combustion (IC) engines and electric vehicles (EV).
For IC engines, the components that operate at higher temperatures, such as engine
blocks, must maintain an even heat distribution and dissipate heat efficiently. If the heat
is not sufficiently dissipated, thermo-mechanical stresses result from an uneven heat
distribution, thereby leading to reduced engine efficiency and decreased component
life. Heat dissipation is important for electric vehicles as well. Heat is generated from
electric motors and batteries that must be dissipated through their housing components.
Insufficient thermal management can reduce the performance of the battery and as a
result the range and product life of EVs [3]. Therefore, maintaining an even heat distri-
bution is crucial for efficient, safe, and reliable operation of EVs and IC-based
automobiles.
Important modern technologies such as ultrasonic processing are also essential for
encouraging the use of lightweight alloys. Recently, ultrasonic irradiation is one tech-
nology that has been used successfully in the refinement of lightweight alloys [4–6].
Potent grain-refining capability, ease of operation, and cost-effective equipment sup-
port its potential for industrial application. As well, since the molten viscosity of Mg is
comparable to that of water, it can be processed with low-wave attenuation [7, 8].
Although the research in this field is growing, the process is largely undeveloped and
requires further study.
In this review, various methods for improving the mechanical and thermal properties
of lightweight alloys will be discussed. The authors wish to encourage a clearer under-
standing of fundamentals and innovative approaches to improving light alloys. The
need for this article originates from the immense potential of lightweight metals in
shaping our future. Materials scientists are constantly contributing to this global effort.
However, the pursuit for developing innovative materials that can address the chal-
lenges of sustainability, environment and energy is still ongoing.
2 Grain Refinement
Grain refinement is an important method for enhancing the mechanical properties of

cast alloys. Grain refinement can be achieved through various techniques. This article
focuses on enabling grain refinement through heterogeneous nucleation from particle
substrates. Grain refinement through heterogeneous nucleation involves the addition of
nucleants to act as substrates for the formation of grains. During past several years,
Processing, Properties, and Applications of Lightweight Al and Mg Alloys 477
many compounds have been evaluated for their grain-refining efficiency in Mg alloys.
In this section, some of the more recent research in this field will be presented.
When considering grain refinement through particle addition, selection of a suitable
particle chemistry is extremely critical. Grain-refining particles must have a higher
melting point than the liquid metal and be stable at elevated temperatures [9]. Grain
refiner particle density should also be near that of the liquid metal [10, 11]. If particle
density is higher than that of the melt, there is risk of rapid settling of refiner particles.
Finally, to minimize the energy needed for effective nucleation, the particles and the
host metal should have high crystallographic registry.
Nitride-based particle addition has shown great promise as nucleating particles for
Mg alloys [12, 13]. In particular, AlN stands out as an excellent candidate for Mg nucle-
ation. This is in part due to its similar density (3.03 g/cm3), high-temperature stability,
and similar crystal structure to Mg (HCP) [14]. A recent study performed by Wang et al.
[14] examined the effects of AlN nanoparticle addition to a Mg-3Sn-1Ca alloy. The
experiments involved addition levels ranging from 0.1 to 1 wt.%. Significant grain
refinement was observed by the researchers, as demonstrated by the 76% reduction in
grain size through the addition of 1 wt.% AlN. This was attributed to heterogeneous
nucleation from AlN particle substrates. However, the highest tensile properties were
achieved through the addition of 0.5 wt.% AlN. This resulted in improvements in ten-
sile strength, yield strength, and elongation of 38%, 34%, and 19%, respectively. The
mechanisms contributing to the improved strength were Orowan strengthening, grain
boundary strengthening, and refinement of CaMgSn-based secondary phases. The
authors also noted that the tensile properties did not increase beyond 0.5 wt.% AlN
addition, due to particle agglomeration.
Apart from direct addition of particles, grain refinement can be achieved through in
situ formation of desired substrates either in the melt or through master alloy addition.
This is often utilized for boride addition, such as TiB2 [15–18]. A recent study per-
formed by Sahoo et al. [17] investigated the use of in situ formed TiB2 particles on the
microstructure and mechanical properties of ZE41 Mg alloy. In the study, they reported
a reduction in grain size from 95 μm in the base alloy to 32 μm in the alloy with
15 wt.% TiB2. However, it was reported that the mechanical properties at 15 wt.% TiB2
were compromised due to grain refiner particle agglomeration. The highest mechanical
properties were attained at the 10 wt.% TiB2 addition level, which resulted in an
improvement in YS, UTS and %Elongation of 138%, 113% and 5.3%, respectively. As
a result, TiB2 was effective in improving the grain size and mechanical properties of
ZE41.
Judicious selection and addition of oxide particles can also lead to effective grain
refinement. For example, a recent study performed by the authors examined the effects
of 1 μm sized Al2O3 particles on the microstructure and mechanical properties of
AZ91E Mg alloy [19]. The authors examined addition levels between 0.25 and 2 wt.%.
Through the addition of 1 wt.% Al2O3, the grain size and yield strength improved by
61% and 11%, respectively (Fig. 1). The maximum levels of improvement for ulti-
mate tensile strength and ductility, achieved through 0.5 wt.% Al2O3 addition, were
15% and 79%, respectively. In addition, refinement of the alloy secondary phases was
also observed. These findings were attributed to a combination of grain refinement
478 P. Emadi et al.
Fig. 1. Optical micrographs, captured post-heat treatment and etching, displaying the typical grain
structure of (a) AZ91E base alloy and (b) AZ91E + 1 wt.% Al2O3. (Reprinted by permission from:
Springer Nature, Emadi et al., copyright (2022) [19]).
and coefficient of thermal expansion mismatch between the matrix and the reinforce-
ment particles.
3 Factors Affecting the Thermal Conductivity of Aluminum

Alloys
Thermal conductivity is a material property that characterizes the ability to transfer heat
internally, from a high- to low-temperature region. Conduction through a material
occurs through lattice vibration waves and free electrons, wherein both act as thermal
energy carriers. Atomic vibrations within a crystal lattice involve the transfer of pho-
nons through a material with a temperature gradient from high to low temperature,
where thermal energy is converted into kinetic energy by free electrons. Hence, an
increase in kinetic energy of free electrons promotes more efficient heat transfer. In this
section, an overview of studies on factors that affect the thermal conductivity of Al
alloys is discussed.
3.1 Alloying and Microstructure
Aluminum alloys are less conductive compared to pure Al due to the presence of alloy-
ing elements or impurities that act as scattering sites for thermal energy carriers. The
solubility limit of alloying elements is a factor that determines if they can be in solid
solution or precipitate as a secondary phase. The size, morphology, and volume fraction
of secondary phases, as well as the concentration of alloying elements in solid solution,
contribute to a reduction in thermal conductivity. It is known that elements in solid solu-
tion independently increase the resistivity of an alloy with increasing solute concentra-
tion. For example, the electrical conductivity of Al-Si-Mg alloys is known to decrease
linearly with increasing silicon (Si) or Mg concentrations [20]. The decreases in con-
ductivity were measured as 12% IACS per wt.% Si and 10% IACS per wt.% Mg. This
was also demonstrated in a separate study where a TiBor grain refiner was added to an
Al-Cu alloy. While grain refinement increased the hardness of the alloy, it was found
that the conductivity subsequently decreased (Fig. 2) [21]. This was to be expected as
both elements were dissolved into the Al matrix.
Secondary-phase morphology also affects the thermal conductivity of an alloy. The
effect of Si morphology on electrical conductivity was previously studied by modifying
the eutectic Si in B319 Al alloy through strontium (Sr) additions. It was found that the
synergy between solidification rate and Sr addition was linked to their shared roles in
eutectic silicon particle modification [22]. The results demonstrated that the electrical
conductivity can improve from approximately 27% to 30% IACS with decreasing Si
particle aspect ratio from approximately 2.4–1.5 or increasing sphericity from approxi-
mately 0.3–1.0. The finer, rounder, and less-elongated particles resulted in more efficient
electron transport through the material compared to an unmodified Al-Si alloy [22]. The
authors also noted that conductivity is diminished to a lesser extent by increasing the
relative amounts of eutectic Si-phase in the alloy matrix rather than by increasing the
level of Si in solid solution. Therefore, it has been concluded that there was clear indica-
tion of the dominant role of the Si modification level on the conductivity of 319 alloy.
The effect of grain refinement on conductivity has also been studied. For example,
Vasquez Lopez et al. [23] demonstrated through casting by unidirectional solidification
that as the solidification rate increased along the casting, dendritic refinement was
achieved. As a result, the increased refinement of dendrites along the casting corre-
sponded to an increase in thermal conductivity. Other studies have also shown that
dendritic refinement results in increased conductivity of Al alloys [24, 25]. Conversely,
Fig. 2. Electrical conductivity of HT200 alloys (Al-Cu) with and without TiBor grain refiner.
(Reprinted by permission from: Springer Nature, Andilab et al., copyright (2022) [21]).
480 P. Emadi et al.
it was found that while previous works determined that refinement by controlling cool-
ing rate can result in increased conductivity, attention should be paid to the associated
porosity formation during solidification as the presence of porosity can lead to reduced
conductivity.
Casting porosity (shrinkage or gas) is known to have an influence on the mechanical
properties of a material. Porosity has been reported to decrease thermal conductivity,
since it effectively impedes heat transfer by acting as scattering sites for thermal energy
carriers. This was demonstrated by Manzano Ramirez et al. [26], who introduced 0.43–
10.35 vol.% of gas porosity through hydrogen dissolution in a 380 Al alloy by increas-
ing the liquid melt temperature. As a result, the thermal conductivity decreased linearly
with increasing pore volume fraction. Similar results were found in a separate study,
where increases in percentage porosity (~1% to 8%) resulted in a decrease in thermal
conductivity (~130 to 80 W/mK) [27]. The results were correlated to the cooling rate of
the castings, where slower cooling resulted in increased porosity and consequently, a
decrease in conductivity.
3.2 Heat Treatment
Heat treatment is an extremely important part of alloy processing. It can improve the
as-cast properties of alloys for applications that require higher performance. Following
casting, post-processing usually begins with solution heat treatment. During the heat
treatment process, a wide variety of material properties can be achieved by controlling
the process parameters [28]. In addition to increased Cu dissolution during solution
treatment, many other microstructural changes can occur during the high-temperature
treatment of Al alloys. For example, acicular eutectic Si particles in hypoeutectic Al-Si
alloy systems have been found to fragment, spheroidize and coarsen during solution
heat treatment [29]. Further, the spheroidization process is known to occur more readily
in alloys containing fibrous Al-Si-eutectic structures, attained by prior chemical modi-
fication and/or increased solidification rates. Such modification of the microstructure,
through heat treatment and Sr additions can lead to improvements in conductivity. This
was demonstrated by Vandersluis et al. [30], where the electrical conductivity of A319
Al alloy was found to increase with increasing solution heat treatment time. The eutec-
tic Si spheroidization processes promoted subsequent conductivity increases by up to
10%, relative to the as-cast conditions (26.7–30.4% IACS).
Age hardening is a well-known method for improving mechanical properties, and it
is also known to improve the conductivity of Al alloys. It was found that artificial aging
promoted significant changes in electrical conductivity [31]. Castings both with and
without Sr were performed in molds preheated to either 500 or 200 °C. The samples
were subsequently solution heat treated for 24 h and age hardened at 250 °C. It was
found that the 0Sr-500C, 0Sr-200C, 150Sr-500C, and 150Sr-200C cast samples fea-
tured conductivities of approximately 33, 38, 35, and 38% IACS, respectively [31].
With increasing artificial aging time, electrical conductivity was consistently improved.
This treatment promoted the highest conductivity achieved in this research, which
corresponded to 25–30% improvements relative to their as-quenched levels. While
these maximum values in conductivity can be achieved, it must be considered that for
most practical purposes, an optimal aging time should be chosen for obtaining rela-
tively high levels of strength and conductivity.
4 Melt Cleanliness
As the efforts toward decreased GHGs and light weighting of components gains
momentum, increased use of Mg alloys is becoming important. However, the high mol-
ten reactivity of Mg at elevated temperatures is a barrier to its application in industry.
Therefore, particular attention must be paid to Mg melt cleanliness. According to Lun
Sin et al. [32], a primary factor in ensuring melt cleanliness is control of inclusions.
Inclusions in Mg alloys can be categorized by two groups, nonmetallic inclusions such
as oxides, chlorides, and nitrides as well as intermetallic inclusions such as Fe-bearing
phases [33]. Inclusions can be formed in several ways, the most common of which are
reactions with oxygen in air, turbulence effects during the casting process and interac-
tions with flux agents or protective atmospheres.
During melting and casting, molten Mg can react with oxygen in the air to form
magnesium oxide (MgO) [33, 34]. Oxide inclusions present themselves in different
morphologies, such as particles or films. Magnesium can also react with the available
nitrogen in air to form nitrides such as Mg3N2. As well, molten Mg can react with mois-
ture in the air or on casting equipment to form MgO and hydrogen. The presence of
these oxides and nitrides are unwanted as they are deleterious to the mechanical proper-
ties of Mg alloys. Further, such an occurrence is dangerous since it could cause a fire or
explosion.
During the casting process, unwanted inclusions can form and become entrained in
the final product. For example, improper mold design can lead to turbulence during fill-
ing [35]. Similarly, stirring using high RPM can break the surface of the molten metal
and trap oxides in the melt. In addition, since Mg is extremely reactive, during pouring,
an oxide film quickly surrounds the liquid metal. Care must be taken to pour in a steady
manner in order to prevent the oxide film from breaking [36, 37]. Therefore, proper
pouring technique, good mold design, and effective dross removal are important for
producing high-quality cast Mg components.
Fluxes and protective atmospheres are used to prevent melt oxidation during casting,
but, they can lead to unwanted inclusions as well [1, 38, 39]. During fluxing, if too
much material is used, it can remain on the melt surface and become trapped in the final
product [40]. Some examples of flux related inclusions are CaCl2, MgCl2, or FeB from
boride-containing fluxes. Due to these concerns, more attention is paid to protective
atmospheres for preventing oxidation. Some protective gases, including sulfur hexa-
fluoride (SF6) and sulfur dioxide (SO2), modify the oxide film on the liquid metal sur-
face, which suppresses Mg vaporization and removes reactive gases [41–43].
Alternatively, the use of steel mesh or ceramic foam filters can be effective in remov-
ing oxides already present in the melt and for controlling the melt velocity [44–46].
Another method that has demonstrated success in removing inclusions from the liquid
metal is sparging with inert gas such as argon [47]. Essentially, inclusions are floated to
482 P. Emadi et al.
the top of the melt by the gas bubbles and are skimmed away prior to pouring. Important
parameters for this technique are processing time, flowrate of gas, bubble size, and melt
temperature [47].
5 Processing
Recently, the treatment of liquid metal with high-intensity ultrasonic waves has gained
considerable attention by researchers. One of the most crucial aspects of ultrasonic
processing is cavitation, which refers to the formation of bubbles in the liquid metal that
can become filled with dissolved gasses [48]. During the treatment process, these cavi-
tation bubbles undergo expansion and compression stages. Ultimately, the bubbles will
collapse, producing pressure pulses, changes in temperature, and increased fluid veloc-
ity. Researchers have reported that cavitation can facilitate the wetting of inclusions for
enhanced nucleation, inclusion removal, melt degassing and secondary phase refine-
ment [49, 50].
As well as cavitation, streaming is also a principal factor in ultrasonic processing.
Ultrasonic processing can generate hydrodynamic flows in the liquid melt [49, 51, 52].
The flow can be characterized in two forms, first are acoustic streams that stem from the
vibratory motion of the ultrasonic horn/sonotrode. The second type are forced convec-
tion flows arising in the bulk fluid and in the vicinity of the crucible walls. The stream-
ing process is effective in distributing grain refiner and reinforcement particles in the
liquid melt.
A study performed by the authors examined the effects of varying ultrasonic pro-
cessing time on the microstructure and mechanical properties of an AZ91E Mg alloy
[53]. The authors applied the treatment for 60–240 s at a constant frequency of
20 ± 1 kHz and an amplitude of 30 μm. The treatment led to significant grain refine-
ment, with 180 s of treatment time, and the grain size decreased from 202 to 50 μm. The
authors observed that the secondary phases of the alloy were also refined. Additionally,
the mechanical properties of the refined alloys were also enhanced. The tensile strength,
yield strength, and %Elongation increased from 138 MPa, 95 MPa, and 1.35% in the
base alloy to 161 MPa, 111 MPa, and 2.2% respectively. The authors attributed the
improvements to grain refinement, decreased eutectic phase volume fraction, and
improved intermetallic distribution (Fig. 3). Therefore, ultrasonic treatment time was
concluded to have a significant impact on the microstructure and mechanical properties
of cast Mg alloys [53].
A similar study, but performed on an Al-Cu alloy, was performed by Li et al. [54].
The researchers used a 200 W sonication system with a frequency of 20 kHz. The treat-
ment time was 15 min, but the amplitude of vibration was not disclosed. Li et al. evalu-
ated the grain size of the resultant castings at three separate locations, the center, ½
radius, and the edge. They reported grain size reduction of 34%, 45%, and 42%, respec-
tively. The researchers also reported a 9% increase in tensile strength relative to the
base condition. The improvement strength was attributed to decreased shrinkage poros-
ity and increased dissolution of Cu in the matrix leading to solid solution strengthening.
Fig. 3. Backscattered electron micrographs of (a) as-cast alloy and (b) alloy sonicated for 240 s.
(Reprinted by permission from: Springer Nature, Emadi and Ravindran, copyright (2021) [53]).
Moreover, the authors also noted that increased Cu in solid solution led to reduction in
the formation of coarse eutectic phases.
6 Alloying and Modification
As mentioned previously, materials such as Al or Mg are seen as potential candidates to

replace steel components, as they have significantly lower density [55]. However, Al
and Mg display reduced strength compared to steel. Therefore, they must be alloyed
with other materials to increase their mechanical properties to a level appropriate for
industrial application.
One method used to improve the strength of Al alloys is to modify the morphology
of secondary phases that are detrimental to the mechanical properties of the material. In
automotive-grade Al alloys (typically the Al-Si-X series), the silicon eutectic phases
form as coarse flakes with sharp features. These phases can act as stress risers and crack
initiation sites. Researchers have reported that reducing the size of these phases and
altering their morphology to a more fibrous form can have a significant impact on the
mechanical properties of cast alloys [56]. As demonstrated by Vandersluis et al. [56],
adding elements such as Sr to molten Al alloys can provide the desired modification.
Strontium content was found to significantly reduce the growth temperature of the
Si-eutectic phases from 577 °C in the unmodified state to 551 °C with 30 ppm Sr. At a
solidification rate of 5.89 °C/s and a Sr content of 40–50 ppm, the diameter of the
Si-eutectic particles decreased from 10.0 to 3.6 μm. As the Sr content increased the
morphology of the Si phases was transformed to a fine fibrous shape as seen in Fig. 4.
Similar studies have been performed with the aim of achieving Si modification.
Researchers have experimented with elements such as Ytterbium (Yb) and Europium
(Eu). Nogita et al. [57] reported that Yb modification can result in a fine-plate-like
structure which was more evenly distributed than the unmodified alloy. However, in the
microstructure, the Si phase did not have the fibrous morphology one would observe if
sodium or Sr were the modifying additive. The authors noted that Yb was not detected
484 P. Emadi et al.
Fig. 4. Optical micrographs of the eutectic Si morphology for casting conditions: (a) 2 ppm Sr
5.89 °C/s and (b) 154 ppm Sr 5.89 °C/s. (Reprinted by permission from: Springer Nature, Vandersluis
et al., copyright, (2020) [56]).
in the Si phases, which could explain why the desired eutectic modification was not
achieved. Europium at an addition level of 2400 ppm showed a morphology more
closely related to the ideal fibrous structure. It was suggested by Li et al. [58] that Eu
may remove or deplete the Si-eutectic nucleation sites which acs as a hindrance to the
growth of Si-eutectic phases.
Cooling rates are also an important factor that can impact the effectiveness of alloy-
ing elements on modifying Si-eutectic phases. It has been shown that at higher cooling
rates, modifiers are more effective, since they are able to increase the angle of branch-
ing and twinning frequency [59]. This leads to an increase in the presence of Si-eutectic
twins and reduces the size of the eutectic particles, leading to a more fibrous morphol-
ogy and increased mechanical properties [59]. To produce Al-Si castings with increased
mechanical properties, it is often recommended to determine the optimum cooling rate,
in addition to using a modifier.
Alloying Al alloys with different elements has shown to be effective in modifying
the Si-eutectic phase of the alloy. A potential method for improving mechanical proper-
ties is using low-density elements, such as lithium (Li), to modify the Si eutectic. As
demonstrated by Chen et al. [60], the addition of 0.3 wt.% Li to an A356 alloy modified
the Si-eutectic phase into small spherical phases that were more evenly distributed in
the microstructure. The modification provided by the Li, combined with a reported
reduction in Al alloy density when Li is added, would further improve the strength to
weight ratio of the Al alloy thus, making it even more attractive as an alternative to
heavy steel alloys in the automotive sector [61].
7 Future of Materials Innovation
The discovery and development of new materials is one of the driving forces of techno-
logical advancement. Materials innovation is currently at the forefront of solving soci-
etal challenges such as climate change. However, development of new materials still
heavily relies on significant experimental work, involving trial and error, where repeated
experiments are required to achieve statistical significance. Materials design is complex
and involves multiple variables that are difficult to control. Hence, a large amount of
data is needed to make informed decisions regarding material design and selection.
Materials and computer scientists across the globe have collaborated and launched a
project in 2011, known as the Materials Genome Initiative [62]. This initiative involves
the development and execution of proven computational techniques to predict, screen,
and optimize materials at a high-scale rate. Through this process, material properties
can be predicted and optimized by solving equations based on the fundamental laws of
physics and quantum mechanics. An example of Computational Computer Science/
Materials Genome can be seen in the work of Zhu et al. [63], where computational
methods were used to optimize the alloy chemistry of a nickel-based super alloy, which
are commonly used for turbine blades. However, optimizing the alloy chemistry is dif-
ficult due to the large number of alloying elements.
Due to the considerable number of possible combinations, a coarse iteration of the
data was performed, where the variation in possible alloy compositions was large. This
was done to identify the potential region where the optimal composition could be
achieved. A second iteration, where the alloy composition was bound within the region
determined from the first iteration was then performed. In the second iteration, the
composition variation was reduced to determine the optimal composition. This enabled
the authors to determine the optimal alloy composition with the lowest consumption of
computing power, resulting in a faster simulation. The authors were able to search
through 357,636 possible alloy compositions and analyze the combinations in terms of
creep resistance, oxidation resistance, density, and cost [63]. As a result, potential alloy
compositions can be selected from the computationally generated results and tested
experimentally.
One of the challenges in the use of phenomenological approaches for calculating/
predicting properties of multicomponent materials systems is high-throughput assess-
ment of databases. An integral component of this method is the ability to produce high
quantities of simulated material properties. This should be followed by experimentation
that validates the results of the simulations. Additionally, open-access databases that
combine various experimental results can aid in improving model predictions.
Therefore, the next step is to leverage open-source development and interactive inter-
net- or application-based technologies to enable user contributions throughout the sci-
entific community.
8 Conclusion
It is clear that sustainability has become the central theme in materials science and
engineering. Not only are researchers contemplating ways to improve the strength and
in-service performance of materials, but they are also actively considering the aspect of
environmental impact. Lightweighting is the focal point of these endeavors. The main
challenge is developing lightweight materials that can outperform current alloys. In
other words, scientists strive to reduce the environmental impact of materials and
simultaneously ensure similar or improved performance. Increasing our knowledge of
alloying, grain refinement and factors that affect thermal properties is crucial.
486 P. Emadi et al.
Researchers must leverage modern manufacturing techniques and technologies that aid
in improving material properties. As well, the development and execution of computa-
tional methods to predict and optimize materials at a rapid rate. To this end, efforts will
continue to encourage tailor-made materials and processes with a sustainable future in
mind.
Acknowledgments The authors would like to acknowledge the work and dedication of the scholars
whose research was used in this review. We would like to thank the alumni and members of the Centre
for Near-net-shape Processing of materials for their research and stimulating discussions. The authors
would also like to acknowledge the financial support of the Natural Sciences and Engineering Research
Council of Canada (NSERC), through Canada Graduate Scholarships to both Mr. Payam Emadi
(CGSD3 – 535728 – 2019) and Mr. Bernoulli Andilab (CGSD3 - 559982 - 2021), and research grant
RGPIN-06096.
References
1. Avedesian M, Baker H. Magnesium and magnesium alloys. Materials Park: ASM International; 1999.
2. StJohn D, Ma Q, Easton M, Cao P, Hildebrand Z. Grain refinement of magnesium alloys. Metall
Mater Trans A. 2005;36:1669–79.
3. Perner A, Vetter J. Lithium-ion batteries for hybrid electric vehicles and battery electric vehicles.
In: Advances in battery technologies for electric vehicles. Woodhead Publishing; 2015. p. 173–90.
4. Jian X, Geer T, Meek T, Han Q. Effect of power ultrasound on grain refinement of magnesium
AM60B alloy. In: TMS Annual Meeting Magnesium Technology. San Antonio; 2006.
5. Gao Q, Wu S, Lü S, Xiong X, Du R, An P. Effects of ultrasonic vibration treatment on particles dis-
tribution of TiB2 particles reinforced aluminum composites. Mater Sci Eng A. 2017;680:437–43.
6. Atamanenko T, Eskin D, Zhang L, Katgerman L. Criteria of grain refinement induced by ultrasonic
melt treatment of aluminum alloys containing Zr and Ti. Metall Mater Trans A. 2010;41:2056–66.
7. Brandes A, Brook B. Smithells metals reference book. Oxford: Butterworth-Heinmann; 1992.
8. Lide R. CRC handbook of chemistry and physics. Boca Raton: CRC Press; 2001.
9. Flemings M. Solidification processing. New York: McGraw-Hill; 1974.
10. Limmaneevichitr C, Eidhed W. Fading mechanism of grain refinement of aluminum silicon alloy
with Al–Ti–B grain refiners. Mater Sci Eng A. 2003;349:197–206.
11. Schaffer P, Dahle K. Settling behaviour of different grain refiners in aluminium. Mater Sci Eng
A. 2005;413:373–8.
12. Fu H, Zhang M, Qiu D, Kelly P, Taylor J. Grain refinement by AlN particles in Mg–Al based
alloys. J Alloys Compd. 2009;478:809–12.
13. Cao G, Choi H, Oportus J, Konishi H, Li X. Study on tensile properties and microstructure of cast
AZ91D/AlN nanocomposites. Mater Sci Eng A. 2008;494:127–31.
14. Wang S, Huang Y, Yang L, Zeng Y, Hu Y, Zhang X, Sun Q, Shi S, Meng G, Hort N. Microstructure
and mechanical properties of Mg-3Sn-1Ca reinforced with AlN nano-particles. J Magnes Alloy.
2021. https://doi.org/10.1016/j.jma.2021.04.002.
15. Emadi P, Rinaldi M, Ravindran C. Grain refinement and fading behavior of MgB2-inoculated magne-
sium. Metallogr Microstruct Anal. 2021;10:367–76. https://doi.org/10.1007/s13632-021-00755-5.
16. Klösch G, McKay B, Schumacher P. Preliminary investigation on the grain refinement behavior of
ZrB2 particles in Mg-Al alloys. Cham: Springer; 2016.
17. Sahoo S, Sahoo B, Panigrahi S. Effect of in-situ sub-micron sized TiB2 reinforcement on micro-
structure and mechanical properties in ZE41 magnesium matrix composites. Mater Sci Eng
A. 2020;773:138883.
18. Elsayed A, Ravindran C. Effect of aluminum-titanium-boron based grain refiners on AZ91E mag-
nesium alloy grain size and microstructure. Int J Met. 2011;5:29–41.
19. Emadi P, Andilab B, Ravindran C. Processing and properties of magnesium-based composites

reinforced with low levels of Al2O3. Inter Metalcast. 2022;16:1680–92. https://doi.org/10.1007/
s40962-021-00738-w.
20. Mulazimoglu M, Drew R, Gruzleski J. Electrical conductivity of aluminium-rich Al-Si-Mg alloys.
J Mater Sci Lett. 1989;8:297–300.
21. Andilab B, Vandersluis E, Emadi P, Ravindran C, Byczynski G, Fernández-Gutiérrez
R. Characterization of a cast Al-Cu alloy for automotive cylinder head applications. J Mater Eng
Perform. 2022;31:1–10.
22. Vandersluis E, Emadi P, Andilab B, Ravindran C. The role of silicon morphology in the electri-
cal conductivity and mechanical properties of as-cast B319 aluminum alloy. Metall Mater Trans
A. 2020;51A:1874–86.
23. Vázquez-López C, Calderón A, Rodríguez ME, Velasco E, Cano S, Colás R, Valtierra S. Influence
of dendrite arm spacing on the thermal conductivity of an aluminum-silicon casting alloy. J Mater
Res. 2000;15:85–91.
24. Argo D, Gruzleski DR, Gruzleski J. A simple electrical conductivity technique for measurement of
modification and dendrite arm spacing in Al-Si alloys. AFS Trans. 1987;95:455–64.
25. Narayan Prabhu K, Ravishankar B. Effect of modification melt treatment on casting/chill interfa-
cial heat transfer and electrical conductivity of Al–13% Si alloy. Mater Sci Eng A. 2003;360:293–8.
26. Ramirez AM, Beltrán FE, Yáñez-Limón JM, Vorobiev YV, Gonzalez-Hernandez J, Hallen
JM. Effects of porosity on the thermal properties of a 380-aluminum alloy. J Mater Res.
1999;14:3901–6.
27. Vandersluis E, Ravindran C. The role of porosity in reducing the thermal conductivity of B319 Al
alloy with decreasing solidification rate. JOM. 2019;71:2072–7.
28. Sjölander E, Seifeddine S. The heat treatment of Al–Si–cu–mg casting alloys. J Mater Process
Technol. 2010;210:1249–59.
29. Wang R, Lu W. Spheroidization of eutectic silicon in direct-electrolytic Al-Si alloy. Metall Mater
Trans A. 2013;44:2799–809.
30. Vandersluis E, Ravindran C. Effects of solution heat treatment time on the as-quenched micro-
structure, hardness and electrical conductivity of B319 aluminum alloy. J Alloys Compd.
2020;838:155577.
31. Vandersluis E, Ravindran C, Bamberger M. Mechanisms affecting hardness and electrical conduc-
tivity in artificially-aged B319 aluminum alloy. J Alloys Compd. 2021;867:159121.
32. Sin SL, Elsayed A, Ravindran C. Inclusions in magnesium and its alloys: a review. Int Mater Rev.
2013;58:419–36.
33. Hu H, Luo A. Inclusions in molten magnesium and potential assessment techniques.
JOM. 1996;48:47–51.
34. Bakke P, Karlsen DO. Inclusion assessment in magnesium and magnesium base alloys. In: Proc
1997 SAE Int Cong & Expo, Detroit; 1997.
35. Emley E. Principles of magnesium technology. Oxford: Pergamon Press; 1966.
36. Griffiths W, Lai N. Double oxide film defects in cast magnesium alloy. Metall Mater Trans
A. 2007;38:190–6.
37. Mirak AR, Divandari M, Boutorabi SMA, Taylor JA. Effect of oxide film defects generated during
mould filling on mechanical strength and reliability of magnesium alloy castings (AZ91). Int J Cast
Met Res. 2012;25:188–94.
38. Cashion S, Ricketts N, Hayes P. The mechanism of protection of molten magnesium by cover gas
mixtures containing sulphur hexafluoride. J Light Met. 2002;2:43–7.
39. Cashion S, Ricketts N, Hayes P. Characterisation of protective surface films formed on molten
magnesium protected by air/SF6 atmospheres. J Light Met. 2002;2:37–42.
40. Friedrich HE, Mordike BL. Magnesium technology. Berlin/Heidelberg: Springer-Verlag; 2006.
41. Chen H, Liu J, Huang W. Oxidation behavior of molten magnesium in air/HFC-134a atmospheres.
J Mater Sci. 2006;41:8017–24.
42. Ha W, Kim Y. Effects of cover gases on melt protection of Mg alloys. J Alloys Compd.
2006;28:208–13.
43. Xiong S, Liu X. Microstructure, composition, and depth analysis of surface films formed on mol-
ten AZ91D alloy under protection of SF6 mixtures. Metall Mater Trans A. 2007;38:428–34.
488 P. Emadi et al.
44. Bakke P, Engh TA, Bathen E, Oymo D, Nordmark A. Magnesium filtration with ceramic foam
filters and subsequent quantitative microscopy of the filters. Mater Manuf Process. 1994;9:111–38.
45. Le C, Zhang ZQ, Cui JZ, Chang SW. Study on the filtering purification of AZ91 magnesium alloy.
Mater Sci Forum. 2009;610-613:754–7.
46. Wang J, Zhou J, Tong W, Yang Y. Effect of purification treatment on properties of Mg-Gd-Y-Zr
alloy. Trans Nonferrous Met Soc China. 2010;20:1235–9.
47. Yim CD, Wu G, You BS. Effect of gas bubbling on tensile elongation of gravity mold castings of
magnesium alloy. Mater Trans. 2007;48:2778–81.
48. Eskin GI. Ultrasonic treatment of light alloy melts. Amsterdam: Gordon & Breach; 1998.
49. Li Y, Feng H, Cao F, Chen Y, Gong L. Effect of high density ultrasonic on the microstructure and
refining property of Al–5Ti–0.25C grain refiner alloy. Mater Sci Eng A. 2008;487:518–23.
50. Liu X, Osawa Y, Takamori S, Mukai T. Microstructure and mechanical properties of AZ91 alloy
produced with ultrasonic vibration. Mater Sci Eng A. 2008;487:120–3.
51. Gao D, Li Z, Han Q, Zhai Q. Effect of ultrasonic power on microstructure and mechanical proper-
ties of AZ91 alloy. Mater Sci Eng A. 2009;502:2–5.
52. Zhang S, Zhao Y, Cheng X, Chen G, Dai Q. High-energy ultrasonic field effects on the microstruc-
ture and mechanical behaviors of A356 alloy. J Alloys Compd. 2009;470:168–72.
53. Emadi P, Ravindran C. The influence of high temperature ultrasonic processing time on the
microstructure and mechanical properties AZ91E magnesium alloy. J Mater Eng Perform.
2021;30:1188–99.
54. Li R, Dong F, Zhang Y, Chen P, Li X. Eutectic phase characterization and mechanical properties of
Al-Cu alloy ingot solidified with ultrasonic treatment. Materials. 2022;15:1067.
55. Ahmed B, Wu SJ. Aluminum lithium alloys (Al-Li-Cu-X)-new generation material for aerospace
applications. Appl Mech Mater. 2013;440:104–11.
56. Vandersluis E, Prabaharan N, Ravindran C. Solidification rate and the partial modification of 319
aluminum with strontium. Int J Met. 2020;14:37–46.
57. Nogita K, Yasuda H, Yoshiya M, McDonald S, Uesugi K, Suzuki Y, Takeuchi A. The role of trace
element segregation in the eutectic modification of hypoeutectic Al-Si alloys. J Alloys Compd.
2010;489:415–20.
58. Li JH, Wang XD, Ludwig TH, Tsunekawa Y, Arnberg L, Jiang JZ, Schumacher P. Modification of
eutectic Si in Al-Si alloys with Eu additions. Acta Mater. 2015;84:153.
59. Hegde S, Prabhu K. Modification of eutectic silicon in Al-Si alloys. J Mater Sci. 2008;43:3009–27.
60. Chen ZW, Ma CY, Chen P. Eutectic modification of A356 alloy with Li addition through DSC and
Miedema model. Trans Nonferrous Met Soc China. 2012;22:42–6.
61. Zeka B, Markoli B, Mrvar P. Production and investigation of new cast aluminium alloy with lith-
ium addition. Mater Geoenviron. 2020;67:13–9.
62. The materials genome initiative. https://www.mgi.gov/. Accessed 14 Feb 2022.
63. Zhu Z, Höglund L, Larsson H, Reed RC. Isolation of optimal compositions of single crystal super-
alloys by mapping of material’s genome. Acta Mater. 2015;90:330–43.
Characterization of the Microstructure of
Al, Mg and Ti Alloys with State-of-the-Art Electron
Microscopy
Raynald Gauvin1 (*), Nicolas Brodusch1, and Stéphanie Bessette1

1
Department of Mining and Materials Engineering, McGill University, Montreal, QC, Canada
raynald.gauvin@mcgill.ca
Abstract The development of light metals, like Al, Mg and Ti alloys, for the transporta-
tion industry relies on controlling the size, distribution and volume fraction of nano-sized
precipitated to improve their mechanical properties. These mechanical properties and
corrosion resistance are also related to the size and texture of the grains of these alloys,
among other factors. Electron microscopy is a key technique to perform such character-
ization and this work will present state-of-the-art results with field emission low-voltage
scanning electron microscopy with EDS and EBSD and with scanning transmission elec-
tron microscopy at 30 keV with a unique electron microscope that has a 0.15 nm spatial
resolution with EDS and EELS capabilities, the SU-9000 from Hitachi [1]. Three-
dimensional electron microscopy imaging of these alloys will be presented with 3D BSE,
EDS and EBSD maps acquired with a state-of-the-art three column focus ion beam, the
Ethos NX-5000 from Hitachi.
1 Research Objectives and Methodology
The development of light metals, like Al, Mg and Ti alloys, for the transportation indus-
try relies on controlling the size, distribution and volume fraction of nano-sized precipi-
tated to improve their mechanical properties. These mechanical properties and corrosion
resistance are also related to the size and texture of the grains of these alloys, among
other factors. Electron microscopy is a key technique to perform such characterization
and this work will present state-of-the-art results with field emission low-voltage scan-
ning electron microscopy with EDS and EBSD and with scanning transmission electron
microscopy at 30 keV with a unique electron microscope that has a 0.15 nm spatial
resolution with EDS and EELS capabilities, the SU-9000 from Hitachi [1]. Three-
dimensional electron microscopy imaging of these alloys will be presented with 3D
BSE, EDS and EBSD maps acquired with a state-of-the-art three column focus ion
beam, the Ethos NX-5000 from Hitachi.
As an example of our characterization capabilities, Fig. 1 shows analytical results
obtained with the Hitachi SU-9000 on a thin specimen of a Al-Cu-Li alloy where lattice
https://doi.org/10.1007/978-3-031-17425-4_63
490 R. Gauvin et al.
Fig. 1. Analytical characterization of n AA2099 Al-Cu-Li alloy with a SU9000 CFE-SEM at 30 kV:
(a) Lattice bright-field image of δ’ a (Al3Li) precipitate with the lattice spacing measurement in (b);
HAADF image of the fine microstructure showing T1 (Al2CuLi) precipitates and experimental ele-
mental map obtained with the jump ratio method by LV-EELS.
Fig. 2. Li EELS edge-to-background ratio map of an Al 2099 alloy taken with the Hitachi SU-9000 at
30 keV.
Characterization of the Microstructure of Al, Mg and Ti Alloys with State-of-the-Art… 491
imaging of the fine δ’ (Al3Li) is resolved (a, b) and an elemental Cu map obtained by
the jump ratio method with the corresponding high-angle annular dark-field (HAADF)
image (c, d), where the T1 plates (Al2CuLi) are clearly evidenced.
Figure 2 shows a Li EELS [2] edge-to-background ratio map of a thin film of the
same Al-Cu-Li 2099 alloy recorded with the three-detectors system on the Hitachi
SU-9000 at 30 keV. This map shows the round δ’ precipitates (Al3Li) of 5–20 nm and
the T1 plates (AlCuLi) that have thicknesses between 1 and 2 nm. The fact that the
edges are always ionized is a strong advantage for EELS since the electron transition
leading to the emission of an X-ray is not necessary to observe the ionization event.
Also, the emission rate is greater than about 10,000 owing to the fluorescence yield.
The downside of EELS is the need for a very thin transparent specimen and beam dam-
age can also occur at 30 kV.
2 Conclusions
It is possible to characterize light metallic alloys with nanometer statial resolution using
scanning transmission electron microscopy (STEM) with transparent specimens at
30 keV. This opens the road for easier and less expensive characterization of these
alloys because typically, such chatacterization is performed at 200 keV with much
expensive electron microscopes.
Acknowledgements The Natural Sciences and Engineering Research Council of Canada (NSERC)
and the Canadian Foundation for Innovation (CFI) are acknowledged for financial support.
References
1. Brodusch N, Demers H, Gellé A, Moores A, Gauvin R. Energy electron-loss spectroscopy (EELS)

with a cold-field emission electron microscope at low accelerating voltage in transmission mode.
Ultramicroscopy. 2019;203:1–36.
2. Egerton RF. Electron energy-loss spectroscopy in the electron microscope. 3rd ed. New York:
Springer; 2011.
Self-Piercing Riveting of High-Ductility Al–Zn–Mg
Casting Alloy (Nemalloy HE700) in F Temper
Yunsong Guo1 (*), Sumanth Shankar1, Mukesh K. Jain1, Laura Baenen2,

Anthony Lombardi3, and Glenn Byczynski3
1
McMaster University, Department of Mechanical Engineering, Hamilton, ON, Canada
guo37@mcmaster.ca
2
Bollhoff Inc., Troy, MI, USA
3
Nemak USA/Canada, Windsor, ON, Canada
Abstract. Nemalloy HE700 (high elongation) automotive Al alloy, a newly developed

Al–Fe hypoeutectic-based alloy for high-vacuum high-pressure die casting, provides
higher strength and elongation even without heat treatment compared to current main-
stream structural automotive Al casting alloy. This study aims to systematically investi-
gate the application of HE700 in the field of self-piercing riveting (SPR): the joinability
of HE700 castings with dissimilar materials, directed by automotive industry standards.
Several rounds of exploratory experiments amply showed the feasibility of obtaining
valid SPR joints of HE700 castings (2.5 mm thick) with Al 6082 alloy (2.5 mm) and
DP600 steel (1.2 mm) sheets with acceptable automotive quality standards. The influence
of the die shape, rivet geometry and maximum punch force were the critical process
parameters investigated. Finally, this article endorsed the viability of significant automo-
tive lightweighting with a lower carbon footprint while integrating novel casting alloys
such as HE700 into the automotive superstructure through economical joining processes
such as SPR.
Keywords: High-Pressure Die Casting · Aluminium Alloys · Self-Piercing

Riveting · Lightweighting · Mechanical Joinability · Process Parameters
1 Introduction
Over the last decade, the adoption of aluminium alloys in the automotive sector has
increased significantly, driven by the rising need for lightweight battery electric vehi-
cles (BEVs) as a result of the global climate policies and actions [1]. Replacing steel
with aluminium alloys and increasing the proportion of aluminium alloys in the body of
the automobile leads to improvements in energy efficiency and travel range, thus, eas-
ing the mileage anxiety and making the BEV an attractive proposition for consumers [1,
2]. Compared to other means of increasing the driving range, such as increasing battery
energy density and/or motor efficiency, the development costs of utilizing aluminium
structural components for significantly reducing the curb weight of an automobile
would be relatively low [3].
https://doi.org/10.1007/978-3-031-17425-4_64
494 Y. Guo et al.
The Nemalloy HE700 is a novel Al structural alloy developed for the near-net shape
manufacturing of automotive castings using the high-vacuum high-pressure die casting
(HVHPDC) process. The HE700 Al alloy is an Al–Fe hypoeutectic alloy with elemental
additions of Zn and Mg for the solid solution and precipitation strengthening of the
casting. As a new family of Al alloys based on the Al–Fe-based eutectic alloy system,
the HE700 castings offer an excellent combination of properties for lightweight vehicle
industries such as high ductility and sufficient yield strength even in the F temper [4–6].
Figure 1a shows a representative engineering stress–strain curve of HE700 at room
temperature, which was obtained after 103 days of natural ageing at ambient condi-
tions. The typical yield stress (YS) (0.2% Proof Stress) and ultimate tensile stress are
approximately 140 and 290 MPa, respectively, along with a typical elongation ranging
between 10% and 13% in the F temper. The self-piercing riveting (SPR) process was
carried out around three to five months after the date of casting the plates for this study.
Figure 1 shows typical photos of the X-ray imaging and microstructure in the cross-
section of the castings.
Due to the high strain rate and large deformation features of the SPR process, the
joinability and joint performance depend on the material toughness which is a combina-
tion of having a high ultimate tensile stress and elongation at fracture. However, with
the presence of pores in the general die castings, surface blisters are generated during
heat treatment, resulting in dimensional instability. Therefore, 80% of automotive die-
casting alloy parts are difficult to improve their mechanical properties through heat
treatment in the automotive industry [7]. Typically, die-cast plates in the SPR process
are prone to a high tendency of cracking in the joint button due to the relatively low
ductility of the material.
Structural plates of HE700 cast using HVHPDC showed notably higher strength and
elongation than the conventional Al–Si–Mg alloy plates in Aural 5 (F and T5 tempers),
Fig. 1. Typical overview of the structural casting plates from Nemalloy HE700 and Aural 5 alloys,
including (a) comparison of typical engineering stress–strain curves at room temperature, (b) X-ray
images of typical casting shot and (c) high-magnification microstructure of a typical casting cross-
section [8].
Self-Piercing Riveting of High-Ductility Al–Zn–Mg Casting Alloy (Nemalloy HE700… 495
as shown in Fig. 1. For further understanding of the casting parameters, microstructure

and mechanical properties, please refer to the recent publications [4, 8, 9].
Within the technological advancements in structural automotive lightweighting,
joining aluminium alloy castings with other wrought Al alloy and steel counterparts is
a critical challenge that is being studied extensively around the world, presently [1, 2].
However, due to the characteristics of aluminium alloy shape castings, joining by tradi-
tional welding processes is energy-consuming and further leads to unwanted defects
such as distortion, cracking, oxidation and porosity. Consequently, as an energy-
efficient joining process, self-piercing riveting has had a significant penetration into the
automotive market in the past decade, as a multi-material joining solution for Al alloys
and steel components [10, 11]. The mechanical behaviour of SPR joints has shown that
joints can provide good impact resistance and fatigue properties compared to spot-
welded joints and that SPR joints are less sensitive to corrosion than welded joints [12,
13]; motivating the continued development of SPR in the automotive industry. The abil-
ity to manufacture Al cast components that could lend themselves to the SPR joining
process is relatively new [14]. This study explores the feasibility of using SPR to join
the Al cast plates in the as-cast (F temper) with the Al 6082 (T6 Temper) [15–17] and
DP600 high-strength steel [14, 18] plates, respectively.
2 Experiments
The HE700 Al alloy has demonstrated excellent mechanical properties as an HVHPDC

structural aluminium alloy [4, 8, 9, 19], yet, severe plastic deformation under high
strain rate conditions experienced in the SPR process still poses a challenge to obtain
an optimized process condition for automotive integration. The specimens for SPR
experiment trials were obtained from the 2.5-mm-thick section of the HE700 alloy plate
castings [8], as shown in Fig. 2. The HE700 plate castings were produced on a Bühler
Carat-type 1200-ton high-vacuum high-pressure die-casting machine assisted by a
Wohlin lubricant sprayer and an ABB six-axis robot that was used for the casting opera-
tions. Figure 2 shows the typical specimen of a 30 mm × 120 mm rectangular section,
used for the SPR trials.
The selections of the material, dimension and stacking order of the sheets are shown
in Table 1. The study revolved around examining the effects of the critical independent
parameters such as the rivet geometry, die design and maximum loading force on the
quality and performance of the joint. The exploratory experimental trials involved inde-
pendent parameters such as a few levels of rivet-die combination set designs and a few
maximum load settings (48–78 kN) for the punch.
Several SPR trials were carried out by using the electrical–hydraulic hybrid RIVSET
tool with a skeletal C-frame and a 78 kN maximum joining force, as shown in Fig. 3.
The entire riveting process lasted less than one (1) second with a displacement-controlled
punch pushing the rivet downward, piercing the two metal sheets. Typically, the even-
tual displacement of the punch was determined empirically by the length of the rivet,
the thickness of the sheets and the preset maximum punch force; but, in this case, the
preset maximum punch force was considered as one of the independent variables to
496 Y. Guo et al.
Fig. 2. Photograph of a typical casting shot along with the schematic of the plate castings showing the
four SPR specimens within the 2.5-mm-thick plate region and the spots used for the joining within
each specimen.
Table 1. Three unique combinations of materials, dimensions and stacking order of sheets in SPR.
Sheet thickness Bottom Sheet thickness
Combinations Top sheet (mm) sheet (mm)
C1/C1Ra 6082 – T6 (Extrusion) 2.5 HE700-F 2.5
C2/C2Ra DP600 1.2 HE700-F 2.5
Steel – GI50/50 U
R repeatability experiment
a
Fig. 3. Photograph of the C-frame RIVSET apparatus for SPR and drawings of typical cross-sections
of the die and rivet.
adequately characterize the effect of the punch load on the riveting outcome [20].
Figure 3 shows the typical rivet shape and two different types of dies (ring groove and
standard die sets). The rivets were made of high-strength boron steel with an ultra-high
tensile strength of 1600 MPa and a hardness of HV480 [21]. The nominal dimensions
of the rivets were 5.3 mm in diameter and 4–7 mm long. The ring groove die was devel-
oped specifically to join low ductility materials such as cast Al alloys using SPR. The
encircling ring groove provided support for the rivet during the joining and assisted in
the spread of the rivet leg. A concavity in the center of the ring groove die allowed lim-
ited material flow into the die and mitigated rivet upsetting and buckling [22]. The
advantages of using the novel ring groove die over the standard flat bottom dies were in
alleviating button-side cracks, improving interlocking of rivets and reducing punch
loads.
The dependent parameters were the quality of the joint as mandated by the visual
inspection for a crack-free joint and an inspection of the cross-section of the joint to
ascertain that the four critical parameters were within unique specifications limits, as
shown in Table 2.
Additional dependent parameters (outcome) that determine a valid SPR joint along
with the method of determination are listed below [23].
• Symmetrical Deformation (Visual inspection),
• A crack-free below the rivet head (Visual inspection),
• A full contact between the rivet head and the base material (no gaps) (inspection of
Sectioned Cross-section),
• A crack-free and uniform deformation of the bottom base metal (visual inspection
and inspection of sectioned cross-section).
3 Results
Figure 4 presents a few typical high-quality SPR joints of HE700+6082 and

HE700+DP600 sets, respectively. The values of mechanical interlock and other critical
dimensions are also shown in the respective cross-section images of the joints in Fig. 4.
These advantages are a testament to the excellent formability and mechanical perfor-
mance of HE700. However, inappropriate setup and combinations could lead to unsat-
isfactory results. Figure 4d shows a typical failed joint between HE700 and 6082 alloys
wherein a very low interlocking values of 0.1 and 0.12 on each side, respectively, would
not provide sufficient joint strength; further, radial cracks appear on the joint button
(circled in red), reducing the fatigue life and corrosion resistance [24]. Furthermore, in
order to further investigate the factors influencing the riveting quality, the load–dis-
placement curves under various preset maximum loads were extracted, as shown in
Fig. 4; the comparison of the curves between a valid and failed joint suggests the sensi-
tivity of the process to the subtle variations in the force–displacement curves. Close
inspection of the cross-sections of the joints affirmed the ease of forming valid SPR
joints with the HE700 plates of 2.5 mm in F Temper (as-cast) condition.
A trial of approximately 60 experiments was conducted using the combinations C1R
and C2R (Table 1), where R denotes the use of repeatability in the design of experi-
ments. These explorations yielded optimized and repeatable levels of critical operating
parameters such as the maximum force of the punch, and the design of the die and rivet.
In the repeatability experiments, a high success rate was achieved, which demonstrated
the excellent repeatability of the results and permit the feasibility of SPR technology for
this material for large-scale industrial applications. These experiments were all carried
out with the HE700 alloy castings in the F temper condition and compared with the
498 Y. Guo et al.
Table 2. Typical specifications for acceptable SPR joints in the automotive industry. The critical
features that require stringent dimensional control are shown in Fig. 4h.
Undercut in button Remaining thickness Rivet head Rivet head

side of material of button side recess projection
Ford Specification: ≥0.20 ≥0.2 <0.3 <0.5
ESBU5A-1B279-AA
(mm)
BMW Specification: ≥0.15 ≥0.1 <0.3 <0.5
GS 96001-2 2010-03
(mm)
SPR joints of Aural 5 (T5 temper) castings (substituting the HE700 plates in Table 1)
[8] as the benchmark.
This work adequately established the feasibility of carrying out successful SPR join-
ing operations on the as-cast plates of Nemalloy HE700 (2.5 mm) to both the Al 6082
(T6) (2.5 mm) and DP600 (1.2 mm) steel sheets. A thorough exploration of several
experiments enabled narrowing the optimal operational window of critical parameters
for a successful joint. The purpose of the work was to comprehensively investigate the
effects of rivet geometry and die type, and maximum load force on the joining and joint
performance, and to analyse the deformation behaviours of the base material (HE700)
during the riveting process. The success of these feasibility trials provided the confi-
dence to continue developing numerical models and property/performance datasets for
the HE700 alloy castings to enable valid simulations of the SPR process and joint per-
formance through finite element analyses and modelling.
4 Conclusions
This study amply demonstrated the feasibility, reliability and repeatability of obtaining
valid SPR joints between HE700 alloy castings (2.5 mm thick) and each of Al 6082
alloy sheets (2.5 mm thick) and DP600 steel sheets (1.2 mm thick), respectively. The
influence of the die type has been investigated, a novel ring groove die was employed
to mitigate cracking tendencies on the button side. In the repeatability experiments,
standard flat dies and ring groove dies were employed, and deformation behaviours of
the bottom sheet (HE700) during the riveting process have thus been investigated. The
results showed that owing to the reasonably high ductility of HE700, it was possible to
obtain satisfactory results even with a standard die. The effects of the critical indepen-
dent parameters such as the rivet shape, die geometry, maximum force, and sheet thick-
ness have been investigated. These experiment trails have provided the necessary
validation to continue developing numerical models and property/performance datasets
Fig. 4. Visual quality assessment of typical SPR joints, showing valid joints of HE700+6082 (a, b)
and HE700+DP600 (e, f), and failed joints (c, d), failed due to a low value of rivet interlocking and
radial cracks on the joint button. A typical force–displacement curve (g) for valid and failed SPR joints
is also presented. (h) shows the critical features in the cross-section image that define the quality of the
joint (see Table 2).
500 Y. Guo et al.
for the HE700 alloy castings to enable valid simulations of the SPR process and joint
performance in the near future.
The ability to successfully join structural Al castings made by the HVHPDC process
to the automotive superstructure would have a significantly positive impact on the auto-
motive structural casting market and lead to significant automotive lightweighting
opportunities with a reduced carbon footprint, soon.
Acknowledgements The authors wish to thank the Natural Sciences and Engineering Council of
Canada for the financial assistance in this project through the Collaborative Research and Development
(CRD) grant program.
References
1. Luo AA. Recent advances in light metals and manufacturing for automotive applications. CIM
J. 2021;12:79–87. https://doi.org/10.1080/19236026.2021.1947088.
2. Ducker Frontier. North American light vehicle aluminum content and outlook 2020. 2020.
Washington, DC, USA. p. 40. Available online: https://drivealuminum.org/wp-content/
uploads/2022/05/DuckerFrontier-Aluminum-Association-2020-Content-Study-Summary-Report-
FINAL.pdf
3. Burd JTJ, Moore EA, Ezzat H, Kirchain R, Roth R. Improvements in electric vehicle battery
technology influence vehicle lightweighting and material substitution decisions. Appl Energy.
2021;283. https://doi.org/10.1016/j.apenergy.2020.116269.
4. Aziz T, Phillion A, Shankar S, Sadayappan K. Understanding pre-solidified grains in structural
castings of Nemalloy HE700 experiments. In: Minerals, metals and materials series. Cham:
Springer; 2022. p. 148–55. https://doi.org/10.1007/978-3-030-92529-1_20.
5. Lombardi A, Byczynski G, Wu C, Zeng X, Shankar S, Birsan G, Sadayappan K. Development
of high ductility Al-Zn-Mg casting alloys for automotive structural components. In: MS and T
2019 – materials science and technology 2019. 2019. p. 1049–56. https://doi.org/10.7449/2019/
MST_2019_1049_1056.
6. Sigworth GK, Howell J, Rios O, Kaufman MJ. Heat treatment of natural aging aluminum casting
alloys. Int J Cast Met Res. 2006;19:123–9. https://doi.org/10.1179/136404605225023198.
7. Lumley RN, Odonnell RG, Gunasegaram DR, Givord M. Heat treatment of high-pressure die
castings. Metall Mater Trans A Phys Metall Mater Sci. 2007;38:2564–74. https://doi.org/10.1007/
S11661-007-9285-4.
8. Yaghi PB, Shankar S, Birsan G, Sadayappan K, Lombardi A, Byczynski G. Al-Fe based eutec-
tic alloy for automotive structural castings. North American Die Casting Association (NADCA);
2021. p. 110. https://digitaleditions.walsworth.com/publication/?m=40762&i=719753&p=112&s
earch=perla&ver=html5
9. Wu C, Zeng X, Shankar S, Birsan G, Sadayappan K, Lombardi A, Byczynski G. Microstructure
and uniaxial tensile properties of heat treatable Al-Zn alloy for structural HPDC components
microstructure and uniaxial tensile properties of heat treatable Al-Zn alloy for structural HPDC
components. In: Die casting congress & exposition, Indianapolis. 2018.
10. Abe Y, Kato T, Mori K. Self-pierce riveting of three high strength steel and aluminium alloy sheets.
Int J Mater Form. 2008;1:1271–4. https://doi.org/10.1007/s12289-008-0134-9.
11. Achira S, Abe Y, Mori KI. Self-pierce riveting of three thin sheets of aluminum alloy A5052 and
980 MPa steel. Materials. 2022;15:1010. https://doi.org/10.3390/ma15031010.
12. He X, Pearson I, Young K. Self-pierce riveting for sheet materials: state of the art. J Mater Process
Technol. 2008;199:27–36. https://doi.org/10.1016/j.jmatprotec.2007.10.071.
13. Robles Hernández FC, Sokolowski JH. Comparison among chemical and electromagnetic stirring
and vibration melt treatments for Al–Si hypereutectic alloys. J Alloys Compd. 2006;426:205–12.
https://doi.org/10.1016/j.jallcom.2006.09.039.
14. Liu Y, Wang G, Weiler J. Padded self-piercing riveting (P-SPR) on magnesium high pressure die
casting. 2022. https://doi.org/10.4271/2022-01-0249.
15. Hoang NH, Porcaro R, Langseth M, Hanssen AG. Self-piercing riveting connections using alu-
minium rivets. Int J Solids Struct. 2010;47:427–39. https://doi.org/10.1016/j.ijsolstr.2009.10.009.
16. Miracle A, Olaleke MO. Predictive model for thermal response during dry machining of Al 6082-
T6 using Fem. Int J Eng Tech Res. 2020;9:1261–71.
17. Mucha J, Kaščák L, Witkowski W. Research on the influence of the aw 5754 aluminum alloy
state condition and sheet arrangements with aw 6082 aluminum alloy on the forming process and
strength of the clinchrivet joints. Materials. 2021;14:2980. https://doi.org/10.3390/ma14112980.
18. Yu H, Guo Y, Lai X. Rate-dependent behavior and constitutive model of DP600 steel at strain rate
from 10-4 to 103 s-1. Mater Des. 2009;30:2501–5. https://doi.org/10.1016/j.matdes.2008.10.001.
19. Wu C. Quantitative metallography, optimization of HVHPC process and modelling solute homog-
enization during solutionizing of Al-4Zn-1Mg-1.2Fe-0.1Ti alloy, Master’s thesis, Mcmaster
University, 2019.
20. Behrens BA, Jüttner S, Brunotte K, Özkaya F, Wohner M, Stockburger E. Extension of the con-
ventional press hardening process by local material influence to improve joining ability. Procedia
Manuf, Elsevier B.V. 2020:1345–52. https://doi.org/10.1016/j.promfg.2020.04.258.
21. Karim MA, Jeong TE, Noh W, Park KY, Kam DH, Kim C, Nam DG, Jung H, Do Park Y. Joint qual-
ity of self-piercing riveting (SPR) and mechanical behavior under the frictional effect of various
rivet coatings. J Manuf Process. 2020;58:466–77. https://doi.org/10.1016/j.jmapro.2020.08.038.
22. Zhao X, Zhang J, Chu YL, Cheng P, Meng D. Research on joining high pressure die casting parts
by self-pierce riveting (SPR) using ring-groove die comparing to heat treatment method. In: SAE
technical papers. 2020. https://doi.org/10.4271/2020-01-0222.
23. Ang HQ. An overview of self-piercing riveting process with focus on joint failures, corrosion
issues and optimisation techniques. Chin J Mech Eng (English Edition). 2021;34. https://doi.
org/10.1186/s10033-020-00526-3.
24. Li D, Han L, Lu Z, Thornton M, Shergold M. Influence of die profiles and cracks on joint buttons
on the joint quality and mechanical strengths of high strength aluminium alloy joint. Adv Mater
Res. 2012:398–405. https://doi.org/10.4028/www.scientific.net/AMR.548.398.
Natural Ageing of Al–Fe(Zn,Mg) Alloy Castings
for Structural Automotive Application
Chimaobi Orji1 (*), Anthony Lombardi2, Glenn Byczynski2, Mohamed Hamed1,

and Sumanth Shankar1
1
Department of Mechanical Engineering, McMaster University, Hamilton, ON, Canada
Orjic4@mcmaster.ca
2
Nemak USA/Canada, Windsor, ON, Canada
Abstract. Novel Al–Fe-based eutectic alloys with Zn and Mg as precipitation strength-

eners were used in the manufacturing of test plates using high-vacuum high-pressure die
casting to evaluate the feasibility of manufacturing structural automotive components
with a higher strength to weight ratio than the present. Natural ageing at room tempera-
ture is a typical behaviour of this new alloy due to the continuous precipitation reactions
of the Zn and Mg atoms dissolved in the primary solid Al phase in the microstructure. The
sequential stages of precipitation include the formation of self-clusters of atoms, dissolu-
tion of self-clusters, formation of GP-I zones, dissolution of GP-I zones, formation of
GP-II zones and stabilization of GP-II zones; all these stages distinctly affect the tensile
properties. Uniaxial tensile tests of castings coupled with the differential scanning calo-
rimetry experiments were carried out to elaborate the mechanisms of natural ageing in
these alloys. The results show that the elongation of the castings is unaffected during natu-
ral ageing and the strength increases to its double, with time, before stabilization. The
properties of as-cast and stabilized state were significantly improved when compared to
the contemporary Al–Si–Mg alloy such as the Aural 5.
Keywords: Natural Ageing · Self-clusters · Dissolution · Precipitates · Zones
1 Introduction
Over the past half a century, there has been a significant surge in global activities and
investments into research and development of technologies that enable reduction of the
curb weight of an automobile. The key motivation of these activities is to reduce the
environmental impact of the transportation sector by cutting down on the emission of
greenhouse gases and drastically reducing the use of fossil fuels. According to a study
published in 2020 by the US Energy Information Administration [1], the transportation
industry is the single largest contributor to the fossil fuel consumption and by conse-
quence, the global CO2 emission, as well. Hence, a reasonable measure would be to
decrease this greenhouse gas volume significantly by finding alternative renewable
energy sources. The deployment of electric vehicles (EV) as a substitute for automo-
biles powered by internal combustion engines has faced various challenges, most of
which were attributed to the increased weight necessitated by the alteration of the
https://doi.org/10.1007/978-3-031-17425-4_65
504 C. Orji et al.
design of the vehicle to accommodate the changes brought forth by the switch in the
drivetrain and energy transport [2, 3]; For example, a 100 kg curb weight reduction in
EV presents an additional 2 km of driving range for the battery charge, which translates
to about 3% energy savings on an average EV. To realize the stringent global targets for
automotive light-weighting [4], which presently mandates in the excess of 30% reduc-
tion in the curb weight, the recent technological thrust has been primarily in developing
new Al alloys and processes such as the high-vacuum high-pressure die casting
(HVHPDC) for near-net-shaped manufacturing of structural automotive components. A
primary challenge in adopting near-net-shaped castings of Al alloys for structural auto-
motive applications has been the lack of an optimum combination of strength and
toughness while using the conventional alloys such as Aural 5 (AlSi8MnMg), Aural 2
(AlSi10MnMg) and Mercalloy 367 (AlSi9MgMnSr), which are based on the Al–Si–Mg
hypoeutectic alloy system and yield a polyphase solidified microstructure [5]. In all
these alloys, typically, heat treatment procedures are undertaken, such as homogeniza-
tion solution treatment (T4) followed by artificial ageing (T7), or alternately, artificial
ageing applied to parts in the F temper (T5). These treatments are aimed at providing
the required strength and toughness for the cast components, after the F temper.
The alloy developed recently is an Al–Fe hypoeutectic alloy with levels of Zn and
Mg that render them in solid solution after solidification at the rapid solidification rates
associated to HPDC, so as to strengthen through subsequent natural and/or artificial
ageing phenomena. This new alloy is commercially termed as Nemalloy HE700 or
merely HE700 in this publication. The HE700 alloy was developed with the intention
of providing an alloy material that would lend itself to automotive application with
negligible heat treatment interference after the casting.
In this publication, the natural ageing process of the HE700 alloy casting after solid-
ification in a HVHPDC process has been shown and discussed. The HE700 alloy, in its
F temper, continues to strengthen through the precipitation reaction mechanisms involv-
ing the Al, Zn and Mg atoms within the primary aluminium solid solution in the micro-
structure. These sequences of precipitation reactions lead to changes in the strength and
toughness of the alloy castings and this publication aims to elucidate the stages in this
natural ageing process at ambient conditions through the use of uniaxial tensile testing
and differential scanning calorimetry (DSC) experiments and analyses. The various
stages of natural ageing phenomenon in HE700 alloy involve progressive formation and
dissolution of several microstructural phases within the primary aluminium grain. It
was notable that during the natural ageing of HE700 alloy, the strength significantly
increases while the elongation remained fairly unaffected.
2 Experiments
A 1200 T HVHPDC machine was used to cast the samples in a die tool made of H13
steel. Figure 1 shows a composite photograph of the casting sample along with the four
flat tensile test specimens (4 mm thick throughout) that were used in this study to
understand the tensile properties of these castings. The HVHPDC process was con-
trolled with automatic lubrication between shots, ladle poured and a maximum vacuum
Natural Ageing of Al–Fe(Zn,Mg) Alloy Castings for Structural Automotive Application 505
Fig. 1. Typical casting of the structural samples using HVHPDC process. Also shown in the inset is a
magnified image of the flat tensile test bar that was used to evaluate the uniaxial tensile properties.
of less than 20 mbar; the shot weight averaged about 3.6 kg each. The typical micro-
structure of HE700 and Aural 5 alloy castings are shown in the past publications [6].
The MTS universal screw-driven tensile testing machine was used with a 50 kN load
cell and the testing was carried out according to the ASTM E8 standard with a cross-
head travel speed of 1 mm/min. The DSC experiments were carried out in a DSCQ100
machine with a heating rate of 20 C/min on a sample of about 15–25 mg each, between
40 and 500 °C.
Figure 2 presents the progression of yield strength and total elongation at failure of the
HE700 alloy casting during natural ageing at ambient conditions. The yield strength
and the % elongation at failure are shown in each of the y-axes, while the time of natural
ageing in days is shown on the x-axis. Also, superimposed in Fig. 2 are the various
stages of precipitation reactions within the primary Al solid solution in the microstruc-
ture; these stages were ascertained from the knowledge in the background literature and
the analyses of the DSC experiments. Notably and uniquely, there is no appreciable
change in the % elongation of these castings during incubation at room temperature.
The changes in strength as a function of natural ageing time occurred in five distinct
stages, which were associated to the progression of the precipitation sequence, as indi-
cated in Fig. 2. The stages observed in the natural ageing process were either hardening
stages, where new precipitate phases formed within the primary Al grains, or softening
stages caused by the dissolution of precipitates from the previous stage to facilitate the
next precipitate stage. The yield strength of the casting on the day of the casting was
approximately 100 MPa, while it naturally ages with time to an asymptotic value of
506 C. Orji et al.
Fig. 2. Yield strength and elongation of HE700 casting (4 mm thick) during incubation at ambient
condition. Also shown are the typical stages of precipitation reaction events.
about 190 MPa over several months. Furthermore, this alloy experienced gradually
increasing toughness with incubation time due to the increased strength and unaffected
elongation.
Figure 3 presents a schematic of stages during the precipiattion reactions, starting
from the net-shaped casting of the alloy component. Upon solidification and cooling,
the supersaturated Al–Fe–Zn–Mg alloy commences the first stage of the ageing pro-
cess. During this very rapid regime, the solute atoms (Zn and Mg) do not have enough
time to settle into lattice sites before solidification occurs. This creates a high concen-
tration of retained vacancies in the primary Al matrix. With continued natural ageing at
ambient temperature, the solute atoms begin to get expelled from the solid solution as
there are not enough lattice sites to accommodate all of them. The more these atoms are
ejected, the higher the tensile properties. This is due to increased imperfections, dislo-
cation densities and lattice strain, hence, the observed rapid increment in tensile proper-
ties within the early stages [7].
These atoms of Mg and Zn independently form atomic self-clusters (Stage I), which
is driven by the concentration of the alloying elements in solid solution being higher than
their equilibrium solubility limit [8, 9]. These self-cluster formations also aid the suste-
nance of the lattice strain. As the clustering rate asymptotes to a plateau (end of Stage I)
due to increasing vacancy annihilation, these initially self-clusters of atoms begin to
dissolve into vacant lattice sites. The Zn atoms tend to be the favoured species to migrate
into vacancies due to their higher potential since they have a lower activation energy than
Mg atoms in aluminium [10], and the Zn atoms outnumber the Mg atoms [8].
Fig. 3. Schematic of the sequence of events that lead to the continuous precipitation reactions within
the primary Al matrix in an HE700 alloy casting during the natural ageing process at ambient
conditions.
As these atoms take up vacant sites on the matrix, the self-clusters continue to break
down and this results in a significant reduction in the lattice strain and tensile properties
measured initially (Stage II). The disappearance of the self-clusters signifies the end of
the Stage II of transformation even though there is no clear-cut distinction between the
various stages as the reactions associated with each stage are seen to overlap into the
succeeding stage and then gradually fade out.
The fast dissolution of the initial self-clusters (especially the Zn self-clusters) result
in an overall reduction in the lattice strain, reflected by the reduction in the yield
strength. However, the Zn atoms cluster dissolution prompts the Mg atoms to migrate
and merge forming co-clusters with some Zn atoms that are still in the precipitate phase
(~MgZn4) [7]. The leads to a resurgence in the strength due to the strain imposed by
these solute-rich superclusters, thus heralding the Stage III of precipitation—the forma-
tion of GP-I zones [7]. This is governed by the migration of Mg atoms which serve as
nucleation sites for the co-clusters [10]. Meanwhile, as the vacancy occupation contin-
ues, the volume of vacancies reduce, so also does the rate of change in the tensile prop-
erties, towards the later stages of Stage III, until the Mg superclusters formation
stabilize.
As time progressed in Stage III, the gradual dissipation of more atoms into encoun-
tered vacancies is the driver for the breakdown of the GP-I zones that were previously
formed [7], which marks the beginning of Stage IV. The breakdown in the GP-I zones
at the beginning of Stage IV is associated with an appreciable reduction in yield
strength, as shown in Fig. 2, due to the loss of lattice strains that accompanied the GP-I
zone formation. However, the Mg-governed cluster dissipation and atom migration will
be reactivated in response to the atoms lost to vacant sites. This results in a rampant
dissolution of the GP-I zones and the accompanying formation of new GP-II zones
shortly thereafter due to mass migration of Zn atoms to capture the Mg atoms that have
been disassociated from the destabilizing GP-I structure (Stage IV). Therefore, Stage
IV is a combination of disassociation and dissolution of the GP-I precipitates to form
new GP-II precipitates, which further strengthen the matrix. The dissolution process of
GP-I leads to the weakening of the matrix strength. The strengthening effect of the
GP-II precipitates is observed in Fig. 2, as there is a gradual increase in yield strength
508 C. Orji et al.
as time progresses in Stage IV after the initial reduction, which is associated to an

increasing content of GP-II zones as they replace the dissolving GP-I zones. The domi-
nating reaction in the tested sample determines the perceived result. After the complete
dissolution of GP-I zones, the natural ageing process results in a continued but slow
increase in the yield strength, which is associated to the nucleation and growth of GP-II
precipitates to their optimal size for strengthening within the primary Al matrix (Stage
V).
The kinetics of the precipitation strengthening in the HE700 alloy during incubation
in ambient conditions (shown in Fig. 2) is not sensitive to the quenching process follow-
ing the removal of the casting from the die, but its kinetics and outcome could be altered
by temperature and timing of treatment relative to the time of solidification and casting
[7]. Typically, in Al–Zn–Mg-based alloys, during the precipitation in natural ageing, the
strength decreases when the Zn is the dominantly mobile atom.
Conversely, when Mg atoms dominate solute atom mobility, the strength increases.
Figure 4 shows the typical results from the DSC experiments. There were three
samples used in these experiments: the pure Al as a baseline reference, HE700 at
57 days of natural ageing and 84 days of natural ageing, respectively. Figure 5 shows
the specific enthalpy and the derivative of specific enthalpy with time (rate of specific
enthalpy change). The results for the pure Al sample show that after the initial transient
of high-rate increase in specific enthalpy during the start of the heating, the value of
specific enthalpy was mostly constant as shown by the nearly horizontal curve for the
rate of change of enthalpy (dh/dt). This is attributed to a lack of dissolved solutes that
could lead to any measurable precipitation reactions within the primary Al grains. In
contrast, there is significant deviation in the specific enthalpy during the heating of
HE700 samples and the amount of total enthalpy needed to heat increases as the num-
ber of days of natural ageing increases. This is due to the larger quantity of precipitates
in the specimens with higher natural ageing time as precipitation requires higher heat
values to enable mobility of atoms and the formation of precipitates. Fig. 4 also shows
the various stages of the precipitation reaction during continuous heating of the sample.
The similarity between the curves shown in Figs. 2 and 4, in the regimes of the five
stages of precipitation, is notable.
Figure 5 shows the results of an experiment, wherein cast samples after no natural
ageing and those with 36 days of natural ageing were subjected to heating at 70 °C for
several minutes and the uniaxial tensile tests were carried out on these samples after
progressively increasing time intervals. The samples that did not naturally age shows a
significantly lower strength after the same treatment than a sample after 36 days of
natural ageing (with beginnings of GP-I precipitates in its microstructure). The results
in Fig. 5 also show that the kinetics of the precipitation reaction at 70 °C depended on
the initial state of the atoms within the primary Al matrix. The presence of the Mg and
Zn self-clusters, formed at temperatures lower than 60 °C, is critical to enable an
increased kinetics of the precipitation process to form the GP-I and GP-II precipitates.
Some remarkable points to note are that through the entire fluctuations of tensile
properties for the duration of the natural ageing process, no significant effect on the
elongation was recorded. Secondly, no matter the treatment given to the HE700 alloy
(Natural ageing, elevated isothermal treatment or constant positive gradient tempera-
ture treatment using DSC), all the mentioned transformational stages and phases were
Fig. 4. Typical response in DSC for specific enthalpy (h) and transient gradient of specific enthalpy
(dh/dt) for HE700 alloy casting after 57 days of natural ageing (NA) (black curves) and 84 days of NA
(grey curve). The response from pure Al is also shown as baseline comparison. Further, the temperature
ramp up was 20 °C.min−1 between 40 and 500 °C; the temperature increase was constant and shown
alongside the time on the x-axis.
clearly observed, while the kinetics of these respective transformations differed. None
of the five stages were skipped during natural or artificial ageing heat treatment. It is
also worthy of note that the number density of initial self-clusters and vacancy concen-
tration present at the time of GP-I formation plays a critical role in the final mechanical
properties after GP-II stabilization, because each phase precipitate is directly influ-
enced by the density and distribution of the phases and atom mobility in the preceding
stages.
4 Conclusion
The results show that the Nemalloy HE700 is an apt example to demonstrate the feasi-
bility of near-net-shaped casting of structural automotive components from Al–Fe
hypoeutectic alloy system with Zn and Mg as solute elements for precipitation strength-
ening process. The alloys castings produced were all of sound quality with no visual
defect. The properties of strength and elongation were superior to the castings made
510 C. Orji et al.
Fig. 5. Transient changes in yield strength during treatment at 70 °C for HE700 alloy casting samples
that were not naturally aged in ambient (0NA) and natural aged for 36 days after casting.
from the Al–Si–Mg alloy such as the Aural 5 and Aural 2 [5]. The HE700 alloy showed
tendencies of natural ageing at room temperature wherein, there are five distinct stages
of atom mobility and precipitation between Mg and Zn in the primary Al phase matrix.
These stages of precipitation lead to continuous improvements to strength while not
affecting the elongation appreciably. The formation of the initial self-clusters of Zn and
Mg atoms, respectively, in the Al matrix is critical to improving the kinetics of the natu-
ral ageing process and realizing the fullest capacity to strengthen. The number density
of self-clusters is directly proportional to the strength attained during natural ageing.
The elongation of the castings seems unaffected by the natural ageing process suggest-
ing a continuous increase in the toughness of the cast components from the precipita-
tion reactions. The unique natural ageing phenomena of the HE700 alloy structural
casting lend itself as a viable alternative for automotive light-weighting through the
near-net-shaped manufacturing of structural components.
Acknowledgements The authors would like to thank the Natural Sciences and Engineering Research
Council of Canada (NSERC) for their continued funding support in this project through their
Collaborative Research and Development (CRD) grant.
References
1. U.S. Energy Information Administration (EIA). Fossil fuel consumption by source and sector.
2020. https://www.eia.gov/totalenergy/data/monthly/pdf/flow/fossil_fuel_spaghetti_2020.pdf
2. Datsen. 2021. https://blogs.synopsys.com/from-silicon-to-software/2021/08/24/electric-vehicle-
design-challenges/
3. Redelbach K, et al. Impact of lightweight design on energy consumption and cost effectiveness of
alternative powertrain concepts. 2012. https://www.researchgate.net/publication/259897053
4. Ducker Worldwide. Ducker Frontier. 2021. https://ducker.com/automotive-transportation/
5. Yaghi P, et al. Al-Fe based eutectic alloy for automotive structural castings. Arlington Heights, IL:
North American Die Casting Association (NADCA); 2021.
6. Aziz T, Phillion A, Sadayappan K, Shankar S. Understanding pre-solidified grains in structural
castings of nemalloy HE700 experiments. In: Eskin D, editor. Light metals 2022. The minerals,
metals & materials series. Cham: Springer; 2022.
7. Kovàcs I. The formation of GP zones in Al-Zn-Mg alloys. Budapest: Institute of General Physics,
Lorând Eötvös University; 1984.
8. Wan L, et al. The natural ageing effect on pre-ageing kinetics of Al-Zn-Mg alloy. J Alloys Compd.
2019. https://doi.org/10.1016/j.jallcom.2018.10.338.
9. Liu C, et al. Reversion of natural ageing and restoration of quick bake-hardening response in
Al-Zn-Mg-Cu alloy. J Mater Sci Technol. 2021. https://doi.org/10.1016/j.jmst.2021.03.070
10. Rosalie JM, Somekawa H, Singh A, Mukai T. The effect of precipitation on strength and ductility
in a Mg-Zn-Y alloy. Structural Materials Unit, National Institute for Materials Science, Tsukuba,
305-0047, Japan. Dept. Mechanical Engineering, Kobe University, 1-1 Rokkodai, Nada, Kobe city,
657-8501 Japan. 2012. https://doi.org/10.1016/j.jallcom.2012.09.027.
Part VI Processing of Critical Materials
Integrating Carbon Capture in Mining Through
Metallurgy. Part 1: Leaching and Reclamation
of Asbestos Tailings: Thetford Mines Carbon Capture
and Remediation Project
Mike Kelland1, Greg Rau1, Brock Battochio1, Jason Vallis1, Mykolas Gladkovas2,
Sridevi Thomas2, Kevin Bradley2, Clive Brereton3, Srinath Garg1,
and Alex Mezei1 (*)
1
Planetary Hydrogen, Dartmouth, NS, Canada
alex@planetaryhydrogen.com
2
SGS Canada Inc, Lakefield, ON, Canada
3
Noram Engineering & Constructors Ltd., Vancouver, BC, Canada
Abstract. Planetary hydrogen (“PH”) is engaged in developing a proprietary carbon cap-

ture technology aimed to be integrated within existing and closed mining operations. The
enabling metallurgical process is based on classic unit operations involving sulfuric acid
leaching of magnesium contained in suitable ores and tailings, solution processing and
electrolytic acid regeneration with simultaneous magnesium hydroxide precipitation. The
resulting hydroxide precipitate is used for either ocean air capture (“OAC”) or direct land-
based carbon (DC) capture or a combination thereof. Column-simulated heap leaching
results to date on a representative Thetford Mines asbestos tailings sample produced
recoveries of 84%, 90%, and 75% for magnesium, nickel, and cobalt, respectively, after
86 days of leaching. Average asbestos conversion to silica was 83%, with complete con-
version (~0% asbestos) within the top layer. Vat leaching results produced comparable
metallurgical performance after 29 days of leaching. Optimization testwork is underway
at the time of writing the paper. The OAC capture potential of the magnesium hydroxide
produced is estimated at 1.65 t dry tailings feed per ton of CO2. Mass and energy balances
of full-scale operation predict specific energy consumption is 9.5 MWh/t captured CO2,
translating into a CO2 carbon–footprint ratio of 4.4:1 as tons captured vs. tons generated.
The capex of the full-scale operation with 1000 ktpa CO2 removal nameplate in the form
of magnesium hydroxide at plant discharge is tentatively estimated to USD$ 1232 Bn.
The annual opex was estimated to USD$ 0.364 Bn pointing to a pre-tax annual ROI of
48% with 7.1 years of capital amortization.
Keywords: Ocean Alkalinity Enhancement · Magnesium hydroxide · Heap

Leaching · Carbonation
1 Introduction
Mine tailings with a high magnesium content present an ideal setting to observe and
promote mineral carbonation due to the presence of reactive magnesium bearing sili-
cates such as Brucite (Mg(OH)2) and serpentine polymorphs (Mg3Si2O5(OH)4) [1–3].
https://doi.org/10.1007/978-3-031-17425-4_66
516 M. Kelland et al.
Ore processing techniques such as milling and crushing further enhance the mineral
reactive surface area to promote passive carbonation [4–6]. The global CO2 sequestra-
tion potential of the ultramafic mine tailings (asbestos, talc, etc.) is estimated to be
about 175 Mt CO2/y [6]. Although the rates of passive carbonation have shown to offset
approximately 11% of the CO2 emissions, accelerating the observed rates requires geo-
chemical treatments or changes [7]. One such geochemical treatment involves the acid
leaching of magnesium-bearing silicates to accelerate the dissolution of magnesium
and other metals from host minerals followed by their sequential precipitation.
Planetary Technologies is developing a proprietary process flowsheet that integrates
acid leaching and subsequent solution treatment with electrochemical salt splitting in
order to recover various pay-metal intermediates along with magnesium hydroxide
from amenable feedstocks. A simplified process schematic is shown in Fig. 1. The tail-
ings are leached with a dilute stream of sulfuric acid produced by the magnesium sul-
fate electrolysis circuit. The full-scale commercial leaching operation will be conducted
in heap configuration; however, investigation of vat leaching as a temporary or back-up
option is also being conducted. The Pregnant Leach Solution (PLS) from either leach-
ing circuit is directed to the Iron removal circuit, wherein iron precipitation is con-
ducted within a narrow pH window of 4.4–4.6 by the addition of a controlled amount
of slaked lime. Oxygen from the electrowinning circuit is injected to significantly
improve the reaction kinetics. Other metals such as aluminum, manganese, and chro-
mium co-precipitate with iron to various degrees. The resulting first barren solution is
directed to the nickel-cobalt mixed hydroxide product (MHP) circuit. Nickel and cobalt
are precipitated by controlled addition of slaked lime at a pH of 8–8.3. The resulting
second barren solution is directed to the electrolysis circuit that produces the sulfuric
acid that is being redirected to the leaching circuit, and the final magnesium hydroxide
salient product intended for carbon capture. Carbon capture is realized by either
scrubbing the carbon dioxide from the atmosphere (land carbonation) or dispersing in
sea water (ocean alkalinity enhancement).
The present work describes the results of the bench-scale proof of concept testwork
conducted on a representative asbestos tailing sample originating from the Thetford
mines, located in the province of Quebec, Canada. The amenability of the sample to
heap and vat leaching was established by conducting a set of bottle roll tests at ambient
Sale
Feed Reclaimed O2 H2 O
tailings tailings
H2
Ca(OH) 2 Ca(OH)2
Fe,Al,Cr,Mn. NiCo MHP MgSO 4

Heap or Vat PLS: Mg, Fe, Ni, EW
Precipitation Precipitation
Leaching Co, Al, Mn, Co...
Filtered
Disposal Sale
Anolyte recycle catholyte
recycle
H2 SO4 Mg(OH) 2
CO 2 Capture and Sequestration
Fig. 1. Simplified process schematic.

Integrating Carbon Capture in Mining Through Metallurgy. Part 1: Leaching… 517
temperature. The responses to vat and heap leaching were investigated using laboratory-
scale column experiments. The solution processing and electrochemical responses were
investigated through baseline bench-top testworks.
2.1 Tailings Mineralogy and Chemical Composition
The mineral feed considered for the process described herein consists of the Thetford
Asbestos Tailings. The asbestos mine tailings piles consist of Normandie, Bell, National,
Lac d’Amiante, British Canadian I & II, and the King Beaver resources. Cumulatively,
they are estimated to contain 246,286,000 metric tons of measured, indicated, and
inferred resources [8]. The tailings consist primarily of Fe, Ni, Si, and Mg along with
minor quantities of Ca, Al, Ti, and Mn. The magnesium silicates include chrysotile and
serpentine as the major constituents, along with other silicate-rich minerals such as
olivine, lizardite, and orthopyroxene. About 90% of the nickel is present in a soluble
oxidized form. Insoluble nickel is primarily deported to Heazlewoodite (Ni3S2) – a
highly refractory nickel sulfide, Awaruite (Ni3Fe) – a naturally occurring alloy of Ni
and Fe, and Garnierite (NiO.MgO(SiO2).xH2O) – a hydrated metasilicate in which the
nickel substitutes for some of the magnesium in the serpentine crystal lattice. Due to the
absence of any iron sulfides, Fe is primarily found in the trivalent state in the minerals
hematite (Fe2O3) or magnetite (Fe3O4). The Normandie tailings comprise a mix of
mainly sand-to-pebble-sized ultramafic to mafic host rock material, some chrysotile
and serpentine fibers, and some powdered rock. Some magnetite and brucite are known
to be collectively associated with the chrysotile in veins.
The chemical composition of a crushed and pulverized Thetford tailings head sam-
ple was determined by XRF (X-ray fluorescence) and ICP (Inductively Couple Plasma
spectrometry). Key analysis is summarized in Table 1. The tailings sample predomi-
nantly contained Mg (22.7%), Si (18.7%), and Fe (5.1%) with some Ni (0.17%) and Co
(0.01%). The asbestos content in the head sample was estimated at 3 wt% according to
the polarized light microscopy (“PLM”) method.
Table 1. Chemical composition (wt%) of the Thetford Tailings.

Elemental analyses, %
Mg Si Al Ni Co Ca Fe Cr Na
22.7 18.7 0.48 0.17 0.01 0.19 5.11 0.24 0.05
Cu CO32− Cl− Mn Zn V Ti P Asbestos
0.002 0.77 0.01 0.08 <0.004 <0.010 0.01 0.004 3
2.2 Metallurgical Thesis
The known challenges with any fibrous feed, including asbestos containing magnesium
silicate tailings relate physically to operability [9]. This is because of the flow-adverse
process response impacted by their fibrous structure. The flowability (rheology) of a
fibrous-leaching slurry prevents mixing at the required shear rates [10].
In addition, chemically, the elevated levels of acidity and overall ionic content in
conjunction with the high silica content further challenge the traditional metallurgical
approach. Silica gel formation tends to commonly generate additional operability issues
during the conventional-agitated leaching and subsequent separation and processing
steps [11].
The metallurgical thesis proposed by Planetary Technologies relies upon changing
the type of leaching from agitated mixing to heap or vat leaching. These techniques
offer certain advantages in that they actually take advantage of the fibrous nature of the
tailings. This physical reality promotes in turn a radically different, process-friendly
flow behavior. The key underlying rheological improvement is due to the fact that it is
only the leach solution that needs to flow freely through the feed, instead of creating
slurry and forcing it to flow by mixing.
The difference between the two methods is the nature of the flow of the leaching
solution, also known as “lixiviant.” Heap leaching is based on gravity-based irrigation
flow of the lixiviant between the mineral particles that are not submersed in said solu-
tion. Vat leaching is realized by forced percolation flow of the lixiviant through the
minerals fully immersed in it. Heap leaching is realized by stacking the feed material in
“heaps” and installing the required irrigation system and drainage ponds. Vat leaching
is realized in large tanks fitted with a solution circuiting and storage system. This dif-
ference allows for the possibility of significantly larger-scale heap operations compared
to vat operations.
Additional advantages of heap and vat leaching methods include the use of dilute
solutions, relative insensitivity to the actual pay-metal grade of the feed material, as
well as lower capital and operating expenses.
2.3 Project Deployment Outline
The overall objective of the metallurgical component of the project is to demonstrate

the feasibility of the electrochemically enabled heap and vat leaching routes, including
the required downstream processing steps for the Thetford asbestos tailings deposit.
The project development is conducted on a staged-gated sequence aimed on continu-
ous due diligence in all relevant aspects [12–19]. Key risk areas being mitigated include
technical, health-safety-environmental, economic, financial, permitting, geopolitical,
marketing, and community relations.
Technically, the methodology applied meets or exceeds the CIM Best Practice
Guidelines for Mineral Processing (BPGMP) [19]. They supplement the Principles of
Process Support for Mineral Resources/Mineral Reserves Estimation for the purposes
of the NI 43-101 Technical Report (“NI 43-101 TR”).
The testwork involves the following main phases:
• Laboratory (“Bench”) scale, non-integrated: Amenability – bottle roll leaching,
small-scale column-simulated heap leaching – single lift, iron and impurities
removal, Ni-Co MHP precipitation, magnesium sulfate electrolysis, and carbonation
of the magnesium hydroxide;
• Pilot testwork: medium- to large-scale column-simulated heap leaching, partially
integrated, including electro-cell prototype, continuous processing, including
carbonation;
• Demonstration plant testwork: fully integrated modular processing capability for
heap and processing sections.
Process and economic models are generated upfront and updated in tandem with the
progress of the testwork progress during and at the end of each phase.
Due diligence is performed continuously on aspects, including technical (metal-
lurgy, electro, handling, storage, delivery logistics), gaps flagging, environmental (per-
mitting, closure, remediation, fatal flaws (all of the above), as well as monitoring of
classic risk factors and uncertainties: financial, geopolitical, environmental, permitting,
marketing, social, etc.
3 Testwork Conditions and Results Summary
3.1 Column-Simulated Heap Leaching Conditions and Results
The setup used for the column used to simulate heap leaching consisted of a polyvinyl
chloride (PVC) column of 15 cm diameter and 1.8 m height provided with a diffuser
membrane tubing at the top and a perforated bottom fitted with a geotextile cloth
(Fig. 2). A tailings sample charge of about 36 kg was loaded into the column. Irrigation
was realized by pumping the lixiviant solution containing 7.3 wt% H2SO4 and 4% wt.
Na2SO4 onto the top of the column at a flowrate of 3.5 mL/min corresponding to 0.2 L/
min.m2 irrigation rate. Fresh lixiviant was fed until the pH reached a value of about 2.5.
At this point, it became “Pregnant Leach Solution,” and it was collected continuously
in a separate storage vessel. The test continued in recirculation mode in order to main-
tain the leaching kinetics until the reaction kept progressing at an increasing cumulative
rate. The column-simulated heap leaching was continued until metal extractions pla-
teaued after 86 days of leaching. The column was washed with DI water until a sample
of the discharge wash water reached a pH value of about 4. The volumetric inflow and
outflow from the column were similar and stable, predictive of a stable heap leach
operation under the conditions as tested.
The extractions for magnesium, iron, nickel, and cobalt were 84%, 53%, 90%, and
74%, respectively. The metallurgical balance indicated complete asbestos removal in
the solid samples withdrawn from the top of the column. The asbestos content in the
solid sample collected from the overall residue was 1 wt% versus 3 wt% in the head
Lixiiv iant and recy cled

solution
Dif f user
membrane
6 in . Dif user
membrane
Leach
Column
Ore
Geotextile membarne
Peristaltic
Pregnant pump
solution
Leach
Solution
Drum PLS Rec
Pregnant
solution Peristaltic Pail
collection pump
Fig. 2. Experimental setup for the simulated column heap leaching.

Table 2. Column-simulated heap leaching results summary.

Stream Mg Si Al Ni Co Ca Fe Asbestos
Leaching, analyses in units as shown
Feed, % 22.7 18.7 0.48 0.17 0.01 0.19 5.11 3
PLS, mg/L 18,100 143 161 174 8 85 2400 –
Wash, mg/L 2020 63.2 13.5 22.7 1.2 10.1 178 –
Residue, % 6.75 32.9 0.55 0.037 0.002 0.16 3.87 1
Leaching, extractions, % by method as indicated
Calc Head 84 1 35 90 74 42 53 –
Direct Head 88 1 36 114 83 41 50 –
Feed vs. Residue 83 −2 33 87 86 53 56 81
Accountability 105 103 103 127 97 89 94 19
sample, determined by PLM. This corresponded to 81 wt% asbestos reduction for a

magnesium dissolution extent of 84%. The results are summarized in Table 2.
A TCLP (Toxicity Characteristics Leaching Procedure) test was carried out on a
“wet” composite sample retrieved from the heap leach column. The results indicated
the dissolved concentrations of deleterious elements were well below the schedule lim-
its (Table 3).
3.2 Column-Simulated Vat Leaching Conditions and Results
The tests column configuration was changed from heap leach to vat leach simulation by
relocating the membrane material at the bottom which allowed flow of lixiviant through
while preventing solids to pass. The lixiviant was pumped into the bottom of the col-
umn from the lixiviant solution drum. The amount of sample was the same as for the
heap leach simulation test, and the composition of the lixiviant was identical as well.
Percolation was realized by pumping the lixiviant onto the bottom of the column at
a flowrate of about 16 mL/min corresponding to about 1 L/min.m2 percolation rate. The
column was flooded throughout the entire test. The addition of fresh and recycled lixivi-
ant was managed similarly to the heap column and with the same pH target of 2.5. The
column-simulated vat leaching was continued until metal extractions plateaued after
29 days of leaching. The column was washed until the wash water reached a pH value
of about 4.
The results are summarized in Table 4.
The metallurgical balance returned recoveries of 79%, 88%, and 66% for magne-
sium, nickel, and cobalt, respectively. The PLM-determined asbestos content in the
“top cut” and “composite samples” were “trace” < 1 wt% and 4 wt%, respectively. The
volumetric inflow and outflow from the column were similar and stable, predictive of a
stable vat leach operation under the conditions as tested.
Table 3. Concentrations of dissolved elements (mg/L) as determined by a TCLP test on a wet

composite sample from the heap leach testwork.
TCLP scan, mg/L
Elements Hg As Ag Ba B Co Cd
Schedule limits, mg/L 0.1 2.5 5 100 500 0.5
Sample <0.00001 <0.002 <0.0005 0.163 0.14 0.0123 0.00035
Elements Cr Mg Na Ni Pb Se U
Schedule limits 5 5 1 10
Sample 0.0081 8.04 1480 0.166 0.00367 0.0016 <0.00002
Table 4. Column-simulated vat leaching results summary.

Stream Mg Si Al Ni Co Ca Fe Asbestos
Analyses, units as shown
Feed, % 22.7 18.7 0.48 0.17 0.01 0.19 5.11 3
PLS, mg/L 11,700 110 105 123 6 68 1750
Wash, mg/L 1770 59.5 23.5 10.3 0.6 6.8 712
Residue, % 9.89 29.7 0.56 0.048 0.003 0.19 4.53 4
Extractions, %
Calc Head 79 2 39 88 66 41 60 –
Direct Head 93 1 42 122 83 48 76 –
Feed vs. Residue 75 10 33 84 82 43 50 24
Accountability 118 92 109 138 101 105 126 76
3.3 Iron and Impurities Removal by Precipitation
A series of four baseline iron and impurities removal testwork was conducted on test
samples of about 2 L pregnant leach solution produced by the heap leach column simu-
lation test. Under the pre-optimized conditions, oxygen was sparged at an average vol-
ume flow rate of 1200 NmL/min into the agitated reactor. The pH was maintained
continuously at 4.6 by controlled additions of 25% wt. calcium hydroxide slurry. The
oxidation potential was maintained at 244 mV versus the silver/silver chloride refer-
ence electrode during the oxygen addition. The reaction temperature was maintained at
80 °C during the three-hour test. Nickel losses corresponding to 95% and 100% iron
removal were 15% and 25%, respectively. The composition of the precipitated residue
is summarized in Table 5.
The results were not as expected, based on previous experience with previous results
produced by comparable systems. It was postulated that the increased level of sodium
in the lixiviant triggered excessive basic salt formation. In addition, the well-known
reality of very low oxygen transfer efficiency in gas–liquid boundaries was substanti-
ated by the excess oxygen required to maximize the redox potential. Optimization
testwork is planned using PLS generated by the sodium-free lixiviant generated

electrolytically.
Table 5. Composition of the iron impurity removal and mixed hydroxide precipitates.
Stream Mg Si Al Ni Co Ca Fe
Iron and impurities 1.58 0.67 0.93 0.22 0.01 16.3 12.3
Ni/Co MHP product 2.89 – 0.22 6.20 0.39 12.9 14.3
3.4 Nickel and Cobalt Precipitation
Two ambient-temperature bench-scale precipitation tests were conducted at two stoi-

chiometry excess additions of calcium hydroxide dispensed as 25% wt. slurry. The tests
lasted 1.1 hours and 2.35 hours, respectively, during which the pH was maintained
within the 8.3–9.1 range. Nickel and cobalt precipitation at 1.98 stoichiometry excess
ranged from 80% to 88%. The composition of the precipitates is summarized in Table 5.
Analogous to the iron and impurities removal, the increased level of sodium in the lix-
iviant triggered excessive basic salt formation, most likely to even a greater extent given
the increased equilibrium pH level. This process stage will be optimized by upcoming
testwork, in integration with the preceding iron and impurities removal stage.
3.5 Magnesium Sulfate Electrolysis
Baseline testwork was conducted to investigate the electrochemical production of the

magnesium hydroxide concurrently with the regeneration of the diluted sulfuric acid to
be used for the heap or vat leaching.
The electrochemical unit consisted of a two-compartment cell allowing for the con-
tinuous withdrawal of the catholyte slurry product containing the magnesium hydrox-
ide produced at the cathode and of the anolyte of concentrated sulfuric acid produced at
the anode.
The electrolysis was conducted using a low-permeability diaphragm. The catholyte
contained 80 g/L MgSO4 for the two typical tests exemplified.
The anolyte contained 50 g/L H2SO4 in test Mg7 and 70 g/L H2SO4 in test Mg10.
The test temperature averaged 30 °C for both tests. The current density was 381 A/m2
for test Mg7 and 907 A/m2 for tests Mg 10.
The current efficiencies estimated based on the 88% purity of the magnesium
hydroxide produced in both tests were 44% and 66% for tests Mg 7 and Mg10,
respectively.
The results indicated that at comparable specific energy consumption values, practi-
cally near the margin of error, the amount of magnesium hydroxide produced increased
by 188% from test Mg 7 to Mg10.
The results are summarized in Table 6. Optimization testing is underway at the time
of submitting this paper.
Based on these results, combined with the full-scale operation mass and energy bal-
ances, the predicted specific energy consumption was 9.5 MWh/t captured CO2, trans-
lating into a CO2 carbon–footprint ratio of 4.4:1 as tons captured vs. tons generated [20].
4 Baseline Process Economics
The bench-scale data package was used to generate tentative economic model
predictions.
Table 6. Comparative example of electrolysis tests results.

Energy
Test Current Grams produced, pure equivalents of consumption
ID A V A/m2 H2 O2 H2SO4 Mg(OH)2 kWh MWh/tOH−
Mg 7 8.0 8.1 381 1.5 23.9 54.5 19.1 0.3 29
Mg 10 19.1 13.2 907 2.9 45.7 88.0 54.8 1.0 32
Table 7. Tentative full-scale process economics summary.

Nameplate, Capex, Opex
Annual capture, CO2 tpa 1,000,000
Daily capture, CO2 tpd 3221
Lifetime capture, CO2 t/deposit 140,389,519
Installed EW Power, MW 1536
Total installed Power, MW 1799
Annual throughput, “as is” tpy 1754,305
Daily throughput, “as is” tpd 5651
Operation Lifetime, years 140
Capex $1,231,961,959
Opex – per year $363,580,705
Additional credits, tpy
Green Hydrogen 40,854
Oxygen 648,493
Ni 2335
Co 110.2
Site remediation 1,754,305
Revenues USD$/year Rev, %
Carbon credit $250,000,000 44
Green Hydrogen $189,971,733 34
Oxygen $0 0
Ni $50,783,282 9.0
Co $4,547,097 0.8
Site remediation – residue only $70,172,190 12.4
Sub-total annual $565,474,302 100
Admin $28,273,715
Annual EBTA $173,619,883
Annual Pre-ROI, % 48
Amortization ex-tax, years 7.1
Inflation / deflation and price fluctuations ignored.
The capital and operating costs (“capex” and “opex,” respectively) were calculated
based on generally accepted industry practices [19, 21–29]. The predicted full-scale
process economics are summarized in Table 7. Accordingly, the capex of the full-scale
operation with 1000 ktpa CO2 removal nameplate was estimated to USD$ 1232 Bn. The
annual opex is estimated to USD$ 0.364 Bn pointing to a pre-tax annual ROI of 48%
with 7.1 years of capital amortization.
The overall level of projections accuracy at this stage of the project is tentatively
estimated to ±60% for the capex and ±40% for the opex. The immediate goal is achiev-
ing the accuracy levels required by the guidelines (BPGMP – NI 43-101) for the
Preliminary Economic Assessment (Capex ±25 to ±50% and Opex: ±25 to ±35%) by
the end of 2022.
From a project development viewpoint, the demonstration plant will create the pos-
sibility of advancing the project to feasibility.
5 Summary and Look Forward
• Column-simulated heap leaching test produced extractions of 84%, 90%, and 74%,
for magnesium, nickel, and cobalt, respectively, in 86 days. The asbestos reduction
was 81% overall, and complete removal at the top section of the simulated heap;
• Column-simulated vat leaching test produced extractions of 78%, 88%, and 66%,
for magnesium, nickel, and cobalt, respectively, in 29 days;
• Baseline iron and impurities removal testwork results produced from 95% to 100%
iron removal with 15–25% nickel losses, respectively;
• Baseline nickel and cobalt precipitation tests recovered from 80% to 88% of the
nickel and cobalt contained in the iron and impurities removal discharge barren
solution;
• The lower-than-expected precipitation performance in both precipitation-based unit
operations was attributed to basic salt formation due to the presence of elevated
levels of sodium in the lixiviant;
• Preliminary electrolysis testwork demonstrated that it was possible to produce a
sodium-free lixiviant for leaching along with magnesium hydroxide for carbon
capture;
• Based on the results to date, the mass and energy balances of the full-scale operation
predicted specific energy consumption was 9.5 MWh/t captured CO2, translating
into a CO2 carbon-footprint ratio of 4.4:1 as tons captured vs. tons generated;
• The capex of the full-scale operation with 1000 ktpa CO2 removal nameplate was
tentatively estimated to USD$ 1232 Bn. The annual opex is estimated to USD$
0.364 Bn pointing to a pre-tax annual ROI of 48% with 7.1 years of capital
amortization.
Subject to successful financing and permitting, the project is envisioned to be
“shovel-ready” in 2026. Following the targeted 2027 successful commissioning and
ramp-up, the required magnesium hydroxide production level for 1000 ktpa CO2 name-
plate capture capacity will be achieved in 2028. Incremental extension will be possible
Table 8. High level project development schedule and forecasting.

Year
completed 2022 2023 2024 2025 2026 2027 2028 2030 2032
Stage Pilot Demo Feasibility, Construction aC & R Expansion
Annual Low 1.2 ECPM and 300 1000 1500 2000
capture, Permitting
CO2 ktpa
Cumulative 1.2 301 1301 2801 4801
capture, kt
CO2
Lifetime 140,390 140,388 140,088 139,088 137,588 135,588
capture,
CO2 t/
deposit
Annual Low 2.1 526 1754 2631 3509
feed
throughput,
ktpa
Feed 246,286 246,284 245,758 244,003 241,372 237,863
reserves, kt
Operation 140 n/a 139 92 68
Lifetime,
years
C & R commissioning and ramp
a
due to the nature of the heap leach operation. On that basis, the nameplate capacity will
be doubled to 2000 ktpa CO2 starting in 2032 (Table 8).
Acknowledgements The authors wish to acknowledge funding support from the Energy Research
and Innovation Newfoundland and Labrador (ERINL). The authors wish to acknowledge the continued
support of the Planetary Technologies management to the ongoing metallurgical testwork program.
References
1. Bobicki ER, Liu Q, Xu Z, Zeng H. Carbon capture and storage using alkaline industrial wastes.
Prog Energy Combust Sci. 2012;38:302–20.
2. Hamilton JL, Sioban AW, Morgan W, Harrison AL, Turvey C, Paterson DJ, Dipple GM, Southam
G. Accelerating mineral carbonation in ultramafic tailings via direct CO2 reaction and heap leach-
ing with potential for base metal enrichment and recovery. Econ Geol. 2020;115:303–23.
3. Oelkers EH, Gislason SR, Matter J. Mineral carbonation of CO2. Elements. 2008;4:333–7.
4. Power IM, Harrison AL, Dipple GM, Wilson SA, Kelemen PB, Hitch M, Southam G. Carbon miner-
alization: from natural analogues to engineered systems. Rev Mineral Geochem. 2013;77:305–60.
5. Seifritz W. CO2 disposal by means of silicates. Nature. 1990;345:486.
6. Wilson SA, Dipple GM, Power IM, Thom JM, Anderson RG, Raudsepp M, Gabotes JE, Southam
G. Carbon dioxide fixation within mine wastes of ultramafic hosted ore deposits: examples from
the Clinton creek and Cassisar chrysotile deposits, Canada. Econ Geol. 2009;104:95–112.
7. Wilson SA, Harisson AL, Dipple GM, Power IM, Barker SSL, Mayer KU, Fallon SJ, Raudsepp M,
Southam G. Offsetting of CO2 emissions by air capture in mine tailings at the Mount Keith nickel
mine, Western Australia: rates, control, and prospects for carbon neutral mining. Int J Greenh Gas
Control. 2014;25:121–40.
8. Dupéré M, et al. Resources estimation of the nickel content in Asbestos Mines Tailings, Thetford
Mines, Quebec, Canada. A NI 43-101 TR prepared for Nichromet Extraction Inc., 15 Oct 2007.
9. Dundee Sustainable Technologies (DST): “Confidential Technical Report” commissioned on
behalf of Planetary Hydrogen, Apr 2021.
10. Mezei A, Ashbury M. Technical risk mitigation through rheometallurgical generation of process
and equipment design criteria. In: 2013 Annual conference of metallurgists, hydrometallurgy:
materials & equipment symposium, Montreal, 27–31 Oct 2013.
11. Mezei A, Ashbury M, Girard M, Canizares M, Bourricaudy E. A rheometallurgical approach to
liquid-solid separation. In: 2011 Annual conference of metallurgists, Montreal, 2–5 Oct 2011.
12. Verbaan N, Yu B, Brown J, Mezei A. Are REE exploration companies doing enough and/or the
right testwork to avoid plant failure? In: Conference of metallurgists, Vancouver, 2014.
13. Mezei A, Todd I, Molnar R. Can complex hydrometallurgical pilot plants effectively reduce project
risks? In: ALTA 2006: Nickel/Cobalt 8 Conference, Perth, May 2006.
14. Mezei A, Ashbury M. Rheometallurgical process risk mitigation through engineering data genera-
tion. In: Hydroprocess 2013 Conference, Gecamines, Santiago de Chile, 10–13 July 2013.
15. Guarnera BJ. Technical flaws in bankable documents. In: Assaying & Reporting Standards
Conference, The Oriental, Singapore, 10–11 Nov 1997.
16. McNulty T. Developing innovative technology. In: Mining engineering, Oct 1998.
17. Nice RW. Ramp up times – why are they never properly considered? In: ALTA 2002: Nickel/Cobalt
8 Conference, Perth, May 2002.
18. Molnar R. Hydromet piloting, what the mineral processor can expect. In: CIM Conference, Ottawa,
Jan 2006.
19. Canadian Institute of Mining, Metallurgy and Petroleum. CIM Best Practice Guidelines for Mineral
Processing (BPGMP). https://mrmr.cim.org/en/best-practices/mineral-processing/
20. Hydropower carbon footprint data from: https://impactful.ninja/the-carbon-footprint-of-hydropower/
21. Dixon S. The challenge of processing gold-copper ores. In: Proceedings of the Heap Leach
Solutions Conference, Reno, 14–16 Sept 2015.
22. Free M. Modelling heap leaching PLS from valuable mineral associations and leaching param-
eters. In: Proceedings of the Heap Leach Solutions Conference, Reno, 14–16 Sept 2015.
23. Echeverria N, et al. Hydraulic performance of a permanent heap leach with an intermediate drain-
age system. In: Proceedings of the Heap Leach Solutions Conference, Reno, 14–16 Sept 2015.
24. Horton M, et al. An investigation of reagent optimization for industrial-scale heap leaching of
chalcocite ore. In: Proceedings of the Heap Leach Solutions Conference, Reno, 14–16 Sept 2015.
25. Milczarek M, et al. Eleven-year performance evaluation of the evapotranspiration cover at Barrick
Goldstrike Mine’s AA Leach Pad. In: Proceedings of the Heap Leach Solutions Conference, Reno,
14–16 Sept 2015.
26. Sauceda J. Estimating costs of agglomeration in gold heap leaching. In: Proceedings of the Heap
Leach Solutions Conference, Reno, 14–16 Sept 2015.
27. Oxley A, et al. The Piauí Nickel heap leach project, Brazil. In: Proceedings of the Heap Leach
28. Short Course. Proceedings of the heap leach solutions conference, Reno, 14–16 Sept 2015.
29. “Confidential Technical Report” commissioned by Noram Engineers and Constructors, on behalf
of Planetary Hydrogen, June 2021.
Developing a Commercial Heavy Rare Earth
Processing Facility at Browns Range
Louis de Klerk1 (*) and Robin Jones1

1
Northern Minerals, West Perth, WA, Australia
rjones@northernminerals.com.au
Abstract. Northern Minerals Limited (NML) has been developing a Heavy Rare Earths
(HRE) project based on the Browns Range xenotime mineral resources since 2011.
The project has followed a rigorous development process going through exploration,
resource development, laboratory test work and laboratory pilot programmes, a large-
scale pilot program and scoping, pre-feasibility and detailed feasibility study stages.
The resource development, laboratory test work programmes, scoping study and
pre-feasibility study were completed from 2012 to 2014. A definitive feasibility study
based on processing 585,000 tpa of ore to produce 3100 tpa of total rare earth oxides
(TREO) in a rare earth carbonate (REC) was completed in 2015. Due to various
external factors, including the rare earth market and discussions with potential off-
take partners, NML formulated a new plan to construct and operate a large pilot plant
at Browns Range to test the technical and economic viability of developing a com-
mercial-scale operation at Browns Range.
The large-scale pilot plant, consisting of a 10 tph beneficiation plant and a
hydrometallurgical plant (cracking, leaching and purification), was constructed in
2018 at Browns Range and has been operated through to 2022, testing both circuits.
This pilot facility has enabled NML to produce sufficient quantities of both xenotime
concentrate (20–30% TREO) as well as a mixed rare earth carbonate (REC) at 50%
TREO for qualification at potential off-take partners with REC sold to several rare
earth refineries.
This paper will discuss the development process starting with the mineralogy
and resource base, the minerals of economic interest and the characteristics of these
minerals that shaped the process flow sheet. The stages of laboratory test work, labo-
ratory piloting and interaction with the project development through scoping, pre-
feasibility and detailed studies will be discussed.
Operating the Browns Range pilot plant has allowed NML to understand the
process in much more detail, and gain operational know-how, than was not possible
from laboratory-scale pilot test work. The decision to pilot at large scale has allowed
NML to resolve process and equipment issues that were not evident from laboratory-
scale test work, and has allowed customers and off-take partners to gain confidence
in the product quality, placing NML in a favourable position to develop a commercial
HRE production facility.
Keywords: Heavy rare earths · Xenotime · Piloting · Project development
https://doi.org/10.1007/978-3-031-17425-4_67
530 L. de Klerk and R. Jones
1 Previously Published Information
The development of the process flowsheet in the lead up to building a pilot plant on site
at Browns Range has previously been described in detail at the COM2014 conference in
a paper written by Hadley and Catovic [1] and a poster written by Griffith, Brown, Hadley
and Catovic [2]. Detail on the hydrometallurgical plant has been published by de Klerk,
Das and Gerbino at ALTA 2020 [3]. This work will only be summarised in this paper.
2 Geology of the Deposits
The NML deposits are unique because they are hydrothermal xenotime deposits, the
only known deposits of this type in the world [4].
Approximately 90% of the total rare earths in the Browns Range deposits occur in
the mineral xenotime with the remainder being light rare earths found in the mineral
florencite, with monazite present only in trace amounts. Xenotime and florencite are
present within a unique host rock mineralisation consisting predominantly of quartz,
muscovite, iron oxide and kaolinite.
This is a vastly different mineralogical style to other known sources of xenotime,
where it occurs as a minor constituent, compared to monazite, in beach placer deposits,
and also as a minor constituent in weathered crust deposits.
The NML deposits include Wolverine, Gambit, Gambit West, Area 5, Cyclops,
Banshee and Dazzler. The Wolverine, Gambit, Gambit West and Cyclops deposits dem-
onstrate strong geological similarities. The higher-grade Dazzler deposit discovered in
2018 has a different style of mineralisation associated with the unconformity contact
between the Gardner Sandstone and the Browns Range Metamorphics.
Both the Wolverine and Gambit West deposits have similar mineralisation styles
being xenotime hydrothermal breccia, but both deposits are complex with numerous
vein and breccia xenotime associations (discrete quartz veins, stockwork veins, crackle
vein selvedges, mosaic breccias and matrix infill). More than 20 different styles and
intensity of alteration, veining and brecciation have been observed in diamond core
demonstrating the complexity of the system. Silica and sericite alteration, with lesser
haematite, kaolinite, barite and sulphides, are variably associated with the mineralisa-
tion, but this differs within each deposit as well as between the deposits [5].
At Wolverine, the hosting Browns Range Metamorphics are a variable sequence of
meta quartz-lithic and arkosic arenites and conglomerates with minor interbedded
schists. The mineralisation at Gambit is hosted within a silica-haematite-sericite altered
structure and is defined by complex structure, alteration, variable silicification and
increased fracturing [5].
The Dazzler deposit is located at the unconformable contact between the
Mesoproterozoic Gardiner Sandstone and the underlying Archean–Palaeoproterozoic
Browns Range Metamorphics, a sedimentary package of meta-arkoses and arenites.
The high-grade mineralisation occurs immediately above the unconformity,
predominantly within an argillite/mudstone unit formed at the base of the Gardiner
Sandstone, and is dipping moderately (30°–40°) towards the southwest. The minerali-
Developing a Commercial Heavy Rare Earth Processing Facility at Browns Range 531
sation has been modelled based on rare earth element geochemistry and logged geology
and is related to the presence of hydrothermal xenotime.
These differences within deposits and from one deposit to the other have been shown
to influence the behaviour in the beneficiation circuit and the gangue minerals play a
large part in the differences in physical flow behaviour of the xenotime concentrate
when treated with sulphuric acid.
3 Mineral Resources
NML’s tenure around the Browns Range Project covers an area of approximately
2750 km2 and is centred on the Browns Range Dome, a major geological feature cover-
ing approximately 1500 km2. The Dome is located in the northern Tanami region,
within both Western Australia (WA) and the Northern Territory (NT).
This massive landholding is highly prospective for rare earth mineralisation with
Fig. 1 showing the existing mineral resources and a number of targets identified for the
ongoing exploration programme.
Since 2009, exploration by NML has been focused on the WA tenements, which has
led to the discovery and completion of rare earth mineral resource estimates at
Wolverine, Gambit West, Gambit, Area 5, Cyclops, Banshee and Dazzler (see Table 1).
From these deposits Wolverine and Area 5 have Ore Reserve estimates.
The Wolverine deposit, in the north of the project area, will form the bulk of the feed
for a commercial concentrator. The detailed mineralogy and recovery data in this paper
are based on processing the Wolverine ore through the pilot plant circuit, and similar
results have been achieved with Gambit West ore, the only other ore that has been pro-
cessed through the entire pilot plant at Browns Range.
4 Mineralogy at Browns Range
NML has several deposits at Browns Range that are planned to be mined in a commer-
cial operation and the mineralogy does change from one deposit to the next, but in all
of the deposits where test work has been done, the heavy rare earths are contained in
xenotime mineralisation and the beneficiation processes selected have been shown to be
effective at treating all the deposits with some variations to the process.
The mineralogy of the beneficiated xenotime concentrate shown in Table 2 is from
the Wolverine deposit, the major source of rare earths at Browns Range, and analysed
using QEMSCAN and by SEM, and is dominated by silica and xenotime.
Xenotime is the dominant rare earth mineral present associated with quartz occur-
ring either as inclusions (Fig. 2) or on quartz margins (Fig. 3). Some quartz also con-
tains muscovite. Xenotime tends to be in fine grained aggregates and only occasionally
as well formed single crystals. Xenotime within quartz ranges from 30 to 120 μm, or
where muscovite was also present, from less than 20 to 170 μm. Xenotime on quartz
margins ranges from 30 to 150 μm.
Fig. 1. Location of browns range pilot plant.
5 Properties of the Minerals Used in Developing the Flowsheet
The properties of the minerals considered in developing the flowsheet include the value
of the various contained elements, the X-ray attenuation, magnetic susceptibility, hard-
ness, density and flotation susceptibility.
Table 1. Northern minerals resources summary as on June 2021 [6].

TREO Dy2O3 Y2O3 Tb4O7 HREO TREO
Deposit Classification Mt % kg/t kg/t kg/t % kg
Wolverine Indicated 2.88 0.84 0.74 4.89 0.11 89 24,195,000
Inferred 1.97 0.89 0.76 5.15 0.11 88 17,588,000
Totala 4.85 0.86 0.75 4.99 0.11 89 41,786,000
Gambit West Indicated 0.12 1.8 1.62 10.98 0.22 94 2,107,000
Inferred 0.13 0.51 0.4 2.67 0.05 81 674,000
Totala 0.25 1.11 0.97 6.56 0.13 91 2,781,000
Pilot Plant Stockpiles Indicated 0.17 0.95 0.83 5.5 0.12 89 1,601,000
Inferred 0.03 0.26 0.2 1.35 0.03 79 89,000
Totala 0.20 0.83 0.72 4.8 0.1 88 1,690,000
Gambit Indicated
Inferred 0.21 0.89 0.83 5.62 0.11 96 1,878,000
Totala 0.21 0.89 0.83 5.62 0.11 96 1,878,000
Area 5 Indicated 1.38 0.29 0.18 1.27 0.03 69 3,953,000
Inferred 0.14 0.27 0.17 1.17 0.03 70 394,000
Totala 1.52 0.29 0.18 1.26 0.03 69 4,347,000
Cyclops Indicated
Inferred 0.33 0.27 0.18 1.24 0.03 70 891,000
Totala 0.33 0.27 0.18 1.24 0.03 70 891,000
Banshee Indicated
Inferred 1.66 0.21 0.16 1.17 0.02 87 3,484,000
Totala 1.66 0.21 0.16 1.17 0.02 87 3,484,000
Dazzler Indicated
Inferred 0.21 2.33 2.17 13.93 0.29 95 5,000,000
Totala 0.21 2.33 2.17 13.93 0.29 95 5,000,000
Totala Indicated 4.55 0.7 0.6 3.98 0.09 87 31,856,000
Inferred 4.68 0.64 0.54 3.67 0.08 88 29,998,000
Totala 9.23 0.67 0.57 3.81 0.08 87 61,857,000
TREO total rare earth oxides – Total of: La2O3, CeO2, Pr6O11, Nd2O3, Sm2O3, Eu2O3, Gd2O3, Tb4O7,
Dy2O3, Ho2O3, Er2O3, Tm2O3, Yb2O3, Lu2O3, Y2O3, HREO heavy rare earth oxides – Total of:
Sm2O3, Eu2O3, Gd2O3, Tb4O7, Dy2O3, Ho2O3, Er2O3, Tm2O3, Yb2O3, Lu2O3, Y2O3, HREO % HREO/
TREO × 100
a
Rounding may cause some computational discrepancies
5.1 Valuable Elements
The rare earth elements are associated with the xenotime and florencite minerals.
Xenotime contains the HRE, half of Nd and Pr, and little La and Ce, and this accounts
for most of the value in the ore (~97% of the basket price at January 2022 prices).
Florencite contains the majority of the light rare earths La, Ce and half of Nd and Pr.
Figure 4 shows the basket value percentage by element and also mineral type.
Dysprosium, terbium, lutetium and yttrium are the major contributors to the basket
value, accounting for 84% of the total value.
Table 2. Mineralogy of beneficiated xenotime concentrate.

Mineral Chemical Formula Weight %
Quartz SiO2 55.9
Xenotime – Y (Y, REE) PO4 35.9
Steel Fe, Cr, Ni 2.45
Florencite (Ce,Nd,La) Al3(PO4)2(OH)6 1.7
Muscovite KAl2(Si3Al)O10(OH,F)2 1.27
Goyazite SrAl3(PO4)2(OH)5(H2O) 0.45
Fe-Oxide Fe2O3/Fe3O4/FeO(OH) 0.42
K-Feldspar KAlSi3O8 0.40
REE Minerals Y,REE,Al,Si,O,OH mixed phases 0.30
Kaolinite Al2Si2O3(OH)4 0.19
Rutile TiO2 0.16
Barytes BaSO4 0.16
Fig. 2. SEM image of xenotime mineralisation in quartz.
5.2 X-Ray Attenuation and Ore Sorting
X-ray attenuation is the reduction in intensity of an X-ray beam as it transverses matter.

The attenuation coefficient is a measure of the quantity of the radiation absorbed by a
given thickness of absorbing material.
The mass attenuation coefficient is strongly dependent on the molecular weight of
elements in the material [7]. In practice this variation in mass attenuation coefficient
allows an X-ray Transmission (XRT) ore sorter to identify particles containing the valu-
able HRE and select these from particles containing mainly gangue elements.
Ore sorting at NML is based on XRT where the ore sorter measures the extent of
X-ray absorbed for each rock and then selects, based on an algorithm that includes the
Fig. 3. SEM image of xenotime mineralisation on margins of quartz.
Basket % Value by
Element
Dy
Tb
Lu
Y
Ho
Er
Fig. 4. Basket value by element and mineral type.
X-ray absorption and the size of the rock (for Steinert sorters that include laser sizing),
whether a rock is rejected or selected for further processing.
A test work programme was run starting in 2015, using a Tomra sorter on Wolverine
high-grade (WHG) ore, but insufficient work on this technology was done when the
pilot plant was built at Browns Range in 2017 to incorporate ore sorting into the
circuit.
The test work programme was later expanded to test both Tomra and Steinert sorters
and small-scale tests were done on the ore that was available on stockpiles at Browns
Ranges including High and Ore Grade material from the Wolverine and Gambit pits and
a Mixed Mineralised Waste stockpile formed from low-grade material from both
Wolverine and Gambit pits.
The small-scale tests were all successful and larger-scale tests were run on all of
these ore types. Stockpiles were sampled from several locations to produce approxi-
mately 20 tonne batches of ore that were fed through the crusher on site. Belt cut sam-
ples of about 50 kg were periodically taken from each ore type to produce a 1.5–2 tonne
sample for ore sorting tests. These samples were screened into two sortable size frac-
tions −75 mm to +25 mm, −25 mm to +10 mm, and an unsortable −10 mm size frac-
tion. Separate sorting tests were run on each sortable size fraction. Similar sorting
results were achieved for both Tomra and Steinert sorters, with the Steinert machine
having a marginally better performance for the larger size fraction.
The sorting results in the vendor laboratory tests for WHG ore are summarised in
Fig. 5. This shows that at a mass recovery of ~33% with over 90% recovery of the
TREO could be achieved.
Based on these favourable results an ore sorter circuit was designed and installed on
site at Browns Range and incorporated into the existing pilot plant. A Steinert sorter
with a 2 m-wide belt was selected and the sorter was set up so that the coarser size frac-
tion (−75 mm to +25 mm) could be fed down one side of the sorter while the finer
fraction (−25 mm to +10 mm) was fed down the other side of the sorter.
A series of sorting test programmes were run to test the sorting behaviour at continu-
ous feed and at rates approaching what was expected on a commercial plant. The sorter
behaved similarly to the performance predicted from vendor laboratory test results with
the exception that the feed contained more fines than measured from the feed material
in the laboratory test work. Figure 6 shows the pilot plant sorting behaviour compared
to that in the laboratory sorting programme. The inability of the ore preparation prior to
the sorter to process the increased fines material limited the rate at which the sorter
could be tested.
The mineralogy of the sorted streams is shown in Table 3 for WHG ore in the size
fraction −25 mm to +10 mm, where Eject Pass 1, 2 and 3 refer effectively to different
ore sorting settings. Drop 3 refers to the final rejected stream after 3 recovery steps.
Ore was repeatedly passed through the sorter at ever-increasing mass recovery set-
tings to obtain the ore recovery curve. Selected material is ejected to a product stream
and the remaining material drops to a reject belt. The rejected material from the first
Fig. 5. Sorting behaviour of Wolverine high-grade ore at vendor laboratories.

Fig. 6. Sorting behaviour of Wolverine high-grade ore at pilot scale compared to vendor laboratories.
Table 3. Mineralogy of sorted Wolverine high-grade ore.

Mineral Phase Concentration (mass %)
WHG WHG WHG WHG
−25 + 10 mm −25 + 10 mm −25 + 10 mm −25 + 10 mm
Eject Pass 1 Eject Pass 2 Eject Pass 3 Drop Pass 3
Quartz, syn (SiO2) 87.6% 93.9% 88.0% 92.3%
Xenotime-(Y) (YPO4) 5.1% 0.2% <0.1% <0.1%
Vivianite, syn 1.9% 0.6% 0.7% 0.7%
(Fe3(PO4)2(H2O)8)
Muscovite-2M1 0.5% 4.2% 4.8% 4.6%
(KAl2(Si3Al)O10(OH,F)2)
Kaolinite-1A 0.3% 0.9% 0.7% 1.0%
(Al2Si2O5(OH)4)
Florencite-(La) (Al3(La0.38 0.2% 0.2% 0.1% <0.1%
Sr0.20Pr0.14Nd0.12Ca0.12Ce0.03(
P0.97Si0.03O4)2(OH)6)
Lanthanite-(Nd) <0.1% <0.1% <0.1% 0.1%
(Nd2(CO3)3·8H2O)
Amorphous 4.4% 0.1% 5.7% 1.3%
pass is used as the feed for the second pass. Most of the xenotime recovery is in the first
cut from the sorter.
Figure 7 shows more detail of the effect of the sorter settings on the sorter perfor-
mance. WHG is high-grade ore (~1% TREO). Setting 1 is high grade and low recovery
and this shifts to setting 5 with low grade and high recovery with settings 2, 3 and 4 in
between these extremes. In practice the “best” settings appear to be in the mid-range
(settings 2–4).
Fig. 7. Sorting behaviour of Wolverine high-grade ore at different sorter settings.
The ore sorter is effective in recovering more than 90% of the heavy rare earth com-
ponents in the sortable sizes of ore in less than half of the mass. This allows low-grade
ore to be rejected without grinding and beneficiating and could also allow lower-grade
ores to be processed or allow the use of alternate mining methods where dilution is
increased to reduce mining costs.
5.3 Magnetic Susceptibility
Minerals are divided into three groups in terms of their magnetic susceptibility: ferro-
magnetic, paramagnetic and diamagnetic ones. Ferromagnetic substances have high
magnetic susceptibility and are easily magnetized by a magnetic field; paramagnetic
substances have low susceptibility and their interaction with the magnetic field is rather
weak; diamagnetic materials also have low susceptibility, but are, by contrast, pushed
out of the area of a strong magnetic field [8]. The target mineral xenotime is paramag-
netic and requires a high intensity magnetic field to be sufficiently magnetised for
extraction by magnetic separation.
The Browns Range ore contains some ferromagnetic material, such as magnetite
(and grinding media), and these must be removed by Low Intensity Magnetic Separation
(LIMS) (magnetic field of 3000 Gauss) before xenotime can be extracted by a Wet High
Intensity Magnetic Separator (WHIMS) (magnetic field of 11,000 to 13,000 Gauss).
The mineral containing the LREE in NML ore is florencite and this is not recovered
effectively by the WHIMS process, but given the low value of the LRE, this is not a
concern.
Table 4. Mineralogy of Wolverine high-grade ore by process step.

Concentration (mass%)
Magnetic Magnetic
separation separation Flotation Flotation
Mineral phase Feed concentrate tail concentrate tail
Quartz, syn (SiO2) 84.72% 50.87% 97.63% 35.23% 62.96%
Muscovite-2M1 2.20% 2.37% 0.21% 2.78% 3.80%
(KAl2(Si,Al)4O10(OH)2)
Kaolinite-1A (Al2Si2O5(OH)4) 0.01% 0.89% 1.27% 0.74% 0.44%
Rutile (Ti0.95Al0.05O1.975) 0.17% Trace Trace 0.43%
Xenotime (Dy), syn, dysprosium 1.82% 8.86% Trace 23.47% 0.25%
phosphate(V) | Dysprosium
Phosphate (Dy(PO4))
Haematite, syn (Fe2O3) 0.70% Trace Trace
Florencite-(Ce), syn 0.10% 0.45% 1.68% Trace
(CeAl3(PO4)2(OH)6)
Amorphous 11.16% 35.69% 0.89% 12.63% 28.31%
Mass 100% 22.7% 75.7% 3.3% 19.3%
The recovery of the mineral phases determined by X-ray Diffraction (XRD) by pro-
cess step is shown in Table 4 and indicates how the rare earth minerals xenotime and
florencite are concentrated in a beneficiation process comprising magnetic separation
with flotation applied to the magnetic separation concentrate.
5.4 Hardness/Grinding
Silica is much harder than xenotime. Measured by the Mohs scale silica is 7 versus 4.5
for xenotime. The ore has to be ground to below 100 μm to liberate the xenotime, but
the softer xenotime can be overground producing very fine xenotime (<20 μm) which
increases recovery losses in magnetic separation and flotation. The losses of recovery
in magnetic separation appear to be related to the particles produced by the pilot plant
grinding circuit that employs a single-stage SAG/Ball mill.
5.5 Density
The dense xenotime crystals have been recovered by gravity circuits in preliminary
tests and a gravity circuit incorporated in the grinding circuit may allow for a reduction
in overgrinding and increased beneficiation circuit recoveries.
Fig. 8. Project development timeline.
6 Project Development Timeline
NML initially followed a conventional project development pathway from resource

development and laboratory test work programmes, and the various study stages but
added the additional step of the large-scale pilot plant in 2016. Figure 8 provides an
overview of the project progression over time.
6.1 Laboratory Test Work Programmes
Much of the early work summarised below has been detailed in the 2014 paper listed as
Ref. [1]. Laboratory test work on a beneficiation circuit started in 2011 and that on the
hydrometallurgical circuit in 2012.
Beneficiation work was done at a range of metallurgical laboratories, including
Nagrom, ALS and SGS in Perth. These progressed from small laboratory-scale work on
drill core and surface sampled material to continuous piloting on larger samples (80t)
obtained from costeaning surface outcrops and drill core. A wide range of physical
separation processes were considered that were narrowed down to a Whole of Ore
Flotation process and a Hybrid process where a concentrate is produced by WHIMS
and that concentrate is upgraded by flotation.
Development of the hydrometallurgical circuit also started with small-scale bench
tests progressing to continuous bench-scale tests and larger-scale laboratory pilot tests.
The flow sheet evolved during this process, but a sulphation bake followed by water
leach and purification was settled on early in the development phase. The purification
process considered various alkalis to precipitate impurities and both carbonate and oxa-
late rare earth product precipitates were considered. These programmes were initially
undertaken at ALS and Australia’s Nuclear Science and Technology Organisation
(ANSTO), with the later detailed work and piloting being done at ANSTO. A range of
other laboratories and vendors were used for mineralogical work and to determine
physical properties of streams in the process to obtain design data.
6.2 Feasibility Study Progression
NML has undertaken a series of feasibility studies that developed in tandem with the
progress in metallurgical the test work. Scoping studies were completed on the benefi-
ciation and hydrometallurgical plants in 2013, and a PFS in 2014. A definitive feasibil-
ity study based on processing 585,000 tpa of ore to produce 3100 tpa of total rare earth
oxides (TREO) in a rare earth carbonate (REC) was completed in 2015.
In 2016, due to a combination of market conditions and discussions with potential
off-take partners, NML decided to build a large-scale pilot plant at Browns Range to
test the technical and economic viability of developing a commercial-scale operation at
Browns Range and to prove product quality and gain market acceptance for the NML
product. This pilot plant was built in 2017/2018 and was operated until early 2022.
7 Browns Range Pilot Plant
NML constructed a 10% scale pilot plant at Browns Range to test the technical and
economic viability of the process flow sheet developed from the laboratory-scale trials
and detailed in the 2015 DFS (refer to Figs. 9 and 10).
The pilot plant consisted of a beneficiation plant with crushing, grinding, magnetic
separation and flotation that was designed to process 10 tph of ore through the mill. An
ore sorter was added in 2020 that allowed up to 50 tph of ore to be crushed and sorted
with 10 tph of sorted ore being fed to the mill. The beneficiation plant produced about
540 kg/h of dried beneficiation concentrate at a grade of up to 30% TREO.
Fig. 9. Pilot plant flow sheet.

Fig. 10. Pilot plant layout.
An indirectly heated rotary kiln was used to bake the concentrate with sulphuric acid
producing a water-soluble sulphate that was water leached in the hydrometallurgical
plant before a purification process consisting of staged additions of ferric and alkali
(lime, magnesia) to precipitate impurities. Uranium was removed by ion exchange
before a REC product was precipitated by addition of soda ash.
A wide range of test programmes were run on the pilot plant. A test work pro-
gramme of a few hundred tests was planned before the pilot plant was commissioned
and additional test work was added as the process became better defined and as problem
areas were exposed.
At the start of the pilot plant programme, work was focused on areas of the process
and plant equipment that did not function well and towards the end of the programme,
the focus shifted to tests on sections of the plant as continuously operating systems
evaluating the effects of different feed types and grades and on the effects of different
operating parameters.
During financial year 2021, the pilot plant milled 3456 tonnes of ore and produced
40,406 kg of rare earth carbonate (REC) containing 1835 kg of dysprosium oxide and
233 kg of terbium oxide. This brought the total production of REC to 30 June 2021
from the Pilot Plant to 211,109 kg. The REC produced contained a total 103,731 kg of
rare earth oxide, which in turn contained 9751 kg of dysprosium oxide and 1245 kg of
terbium oxide.
Fig. 11. Rare earth recovery by element and process step.
Fig. 12. Major gangue recovery by element and process step.
7.1 Beneficiation Plant
Figures 11 and 12 show the recoveries of the beneficiation circuit, detailing the rare
earth and gangue elements reporting by process step.
The recoveries of the HRE are about 75% with the major losses occurring in mag-
netic separation. Most of the gangue elements are effectively separated from xenotime
as the TREO grade increases from ~1% in the feed to 30% in the final concentrate.
Uranium is an exception as U follows the REE in the beneficiation circuit.
The mineralogy shows that xenotime is disseminated in the ore and on the margin of
silica crystals as fine particles.
A Tescan Integrated Mineral Analyzer (TIMA), a scanning electron microscope
automated minerals analysis system, was used to measure mineral chemistry, libera-
tion, associations and grain size analysis. The TIMA analysis of the mill product (LIMS
Feed) shows very clearly how the xenotime is distributed as liberated particles and in
association with other minerals. Figure 13 shows that to a large extent xenotime is liber-
ated from other minerals, although about 20% is associated with other minerals.
The distribution of liberated xenotime by process step in Fig. 14 clearly shows that
the WHIMS Tail stream is predominantly fine particles and it is this tails stream which
accounts for most of the losses of xenotime from the circuit. Some of the production of
fine xenotime is during the mining and ore preparation processes for ore sorting but the
majority of fines generation is during grinding. Due to the fine nature of the xenotime
mineralisation, comminution is required to liberate xenotime, but it is clear that if the
Hybrid circuit is continued to be used, the generation of fines should be reduced as far
as possible to avoid product losses.
Liberated Xenotime
Xenotime plus Quartz
Xenotime plus Florencite
Xenotime plus White Mica
Xenotime plus Florencite plus

Quartz
Fig. 13. Mineral association of xenotime in milled ore.
8 Sulphation Kiln Operation
An indirectly heated rotary kiln was used for sulphation baking. An indirect unit was
selected as it allows for a much smaller kiln, a much reduced requirement for gas clean-
ing and reduced capital cost. The process consists of drying xenotime concentrate, inti-
mately mixing the concentrate with 98% sulphuric acid, and heating the mixture to
275 °C and holding that temperature for an hour.
The LIMS Non Mags are

fed to the Magnetic
separation and show a
coarser material than the
LIMS Non Mags
LIMS feed. Particles are
arrange from large to
small so some small
particles may not show up
The WHIMS Con is the

feed to flotation and
WHIMS Con
shows a lack of finer
particles
The WHIMS tail is lost

from the process and
WHIMS Tail clearly shows that most of
the losses are as fine
particles
The flotation concentrate

Flotation Con is composed of only large
particles.
Flotation tails are lost

Flotation Tail
from the process
Fig. 14. Mineral distribution of liberated xenotime by process step.
The process works well at the designed process conditions, but major problems
occurred with material flow and corrosion.
When the concentrate and acid is mixed, the concentrate slowly reacts with the acid,
and forms a toothpaste consistency that continues to harden with time and as the material
is heated. This mixture tends to build up a scale on surfaces, particularly hot surfaces,
and impedes the flow of the mixture into the kiln, along the kiln and also out of the kiln.
Heating the concentrate acid mix in the kiln causes a reaction between the concen-
trate and the acid but a substantial portion of the acid evaporates and is present in the
kiln atmosphere. Acid vapour condenses on all cool surfaces and condensate plus con-
centrate dust forms mud that sticks to all cool parts of the kiln and gas outlet piping.
An example of the extent of scale build-up is shown in Fig. 15 where the scale has
formed an internal ring in the kiln. In an indirect kiln, the heat has to be conducted
Fig. 15. Scale build-up on kiln internal surfaces.
through the kiln wall and the scale forms an insulating patch. This means that the tem-
peratures in the fire box have to be increased to maintain internal temperature and the
scale build-up continues at an ever-increasing rate.
The kiln descale cycle was initially a week to 10 days, but this was eventually
extended to a few months as a result of changes in feed system, discharge chutes, gas
scrubbing systems, operating conditions and the installation of a custom-scale removal
device.
Materials of construction remain a problem in the gas lines as the hot gases exiting
the kiln contain concentrated sulphuric acid/sulphurous vapours that condense and
become very corrosive at an intermediate temperature and high acid content.
9 Hydrometallurgical Plant Operation
The hydrometallurgical plant process that was designed from laboratory test work was
proved to work effectively. Getting the plant to operate continuously and produce on
specification product took a long time to achieve and this was partly as a result of the
kiln not being able to produce a continuous stream of product, forcing the hydrometal-
lurgical plant to run intermittently.
Production of high-purity product to tight specifications and achieving good recov-
eries relies to a large extent on effective solid liquid separation steps, continuous pro-
cess operation and consistent process control.
A large number of tests were done on the hydrometallurgical circuit testing different
types of solid liquid separation systems, feed strategies for sulphated ore to water leach,
and ion exchange circuit configurations to improve uranium removal.
Aluminium occurs in large quantities in the ore and although the bulk is removed in
the beneficiation circuit, sufficient Al remains to be a major gangue element in the
hydrometallurgical circuit. Solvent Extraction (SX) plants that separate the REC into
individual elements have very tight specifications on the Al content in REC. NML has
proven a precipitation step that removes the Al to acceptable levels but also results in
some REE losses that are partly recovered by recycling this material to the circuit.
Alternate SX processes to remove Al have also been developed.
Thorium is removed progressively during the purification process, with all of the
thorium being combined in the solid tails from water leach and the various precipitation
steps solid streams. Uranium is removed by ion exchange.
Treatment of off-gases and waste liquids and slurries from the hydrometallurgical
plant had to be re-evaluated and a more effective treatment system was required to man-
age water and tailings from the combined hydrometallurgical and beneficiation plant
tailings system.
10 Conclusion
Due to the unique hydrothermal style of xenotime mineralisation at Browns Range and
the inherent complexity of processing rare earth minerals, the development of a com-
mercial HRE processing facility at Browns Range has followed a more rigorous process
than would normally be required for more well established commodities.
Operating a large-scale pilot plant at Browns Range for 3 years has been a key addi-
tional step in the development and confirmation of the process flowsheet. It has pro-
vided NML with a much higher level of understanding of the process during continuous
operations which was not possible during laboratory-scale pilot test work. Additionally,
a significant amount of operational know-how has been gathered which will be included
in the detailed design of the commercial-scale HRE production facility.
Furthermore, production of substantial quantities of rare earth carbonate has pro-
vided NML the opportunity to engage with potential customers and off-take partners
both commercially and technically, allowing confidence in the product quality to be
established.
With the demand for rare earths set to grow in years to come, NML is now very well
placed to develop a commercial HRE production facility.
References
1. Hadley T, Catovic E. Beneficiation and Extraction of REE from Northern Minerals’ Browns Range
Heavy Rare Earth Project. In: COM2014 – Conference of metallurgists proceedings.
2. Griffith CS, Brown S, Hadley T, Catovic E. Browns Range Project. Radioactivity deportment stud-
ies – beneficiation and hydrometallurgical piloting. In: Conference of metallurgists poster 2014.
3. de Klerk L, Das G, Gerbino A. Rare earth concentrate processing and purification. Comparison and
optimization using OLI simulation. ALTA. 2020.
4. Gupta CK. Extractive metallurgy of rare earths. Boca Raton: CRC Press; 2005.
5. Northern Minerals Wolverine and Gambit West Geological summary.

6. Northern Minerals Limited Annual Report 30 June 2021 – News, Reports and Presentations –
Northern Minerals.
7. X-ray mass attenuation coefficients. https://physics.nist.gov/PhysRefData/XrayMassCoef/
tab3.html
8. Demidov IV, Vaisberg LA, Blekhma II. Vibrational dynamics of paramagnetic particles and pro-
cesses of separation of granular materials. Int J Eng Sci. 2019;141:141–56.
Integrating Carbon Capture in Mining Through
Metallurgy. Part 2: Integration into the Piaui
Nickel–Cobalt Laterite Operations
Anne Oxley1, 2 (*), Brock Battochio3, Mike Oxley1, Mike Kelland3,

Mykolas Gladkovas4, Sridevi Thomas4, Kevin Bradley4, Srinath Garg3, Greg Rau3,
and Alex Mezei3 (*)
1
Brazilian Nickel Plc, Kent, UK1
aoxley@brnickel.com
2
The Natural History Museum, London, UK
3
Planetary Technologies, Dartmouth, NS, Canada
alex@planetarytech.com
4
Abstract. Brazilian Nickel PLC (“BRN”) is developing its flagship Piauí Nickel–Cobalt
Battery Metals project in Brazil. The project is in an advanced staged with small-scale
early production commencing in mid-2022 at the PNP1000, which will produce 1400 tpa
nickel. The full-scale project will produce an average of 25,000 tpa of nickel and 900 tpa
of cobalt over the life of mine. Planetary Technologies (“PT”) is engaged in developing a
proprietary carbon capture technology aimed to be integrated within mining operations.
The enabling metallurgical process is based on classic unit operations involving sulfuric
acid leaching of the pay metals contained in suitable feedstocks, solution processing and
electrolytic acid regeneration concomitantly with alkalinity generation. When magne-
sium is present, it is leached along with other species, and it is by and large an undesired
side-reaction. According to the PT process, the leached magnesium is recovered electro-
lytically as hydroxide which is used for either ocean air capture (“OAC”) or direct land-
based carbon (DC) capture or a combination thereof. The main objective is to reduce the
atmospheric CO2 levels while generating credits from pay metals, such as nickel and
cobalt, along with the hydrogen produced during the electrolytic regeneration. The oxy-
gen produced during the same process can be used for the removal of iron and impurities
during the solution processing. The two companies are engaged in a partnership aimed to
capture circa 300,000 metric tons of carbon dioxide (CO2) per year expected to be emitted
by the full-scale Piauí operation. This target will be achieved in several stages. The paper
summarizes the results of the initial technical de-risking stage that produced the prelimi-
nary design criteria for the magnesium hydroxide scrubbing system.
Keywords: Ocean Alkalinity Enhancement · Magnesium hydroxide ·

Rheometallurgy · Critical Solids Density · Carbonation · Scrubbing · Piaui · Heap
Leaching · Planetary Hydrogen · Brazilian Nickel
https://doi.org/10.1007/978-3-031-17425-4_68
550 A. Oxley et al.
1 Introduction
Increasing energy demand has led to an increase in the use of fossil fuels, which has
consequently led to an increase in the atmospheric CO2 levels [1–3]. The advent of
Carbon Capture and Storage (CCS) technologies, such as those based on mineral car-
bonation aim to enable meeting the current and future energy demands in a “carbon-
neutral” manner [1, 2]. Serpentine-rich minerals are amenable feedstocks from which
Mg can be extracted and then transformed into hydroxide intended to be used for car-
bon capture and storage. For example, the acid leaching of serpentine and other
Mg-bearing minerals provides a plausible route for the generation of Mg(OH)2 which
can then be used to sequester CO2 [4, 5].
Planetary Technologies (PT) is developing a proprietary process aiming to treat
amenable Mg-bearing feedstocks using a dilute sulfuric acid stream produced from a
salt splitting electrochemical cell. The magnesium sulfate contained in the pregnant
leach solution (“PLS”) is precipitated as hydroxide, Mg(OH)2 within the electrochemi-
cal cell. The acid is used for leaching. The hallmark of the technology is its potentially
seamless integration ability into existing heap and conventional leach-based metallurgi-
cal operations. The PT process offers the advantage of producing the lixiviant (sulfuric
acid) as a by-product of the electrochemical process. The hydrogen produced during the
process is compressed and sold, while the oxygen can be used as an oxidant in the
impurity removal circuit. The Mg(OH)2 produced is used as a scrubber feed to offset the
CO2 emissions from the impurity removal circuit as well as from other emitter gas
streams.
Brazilian Nickel PLC (“BRN”) is developing its flagship Piauí Nickel–Cobalt
Battery Metals project with a JORC resource of 72 million tonnes at 1.0% Ni and
0.05% Co, with 73% of the resource in the JORC Measured category in Brazil [6, 7].
Extensive testing indicated good amenability to heap leaching. The results were vali-
dated at a large demonstration plant scale. As a result, the project is at an advanced
stage with small-scale early production commencing in mid-2022 at the PNP1000
plant, which will produce 1400 tpa nickel. The full-scale project will produce an aver-
age of 25,000 tpa of nickel and 900 tpa of cobalt over the lifetime of the mine (“LOM”).
The operation treats the laterite ore through a heap leach-based process in order to
recover nickel and cobalt. The heap leach PLS is directed to the iron and impurity
removal circuit using limestone for pH control. The resulting solution is subjected to
ion exchange followed by nickel and cobalt precipitation with magnesium oxide (MgO).
Certain processes in the BRN heap leach circuit have been identified as sources of
carbon emissions. They consist primarily of the lime kiln and the impurity removal
circuit. The purchased magnesia (MgO) used for the precipitation is a natural MgO
product and thus not an indirect CO2 emitter. An important objective of the project is to
include the capture of approximately 300,000 metric tons of carbon dioxide (CO2) per
year expected to be emitted by the full-scale Piauí operation. The PT-BRN partnership
seeks to integrate the PT technology into the heap leach operation at Piaui with the
objective of precipitating the magnesium contained in the barren-recycle bleed stream.
The resulting magnesium hydroxide will be used as the scrubbing agent.
Integrating Carbon Capture in Mining Through Metallurgy. Part 2: Integration… 551
7-10% wt. H2 SO4 eluant Mg(OH)2 pulp/slurry
Fe removal -
purification d/c
Ni and Co
Co and Ni IX Precipitation PT Tankhouse Mg(OH)2 Scrubber
circuits
Fe/Impurity
Removal off-
Bleed- gas CO2
Eluate streams (x2) to stream H2
Barren
Ni and Co PPTs
O2 to Iron and impurity removal
Raff-pond
Co and NI IX and PPT combined raffinate d/c stream
7-10% wt. H2 SO4 to HL
Fig. 1. PT module integrated into the BN-Piauí process flowsheet.
The schematic of the PT module integrated into the Piauí process flowsheet is
depicted in Fig. 1. Accordingly, the combined bleed stream that includes the discharge
barren streams from the nickel and cobalt precipitation circuits is directed to the elec-
trochemical cells. The acid stream is split as required between the ion exchange and
leaching circuits. The magnesium hydroxide is directed to the scrubber.
The paper summarizes the results of the initial technical de-risking stage that pro-
duced the preliminary design criteria for the magnesium hydroxide scrubbing system.
A previous paper [8] summarizes the results to date of the magnesium sulfate
electrolysis.

2.1 Metallurgical Thesis
Carbonation of Mg(OH)2 is thermodynamically possible but kinetically impaired due to

its low solubility in general, and compared to the magnesium carbonate in particular
(Table 1). However, kinetics can be improved based on reactive dissolution, a phenom-
enon derived from first principles [9]. This is because in case of a high-solid-density
reacting pulp, the system is in dynamic equilibrium, hence subject to Le Chatelier’s
principle. Accordingly, if a dynamic equilibrium is disturbed by changing the condi-
tions, the equilibrium shifts to counteract the change in order to re-establish itself. The
changing condition in this case is the carbonation, which consumes the existing dis-
solved hydroxide, instantly “making room” for more of it in solution. In addition, the
formation of water during the reaction accommodates for additional amounts of hydrox-
ide to dissolve as the reaction progresses.
In pure theory, ensuring a copious excess of Mg(OH)2 even as a pulp should result in
sufficient dissolved hydroxide in any moment in time as to significantly improve the
552 A. Oxley et al.
Table 1. Relevant solubility data at 25 °C.

Compound g/kg water Ksp SG solid
Mg(OH)2 0.0122 6E-12 2.345
Anhydrous MgCO3 0.139 2E-08 2.958
Dihydrate MgCO3*2H2O 2.825
Trihydrate MgCO3*3H2O 1.837
Pentahydrate MgCO3*5H2O 1.73
Mg(HCO3)2 0.77 -
Note: the specific gravity of various MgCO3 forms decreases with hydration
RHEOMETALLURGICALLYDEFINED FLOW-STABILITY DOMAINS
Ultra-stablewith near zero flow and separation

Resistance to flow, Pa ~ Stability
Stable plug-flow: grind discharge, paste, near

zero shearing
Critical Solids Density CSD~ Near

stable: low shear rates, thickener
underflow
Meta-stable: most high-to medium

shear unitops: adsorption, absorption,
carbonation, heat exchange-medium
shear rates
Unstable, fast settling, requires

high shear rates to maintain
suspension
Solids content, % wt. inpulp
Fig. 2. Rheometallurgically defined flow stability domains.
reaction rate. In practice, however, it is necessary to ensure that the basic requirements
for slurry handling, particularly for mixing and separation are met. This can be done
through the determination of the Critical Solids Density (CSD) of representative high-
density magnesium hydroxide pulp samples. CSD is an empiric rheometallurgical
parameter defined as the solids content above which a small increase of the solids con-
tent causes a significant decrease of its yield stress or flowability [10–12].
The CSD defines the most essential rheometallurgy domains common to metallurgi-
cal unit operations since they are all reliant upon shearing induced flow behavior. The
general example depicted in Fig. 2 illustrates the stability and flow profiles characteris-
tic for the various unit operations performed in commercially operated processing
plants. These unitops include mixing, separation, mass, and heat transfer. Common to
the CSD-defined stability domains is the continuous flow linking them. The degree of
stability is determined by several factors, including particle size, density, and solution
properties. The underlying flow behavior at a certain temperature is also determined
primarily by solids density and the rate of shearing.
In the particular case of this project, the pulp solids content (% wt.) corresponding to
the meta-stable flow domain was selected for carbonation based on experience. It is
located in the proximity of the CSD, immediately below the thickening zone. Subsequently,
the CSD of the magnesium hydroxide pulp test sample was determined by rheology test-
ing using the Concentric Cylinder Rotational Viscometry (CCRV) method.
This approach was consistent with the Real Data Modeling (RDM) methodology, a
commercial testwork, and flowsheet development approach that relies on defining the
overall metallurgical systems behavior based exclusively on both chemical and physi-
cal process responses as determined by testing. This method reduces the risks associ-
ated with the modeling-only-based approach [10, 12]. As described in this paper, the net
result was a technically derisked laboratory-scale design criteria package that justified
the advancement of the project to pilot stage.
2.2 Project Deployment Outline and Laboratory Test Matrix
A staged-gated approach was applied to minimize the initial and maximum risk
involved in process development [13–19]. The key stages and project deliverables
included the following:
• Laboratory-scale batch with two stage gates
Deliverables: decision enabling data
• Continuous laboratory-scale mini pilot with one stage gate
Deliverables: preliminary process design package (PDP)
• Continuous pilot with one initial stage gate
Deliverables: pre-feasibility and Demo PNP1000 implementation
This paper summarizes the results of the laboratory-scale testwork performed
according to the following laboratory-scale testwork program:
• Bulk magnesium precipitation under un-optimized conditions, to produce a syn-
thetic magnesium hydroxide cake that mimics the product generated by an electroly-
sis cell;
• Rheological characterization of the magnesium hydroxide pulp using concentric
cylinder rotational viscometry (CCRV) to determine its critical solids density (CSD)
and general flow-response;
• Determination of the reaction kinetics of the synthetically produced high-density
magnesium hydroxide pulp at % wt. slightly below the CSD, with carbon dioxide
fed at various flowrates according to pre-determined stoichiometry additions.
554 A. Oxley et al.
3 Testwork Conditions and Results Summary
3.1 Bulk Magnesium Hydroxide Precipitation
A bulk magnesium hydroxide (Mg(OH)2) precipitation test was conducted from a syn-
thetic solution of magnesium sulfate (MgSO4) containing about 50.9 g/L Mg. The
reagent consisted of a sodium hydroxide solution containing about 8% wt. NaOH. The
test was conducted on a dual-control basis: (i) stoichiometry excess of 5% (K = 1.05)
while ensuring that the final pH exceeded 10.5.
About 10.2 kg equivalent dry magnesium hydroxide precipitate was produced at a
recovery of 99.5%. It contained about 27% Mg corresponding to about ~65% pure
equivalent Mg(OH)2 grade product. The main impurities consisted of sodium (~5.2%
Na) and sulfate (~13% SO4). On that basis, it was estimated that a tentative formula for
the precipitate could be Mg(OH)2*Na2SO4*H2O.
The first four carbonation tests (CT1 through CT4) were conducted using the 65%
low-grade grade product. It was learned during the first four carbonation tests that a
significant portion of the sodium sulfate could be removed by washing. The resulting
precipitate displayed a purity of 89% Mg(OH)2. This sample was used for carbonation
tests CT5 through CT8.
3.2 Concentric Cylinder Rotational Viscometry
Concentric cylinder rotational viscometry (CCRV) rheology tests were conducted over a
range of flowable pulp densities of the washed magnesium hydroxide precipitate. The
resulting CSD was 23 wt% solids corresponding to a yield stress of approximately 65 Pa
under both sheared and un-sheared conditions. The sample was generally thixotropic up
to 18 wt% solids and transitioned into rheopectic behavior at 23.7% wt. solids. The rhe-
ometer experienced a torque overload when testing at 31 wt% solids. As a result, it was
decided to perform the initial carbonation tests at 23% wt. solids and at 300 RPM.
3.3 Bench-Scale Magnesium Hydroxide Pulp Carbonation
3.3.1 Test Setup
The schematic of the apparatus used to investigate land-based carbonation using mag-
nesium hydroxide pulp is shown in Fig. 3. The apparatus included an agitated reactor
containing the magnesium hydroxide pulp. A mixture of carbon dioxide and air at pre-
determined compositions were sparged into the reactor at pre-determined flow rates.
The unreacted carbon dioxide was scrubbed in a series of scrubbers using a sodium
hydroxide solution. This was necessary in order to accurately quantify the extend of the
actual carbonation reaction occurring in the reactor. The carbon dioxide volumetric
flow rate at the reactor exit was measured continuously. Tests CT1 through CT4 were
Fig. 3. Schematic of the batch bench-scale magnesium hydroxide pulp carbonation setup.
conducted without internal pulp recirculation. Tests CT5 through CT8 were conducted
with internal pulp recirculation.
3.3.2 Test Conditions
Base-line bench-top pulp carbonation kinetics tests CT1 through CT4 were conducted
in batch mode, aiming to investigate the reaction kinetics of the synthetically produced
magnesium hydroxide of 65% Mg(OH)2 purity. The hydroxide was repulped to 23%
wt. solids content. The agitation was set at 300 RPM during the carbonation test. The
carbon dioxide was introduced at various stoichiometry (K factor) of CO2 additions
versus the available equivalent pure Mg(OH)2 present in the reactor (“scrubber”).
The tests were conducted at ambient temperature and normal pressure, in closed /
baffled reactors. The gas mix was sparged near the tip of the impeller. Continuous pH
monitoring of the main reaction (scrubber) was performed. The circuit was fitted with
off-gas capture via an agitated NaOH scrubber.
The reaction progress was assessed based on slurry kinetics samples, with inorganic
carbon analyzed in both the liquid and solids phases. In addition, carbonate/bicarbonate
titration of the NaOH scrubber was performed for carbon balance checks. Metallurgical
balance was performed around each test.
Pre-optimization bench-top pulp carbonation kinetics tests CT5 through CT 8 were
conducted as a single extended pre-optimization run of about 4.5 h combined duration.
The test sample consisted of a purified magnesium hydroxide of 89% Mg(OH)2 purity
repulped at 15% wt. solids content. The agitation was reduced to 200 RPM. This was
because the removal of the sodium ions impacted a flow-adverse behavior to this sam-
ple. The suspected reason was that sodium ion being a surfactant, by its removal, the
magnesium content increased proportionally. As opposed to sodium ion, the magne-
sium ion is a coagulant and at times flocculant, somewhat comparable to calcium. The
net rheometallurgical outcome of this change in chemical composition was an increase
of the yield stress which in turn forced the reduction of the pulp solids content from
23% wt. to about 15% wt. in order to allow for adequate flowability (needed for mixing
and gas transfer). At the same time, the agitation rotation speed was cut back to 200
RPM in order to mitigate the expected rheopectic tendency. The test conditions are
summarized in Table 2.
556 A. Oxley et al.
Table 2. Summary of carbonation test conditions.

Test Hydroxide pulp Carbon dioxide gas stream
ID Run Solids a
Purity Equivalent CO2 Gas mix
<−> min % wt. % b
K NL/min NL/min c
vvm
CT1 60 23 ~65 0.89 3.39 3.42 1.20
CT2 0.10 0.339 0.3 0.12
CT3 0.49 1.694 1.7 0.60
CT4 0.05 0.169 0.2 0.06
The following four tests part of one single extended pre-optimization run
CT5 30 15 ~89 0.90 6.389 6.4 1.82
CT6 60 0.50 1.779 1.8 0.51
CT7 90 0.10 0.237 0.24 0.07
CT8 105 0.05 0.102 0.10 0.03
All tests were conducted at 25° C. The CO2 concentration in the gas mix was ~99% in all tests
Agitation at 300 RPM for tests CT1 through CT 4 and 200 RPM for tests CT 5 through CT8
a
Purity in dry material as equivalent Mg(OH)2
b
X – stoichiometry coefficient, <1 means deficit
c
vvm – volume gas per volume slurry per min
3.3.3 Tests Results and Interpretation

The results of the base-line tests CT1 through CT4 indicated the following:
• Absorption efficiencies varied inverse proportionally to stoichiometry factor K,
ranging from 1.7% to 16.7% for K values ranging from 0.9 and 0.05, respectively,
during the 1 h total runtime.
• The effective absorption rate ranged from 6.7 to 33.6%/h for the respective elapsed
times of 60 and 15 min, respectively.
• The interpretation of the results allowed initial establishment of the design criteria
strategy. Accordingly, the data indicated that at least three absorption stages would
be required in order to quantitatively remove the carbon dioxide during 12 h resi-
dence time.
The results of the pre-optimization tests CT5 through CT8 indicated the following:
• Absorption efficiencies varied inverse proportionally to stoichiometry factor K,
ranging from 2.3% to ~100% for K values ranging from 0.9 to 0.05, respectively,
during the various runtimes tested.
• The corresponding effective absorption rate ranged from 4.7 to 63.7%/h.
• The results allowed better definition of the preliminary design criteria derived from
the base-line testwork.
• The McCabe-Thiele diagram reflective of the data produced by tests CT5 through
CT8 indicated that at least three absorption stages would be required in order to
quantitatively remove the carbon dioxide during 6.3 h residence time (Fig. 4).
• The results allowed for the design of the small-scale continuous optimization run.
The results and interpretation are summarized in Tables 3 and 4, respectively.
CO2 Absorption efficiency versus its stoichiometry K

100
90
80
Pulp feed "Head tank" (stage #1) /
CO2-free gas OUT
Absorption rate, %/h
70
60
50 Pulp d/c tank
40 (stage 3) / CO2 IN
30
20
10
0
0.00 0.20 0.40 0.60 0.80 1.00
K
Fig. 4. McCabe-Thiele diagram for pre-optimized CO2 absorption conditions.
Table 3. Summary of carbonation test results.

Test a
Actual CO2 absorption Absorption ratios in grams CO2 per
ID Elapsed Eff’cy Rate kg L kg Eq’nt Mg(OH)2
<−> min % %/h pulp pulp Dry solids per kg per kg.h
CT1 15 1.7 6.7 0.5 0.5 2.0 3.1 12.3
CT3 60 7.61 7.6 1.1 0.6 5.0 7.7 7.7
CT2 60 16.7 16.7 0.5 0.6 2.3 3.6 3.6
CT4 30 16.78 33.6 0.3 0.6 1.2 1.8 3.6
The following four tests part of one single extended pre-optimization run
CT5 30 2.3 4.7 0.6 0.6 3.8 4.3 8.7
CT6 60 9.9 9.9 1.4 1.5 9.3 10.4 10.4
CT7 90 63.4 42.3 1.6 1.7 10.5 11.8 7.9
CT8 105 111.6 63.7 1.8 1.9 11.8 13.3 7.6
Actual elapsed time when the tests balance was calculated based on the kinetics samples
a
Overall, the results confirmed the metallurgical thesis postulating that carbonation is
possible by using a high solids loading magnesium hydroxide pulp subjected to ade-
quate shear mixing during the gas sparging.
4 Conclusions and Look Forward
• Base-line and pre-optimization bench-top pulp carbonation kinetics tests were con-
ducted in batch mode, aiming to investigate the reaction kinetics of a synthetically
produced magnesium hydroxide pulp. The tests were conducted under rheometal-
lurgically predefined conditions.
558 A. Oxley et al.
Table 4. Interpretation of the carbonation test results.

Residence Actual data and extrapolation to ~100% efficiency
Time Test K Rate Rest time Efficiency
h ID <−> %/h K*rate h %
0.25 CT1 0.89 6.7 6.0 3.0 20
1.00 CT3 0.49 7.6 3.7 3.0 23
1.00 CT2 0.10 16.7 1.73 3.0 50
0.50 CT4 0.05 33.6 1.76 3.0 100
Total residence time at equal pulp flowrate/tankage 11.9
0.5 CT5 0.90 4.7 4.20 1.6 7
1.00 CT6 0.50 9.9 4.97 1.6 16
1.50 CT7 0.10 42.3 4.23 1.6 66
1.75 CT8 0.05 63.7 3.19 1.6 100
Total residence time at equal pulp flowrate/tankage 6.3
• Base-line bench-top pulp carbonation kinetics tests CT1 through CT4 were con-
ducted using a magnesium hydroxide of 65% Mg(OH)2 purity. The hydroxide was
repulped to 23% wt. solids content and the carbonation was conducted at 300 RPM
agitated mixing. The results of these tests indicated that at least three absorption
stages would be required in order to quantitatively remove the carbon dioxide during
12 h residence time.
• Pre-optimization bench-top pulp carbonation kinetics tests CT5 through CT 8 were
conducted as a single extended pre-optimization run of about 4.5 h combined dura-
tion. The test sample consisted of a purified magnesium hydroxide of 89% Mg(OH)2
purity repulped at 15% wt. solids content. The agitation was reduced to 200
RPM. The results of these tests indicated that at least three absorption stages would
be required in order to quantitatively remove the carbon dioxide during 6.3 h resi-
dence time.
• Overall, the results confirmed the metallurgical thesis postulating that effective car-
bonation is possible by using a high solids loading magnesium hydroxide pulp sub-
jected to adequate shear mixing during the gas sparging. At the same time, the results
warranted the decision of advancing the project forward to the continuous optimiza-
tion stage.
Acknowledgments The authors wish to acknowledge the continued support of the BRN Board and
PT management to the ongoing metallurgical testwork program. The authors wish to acknowledge
funding support from the Energy Research and Innovation Newfoundland and Labrador (ERINL).
References
1. Alexander G, Maroto-Valer MM, Aksoy PG. Evaluation of reaction variables in the dissolution of
serpentine for mineral carbonation. Fuel. 2007;86:273–81.
2. Huesemann M. Can advances in science and technology prevent global warming ? A critical review
of limitations and challenges. Mitig Adapt Strateg Glob Chang. 2006;11:539–77.
3. Nduagu E, Bjorklof T, Fagerlund J, Makila E, Salonen J, Salomon J, Zevenhoven R. Production

of magnesium hydroxide from magnesium silicate for the purpose of CO2 mineralization-Part2:
Mg extraction modeling and application to different Mg silicate rocks. Miner Eng. 2012;30:87–94.
4. Tier S, Revitzer H, Eloneva S, Fogelholm CJ, Zevenhoven R. Dissolution of natural serpentinites
in mineral and organic acids. Int J Miner Process. 2007;83:36–46.
5. Tier S, Kuusik R, Fogelholm CJ, Zevenhoven R. Production of magnesium carbonates from ser-
pentinite for long term storage of CO2. Int J Miner Process. 2007;85:1–15.
6. Oxley A, et al. The Piauí Nickel heap leach project, Brazil. In: Proceedings of the Heap Leach
7. Brazilian Nickel Piauí Nickel Project – Brazilian Nickel PLC Website.
8. Kelland M, et al. Integrating carbon capture in mining through metallurgy, part 1. Leaching and
reclamation of Asbestos tailings: Thetford Mines carbon capture and remediation project. In:
Proceedings, 61st annual conference of metallurgists, COM 2002, Montreal, 21–24 Aug 2022.
9. Zao W, et al. Mathematical model of precipitation of magnesium carbonate with carbon dioxide
from magnesium hydroxide slurry. Comput Chem Eng. 2016;87:180–9.
10. Mezei A, Ashbury M. Technical risk mitigation through rheometallurgical generation of process
and equipment design criteria. In: 2013 Annual conference of metallurgists, hydrometallurgy:
materials & equipment symposium, Montreal, 27–31 Oct 2013.
11. Mezei A, Ashbury M, Girard M, Canizares M, Bourricaudy E. A rheometallurgical approach to
liquid-solid separation. In: 2011 Annual conference of metallurgists, Montreal, 2–5 Oct 2011.
12. Mezei A, Ashbury M. Rheometallurgical process risk mitigation through engineering data genera-
tion. In: Hydroprocess 2013 Conference, Gecamines, Santiago de Chile, 10–13 July 2013.
13. Verbaan N, Yu B, Brown J, Mezei A. Are REE exploration companies doing enough and/or the
right Testwork to avoid plant failure. In: Conference of metallurgists, Vancouver, 2014.
14. Mezei, Todd I, Molnar R. Can Complex hydrometallurgical pilot plants effectively reduce project
risks?. In: ALTA 2006: Nickel/Cobalt 8 Conference, Perth, May 2006.
15. Guarnera BJ. Technical flaws in bankable documents. In: Assaying & reporting standards confer-
ence, The Oriental, 10–11 Nov 1997.
16. McNulty T. Developing innovative technology. Min Eng. 1998;50:51–5.
17. Nice RW. Ramp up times – why are they never properly considered?. In: ALTA 2002: Nickel/
Cobalt 8 Conference, Perth, May 2002.
18. Molnar R. Hydromet piloting, what the mineral processor can expect. In: CIM Conference, Ottawa,
Jan 2006.
19. Canadian Institute of Mining, Metallurgy and Petroleum. CIM best practice guidelines for mineral
processing (BPGMP). Mineral Processing (cim.org)
Deep Eutectic Solvents for the Recovery of Metals:
Electrodissolution of Magnets for Rare Earth
Recycling
Guillaume Zante1 (*), Shannon Stodd1, Rodolfo Marin Rivera1, Jennifer Hartley1,
and Andrew P. Abbott1
1
School of Chemistry, University of Leicester, Leicester, UK
gz45@leicester.ac.uk
Abstract. Rare earth elements are essential for several key technologies, including
permanent magnets. High environmental impact of mining, together with increasing
volume of wastes, makes magnet recycling an interesting alternative for the supply of
the rare earth elements Nd, Pr, and Dy.
Keywords: Deep eutectic solvents · Rare earth recycling · Ionometallurgy
1 Introduction
Rare earth elements are essential for several key technologies, including permanent
magnets. High environmental impact of mining, together with increasing volume of
wastes, makes magnet recycling an interesting alternative for the supply of the rare
earth elements Nd, Pr, and Dy.
Hydrometallurgy allows the recycling of waste materials containing low concentra-
tions of rare earth elements, without needing high temperatures [1]. However, hydro-
metallurgical processing requires several pre-treatment steps and a large consumption
of hazardous chemicals to solubilize and purify metals.
Usually, magnets are first demagnetized and any adhesive remaining is removed.
Then, magnets are subjected to a high temperature process and crushed for metal libera-
tion, which leads to safety issues since the powder obtained can be very flammable [2].
Finally, the powder is dissolved with toxic and harmful mineral acids, allowing separa-
tion of the rare earths from iron with various techniques.
Deep Eutectic Solvents (DESs) are a mixture of a hydrogen bond donor and a hydro-
gen bond acceptor and are an environmentally friendly alternative to mineral acids, that
can be used to dissolve metals [3]. In this study, our aim is to reduce the number of
pre-treatment steps by recovering rare earths directly from the waste magnets using a
DES-based electroleaching process.
https://doi.org/10.1007/978-3-031-17425-4_69
562 G. Zante et al.
Demagnetized NdFeB magnets were provided by the University of Birmingham. These

magnets were surrounded by a black adhesive, and the metal content (wt.%) in the
magnet is 23% Nd, 67% Fe, 3% Pr, and 3% Dy, with the remaining mass being boron
and traces of other elements (Co, C, Al, etc.). After adhesive removal, the magnet is
dissolved using the electrochemical device described in Fig. 1.
Both electrodes are immersed in Ethaline (150 mL), a DES made of choline chloride
and ethylene glycol (1:2 molar ratio). The magnet (20 g) is placed in an Ir-oxide/
Ti-mesh basket, used as an anode. A copper plate (30 cm2) is used as a cathode. A con-
stant voltage of 5 V is applied between both electrodes, anodically dissolving metals
(M) from the magnet at the anode, according to Eq. (1):
Mmagnet

0
M nDES ne . (1)
The application of voltage also allows selective electrodeposition of iron at the cop-
per plate, according to Eq. (2):

ne Mcathode
0
M nDES . (2)
Although DESs have a large electrochemical window, oxidation of the solvent could
occur due to the high voltage applied. The current density is estimated to be 65 mA cm−2,
while the ohmic loss is around 430 mV.
Fig. 1. Scheme of the experimental device used for the electrodissolution of magnets in DES.
Deep Eutectic Solvents for the Recovery of Metals: Electrodissolution of Magnets… 563
The flowsheet developed is given in Fig. 2. First, the black adhesive surrounding the
magnet is dissolved using dimethylformamide (DMF), as illustrated in picture (A) of
Fig. 2, which shows a magnet before (left) and after (right) adhesive dissolution.
Adhesive removal allows a direct contact between the magnet and the electrode, man-
datory during electrodissolution. Under optimized conditions, removal of the adhesive
is complete in 15 min at 50 °C, using an ultrasonic bath. Without ultrasounds, removal
of adhesive usually takes several hours at temperatures higher than 250 °C [4]. Simple
filtration separates solid adhesive sheets from the solvent, while refluxing DMF allows
its re-use over several cycles.
Then, the magnet is placed in the electrochemical cell described in Fig. 1. Dissolution
of the magnet is evidenced by the colour of the Ethaline electrolyte which turns from
transparent to black during electrodissolution (picture (B)). The dissolved metal ions
are coordinated by chloride, while a part of the iron content of the magnet is deposited
on the cathode plate, as evidenced by the black deposit formed at the surface shown in
picture (C). Under these conditions, the dissolution rate can reach 1 g/h.
Separation of rare earth elements from iron can be achieved by the addition of oxalic
acid, since the rare earth will precipitate as oxalate complexes, while iron(III) oxalate
complexes remain soluble, as evidenced by the green colour of the solution (picture
(D)). When adding 10 wt.% of oxalic acid, the purity of Nd oxalate recovered reaches
64% but can be improved by reducing the amount of oxalic acid added. Upon exposure
to light, the iron(III) complexes will be gradually reduced into iron(II) oxalate complexes
Fig. 2. Scheme of the experimental device used for the electrodissolution of magnets in DES.
564 G. Zante et al.
that precipitate at the bottom of the solution (picture (E)), thus allowing to recover iron
at the end of the process. One can envisage the re-use of the Ethaline solvent after the
metals and oxalates are removed.
Overall, electrodissolution with DESs allows the leaching of metals from wastes
efficiently in the absence of mineral acids. Reuse of the solvents and the simple process
helps make the process economically attractive.
References
1. Jowitt SM, Werner TT, Weng Z, Mudd GM. Recycling of the rare earth elements. Curr Opin Green
and Sustain Chem. 2018;13:1–7. https://doi.org/10.1016/j.cogsc.2018.02.008.
2. Bandara HMD, Field KD, Emmert MH. Rare earth recovery from end-of-life motors employ-
ing green chemistry design principles. Green Chem. 2016;18:753–9. https://doi.org/10.1039/
C5GC01255D.
3. Pateli IM, Abbott AP, Jenkin GRT, Hartley JM. Electrochemical oxidation as alternative for dis-
solution of metal oxides in deep eutectic solvents. Green Chem. 2020;22:8360–8. https://doi.
org/10.1039/D0GC03491F.
4. Önal MAR, Dewilde S, Degri M, Pickering L, Saje B, Riaño S, Walton A, Binnemans K. Recycling
of bonded NdFeB permanent magnets using ionic liquids. Green Chem. 2020;22:2821–30. https://
doi.org/10.1039/D0GC00647E.
Heavy Rare Earth Ore Sorting at Browns Range
Louis de Klerk1 (*), Bill Rayson2, Paul Beukes3, and Angus Tod4
1
Northern Minerals, West Perth, WA, Australia
2
Total Earth Science Pty Ltd, Hillarys, WA, Australia
3
Steinert, Bibra Lake, WA, Australia
4
Minalyzer Pty Ltd, Belmont, WA, Australia
Abstract. Northern Minerals Limited (NML) has been developing a Heavy Rare Earths
(HRE) project based on the Browns Range xenotime mineral resources since 2011.
A pilot plant on site has tested metallurgical processes with ore obtained from near
surface deposits during a trial mining campaign. The process to upgrade the ore to a con-
centrate includes crushing, wet screening, an air knife to remove surface water, ore sorting
and then feeding an upgraded ore to a mill for grinding before magnetic separation and
flotation. The mill can process 10 tph of ore.
The ore sorting process used is based on a combination of X-ray Transmission (XRT)
and 3D laser. The benefit of ore sorting is that selection of material for further processing
can be done after crushing, but before fine comminution – this allows for both reduced
capital costs and operating costs, for an equivalent amount of saleable concentrate
produced.
Whilst conventional processes, such as magnetic separation and flotation, are nor-
mally well understood and metallurgical performance of these processes are routinely
used in determining modifying factors for Mineral Resource and Ore Reserve modelling,
this is not the case for ore sorting.
This paper will discuss the development of a modelling and test work program to link
X-ray Fluorescence (XRF) analytical data and laboratory-scale radiodensity measure-
ments from hand specimens and diamond cores, together with results of trial mining and
pilot-scale sorting test work, to prediction of how mined ores will sort at pilot and produc-
tion scales and how this information can be incorporated into the geological model for
resource modelling.
The aim of this process is to generate reliable ore sorting modifying factors for use in
mine planning for areas of deposits which are not available for trial mining, and more
broadly for deposits where only bench-scale ore sorting data are available.
Keywords: Ore sorting · Rare earths · Xenotime
1 Requirements for Incorporating Metallurgical Factors

into Mineral Resource Estimations
There is an excellent theoretical and first principles basis for XRT-based ore sorting to
work on the ores of Browns Range.
https://doi.org/10.1007/978-3-031-17425-4_70
566 L. de Klerk et al.
Fig. 1. X-ray images of mineralised and non-mineralised ore samples.
Radiodensity is the relative inability for photons at the X-ray energies of the electro-
magnetic spectrum to pass through a particular material, analogous to the term opacity
for the visual light spectrum.
The mineralisation at Browns Range is a relatively simple combination of silicates
and aluminosilicate host rock, with the valuable metals being directly contained within
xenotime mineralisation. At X-ray energies typically used in ore sorting applications,
the xenotime at Brown Range provides 10 times the X-ray radiodensity (‘brightness’ in
traditional X-ray images) than the un-mineralised host rock. Hence, xenotime minerali-
sation can easily be identified in X-ray images of Browns Range ore, and mechanical
ore sorters based on XRT principles should work well.
Figure 1 shows the difference in X-ray images for mineralised material (top row)
and un-mineralised material (bottom row).
1.1 Is the Metallurgical Process Well Tested in Technology or Novel

in Nature?
The use of ore sorting although not new is novel for xenotime mineralisation and the
process of linking the samples taken for test work to the whole ore body has to date
been insufficiently understood for these results to be incorporated into Ore Reserve
estimates.
1.2 Any Assumptions or Allowances Made for Deleterious Elements
With respect to ore sorting, deleterious elements are those that may mimic the effect of
the mineralisation, in this case providing a high density response where no valuable
metal exists. The test work being undertaken will attempt to identify deleterious ele-
ments by direct analytical methods as well as by directly measuring the pattern of
relationships between XRT density and valuable metal content.
Heavy Rare Earth Ore Sorting at Browns Range 567
1.3 The Existence of Any Bulk Sample or Pilot Scale Test Work

and the Degree to Which Such Samples Are Considered Representative
of the Orebody as a Whole
This point is the most pertinent to ore sorting and is the focus of current research at
Browns Range. Unlike typical metallurgical processes which act upon highly commi-
nuted and homogenised ores at micron to millimetre scale, ore sorting is conducted at
hand specimen scale (typically 10–90 mm) on crushed and screened heterogeneous,
unblended material.
Consequently, ore sorting pilot-scale tests require significant volumes of material,
typically several tonnes to several hundred tonnes, to be valid.
The large sample volumes required for testing ore sorting at pilot-scale preclude reli-
able ore sorting trials from being directly conducted on drill cores, especially on parts
of deposits at significant depth.
The approach adopted by Northern Minerals at Browns Range was to devise a set of
small-scale tests to test first principles parameters known or suspected to affect the
viability of XRT sorting; then build a numerical model which accounts for these param-
eters; thereafter demonstrate the validity of the numerical model by comparing to bulk-
scale tests; then analyse the first principles parameters across the deposit upon drill
cores and use this information to assess likely changes in performance of ore sorting
throughout the deposit.
The JORC CODE 2012 edition (Reporting code effective in Australia) provides spe-
cific requirements and guidance for considering metallurgical factors and assumptions,
and much of this test work is aimed at answering questions in line with this code [1].
2 Requirements for Ore to Sort
There are, essentially, four fundamental requirements to a successful ore sorting

outcome:
• There must be heterogeneity in the ore.
• A portion of the ore must be above cut-off grade.
• A portion of the ore must be below cut-off grade.
• The ore sorter must be able to identify, hence select, particles of ore above. a nomi-
nated cut-off grade reliably.
2.1 Distribution of Particle Grades
Not all material has a suitable grade distribution for ore sorting. Suitable grade distribu-
tions for ore sorting are generally extremely positively skewed.
Fundamentally, ore sorting will only work where the feed ore is suitably distributed
with respect to the mill cut-off grade. In most applications there must be a relatively
large population of material both above and below the mill cut-off grade to obtain ben-
efit from ore sorting.
If the input average grade is below the mill cut-off grade, hence technically waste,
ore sorting can only produce ore if some sub-portion of the feed is above the mill cut-off
grade.
Conversely, if the average grade of the input material is above the mill cut-off, hence
technically ore, the ore sorter can only reject waste if some sub-portion of the feed is
below the mill cut-off grade.
We can only sort ore that has a relatively large degree of heterogeneity at the indi-
vidual particle level, and the heterogeneity must spread the distribution of particle
grades across either side of the mill cut-off grade.
To illustrate this point, we have provided 6 Monte-Carlo simulations modelled in
Excel which show simulated grade distributions and cut-off relationships. We have pro-
vided both ore cases with a nominal 1% TREO average grade, and waste cases with a
nominal 0.25% TREO grade. It is assumed for this illustration that the mill cut-off
grade is 0.5% TREO. It should be noted that these parameters are chosen for ease of
illustration and do not relate to actual operating parameters at Browns Range (Fig. 2).
2.2 Quality of Relationship Between Particle Grade and Particle Sorting

Parameter
Northern Minerals is mainly using XRT principles to sort its ore at Browns Range. In
order to sort an ore, the relationship between radiodensity and grade must be of suffi-
cient quality so that the process of characterising an individual particle’s grade based on
its observable XRT response is reliable.
Bivariate analysis can be used to characterise the quality of this relationship and
make predictions about if an ore is sortable or not.
The charts in Figs. 3 and 4 have been left unitless. The important metric is the quality
of the relationship between grade and radiodensity, and the absolute units of grade and
radiodensity is not particularly important.
The units of radiodensity are in site specific units and are somewhat dependent on
experimental setup. Care must be taken to ensure the experimental setup is an accept-
able analogue to the Steinert ore sorter, as it is used on the particular ore type.
Figure 3 shows a radio density vs. grade plot from a deposit that is likely to have
acceptable performance using XRT ore sorting.
Figure 4 Radio density vs grade for an ore unlikely to sort well. Note, this is a syn-
thetic example to demonstrate an ore unlikely to sort well.
Fig 2. (continued) significant heterogeneity, would be a good candidate for ore sorting. (d) The log-
normal distributed ore grade distribution, hence with significant heterogeneity, would be an ideal can-
didate for ore sorting. Importantly, this distribution is analogous to Browns Range and many other ore
types. (e) The normally distributed waste grade distribution with limited heterogeneity is generally a
poor candidate for ore sorting. (f) The log-normal distributed waste distribution, hence with significant
heterogeneity, would be a candidate for ore sorting
Fig. 2. Simulations with different grade distributions. (a) The normally distributed ore grade distribu-
tion with limited heterogeneity would in most cases be a poor candidate for ore sorting. (b) The nor-
mally distributed ore grade distribution with mild heterogeneity would still be a relatively poor
candidate for ore sorting. (c) The bimodal distributed ore grade distribution, hence with
ORETYPE ‘A’
y = 6E-06x - 0.0068
R2 = 0.8456
RADIO DENSITY
GRADE
Fig. 3. Radiodensity vs. grade for an ore suitable for sorting.
ORETYPE ‘B’
RADIO DENSITY
y = 3E-06x - 0.0002
R2 = 0.0449
GRADE
Fig. 4. Radio density vs. grade for an ore unlikely to sort well. Note that this is a synthetic example to
demonstrate an ore unlikely to sort well.
3 Putting the Principles Together: Creating a Predictive

Numerical Model of Browns Range Sortability
In assessing sortability, quantitatively, the first principles parameters we attempt to

measure and evaluate are as follows:
• The grade distribution at 1 m sample support, which can be estimated conventionally
from conventional drilling and assaying.
• The grade distribution at smaller support (e.g. 10–75 mm) can be evaluated by
applying a change of support correction to the conventionally assayed drill sample
composites.
• The change of support correction can be informed by analysis using XRF directly
upon diamond cores at sample supports difficult to attain by conventional assays.
• The sorting process can be modelled as a function of selection threshold and prob-
ability of selection, which can be derived from bivariate analysis between paired
conventional geochemical analysis and XRT analysis.
It is a commonly understood principle that if a deposit is divided into many equiva-
lent volume parcels, then the variability of the population of the parcels is inversely
proportional to the size of the parcel.
When investigating a deposit for the purpose of developing a Mineral Resource
Estimate, Ore Reserve or mine plan, it is typical to perform a large amount of drilling
to sample the deposit. These drill holes are typically sampled and composited for esti-
mation into intervals around 1 m in length and 5–10 cm in diameter, representing
5–30 kg of material. These raw ‘composites’ are then averaged together to estimate the
grade of even larger parcels of the deposit, say 5 × 5 × 5 to 100 × 100 × 100 m repre-
senting hundreds of tonnes to thousands of tonnes of material.
In contrast to conventional Mineral Resource Estimation, we need to estimate the
grade distribution of a deposit at the ore sorting particle size (+10–75 mm in the case of
Browns Range), with masses nominally 5–500 g. It is therefore required that we run a
typical Mineral Resource Estimation in reverse, estimating the grade of the deposit at
smaller volumes than the informing drilling samples, and increasing the variability of
the estimated volume population.
To do that, rather than averaging drill hole composites together into blocks, we take
the raw drill hole composite grades and perform a volume variance correction to
increase the variability to a suitable distribution for simulating the ore sorting process.
This volume variance relationship was investigated at various deposits at Browns
Range by using XRF analyses on drill cores. Figure 5 shows the relationship at the
Wolverine deposit.
This relationship, once estimated, can then be used to derive an expected ore sorting
particle grade distribution from the 1 m composite sample distribution. Figures 6 and 7
show both 1 m and resulting estimate of small particle histograms for ‘Wolverine Ore
Grade’ and ‘Wolverine High Grade’ stockpile materials. Both distributions are highly
positively skewed and are good candidates for ore sorting.
The next part in the evaluation is to perform a paired analysis of a significant number
of ore sorting sized particles for both X-ray response and chemical composition so that
a bivariate analysis of the relationship between rare earth content and radiodensity
response can be conducted, such as that in Fig. 3. The material for this analysis can be
from diamond core.
The likely performance of ore sorting can then be predicted by applying a selection
threshold and error assumption to the particle grade distribution. Figure 8 shows the
model estimated recovery curves produced for Wolverine ore sorting with individual
Wolverine ore sorting test results also plotted.
Fig. 5. Change in variance with composite length at Wolverine.
Fig. 6. Wolverine Ore Grade histogram of TREO% grades.
Fig. 7. Wolverine High Grade histogram of TREO% grades.

Predicted Sorting Results

1
Cumulative Metal Recovery 0.9
0.8
0.7
0.6 WV HG
0.5
WV OG
0.4
0.3 WV HG TRAIL DATA
0.2 WV OG TRAIL DATA

0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Cumulative Mass Yield
Fig. 8. Predicted sorting results.
The performance of ore sorting can then be predicted by applying a selection thresh-
old and likelihood of selection assumption to the particle grade distribution. The spe-
cific model at Wolverine is constructed as follows:
• Each original drill composite, at 1 m support, is put into a spreadsheet list.
• The grade of the individual composite in the list is transformed using a change of
support procedure such that the population grade and variance now matches the vari-
ance expected at sortable particle size.
• The variability of expected radiodensity at any given TREO grade is modelled from
the bivariate analysis.
• A selection threshold is assumed.
• The likelihood of particle in the list to be selected is then assessed. The probability
of the particle being selected ranges from 0 to 1 and is a function of its listed TREO
grade, the distribution of possible XRT values at that given TREO grade and the
selection setting assumed.
• The likely metal recovery (expressed as a ratio) is the sum of all listed grades multi-
plied by the probability of selection, divided by the sum of all listed grades.
• The likely mass yield is the average of the probability of selection of all items in the
list.
Figure 8 shows the model estimated recovery curves for Wolverine ore sorting with
individual Wolverine ore sorting test results also plotted. The model curve is pro-
duced by running the method described above at multiple selection thresholds to
generate a curve representing various potential sorting outcomes.
Fig. 9. Hyperspectral scan image.
4 Grade Variation in Core
NML has investigated several alternatives to quantify the variation of the ore at a scale
suitable to predict sortable performance including manual pXRF analyses on a 10 mm
spacing, continuous core scanning using XRF and hyperspectral core scanning.
Examples of the information available for the two continuous core scanning technolo-
gies are compared below for the same box of core from the Banshee deposit.
The Hyperspectral Core Imager is an automated core imaging system that integrates
imaging spectroscopy, core photography and 3D laser profiling using a high-resolution
imaging spectrometer to map the material surface at a fine pixel scale delivering detailed
mineralogical image maps, mineral assemblages and textures.
The hyperspectral scan shown plots the HREE/Xenotime on a photo of the core in
Fig. 9. A variety of different properties of the core can be plotted, but this plot gives an
indication of the distribution of the HREE/xenotime in the core. The data provide rela-
tive grades of ores and this is available on a 10 mm scale over the length of the core. It
is evident that at a 10 mm size scale there are sections of ore containing little HREE and
there are sections of ore rich in HREE. The core is PQ, with an 85 mm diameter, and
ore breakage into sortable fractions 10–25 mm and 25–75 mm would clearly provide
pieces with high grade and some with low grade.
The continuous core scanning XRF scans a line along the core and the measurement
over a length is the accumulated analysis over that length. The assumption is that the
variation along a line would be representative of the variation in the other two dimen-
sions at that point. The analysis shown in blue columns is for yttrium (Y), the major
HREE in xenotime, but a range of other elements and physical properties are also mea-
sured. These data can be graphed on top of a photo of the core as shown in Fig. 10 and
are available as a data file. As with the hyperspectral core scan, ore breakage into sort-
able fractions 10–25 mm and 25–75 mm would clearly provide some material with high
grade and some with low grade. These data can be quantified as shown in the data from
BRBD004 in Sect. 5 in this paper.
5 XRF Scanning
Initially portable XRF devices were used, but labour and productivity requirements
needed NML to find a more productive solution.
Fig. 10. XRF scanning XRF.
Fig. 11. Minalyzer CS operating in a core shed.
The workings of the Minalyzer core scanner (Fig. 11) described in patent [2] are
designed for handling large volumes of drill samples and are capable of scanning drill
cores directly in core trays. It further describes the use of a laser technique generally
referred to as light detection and ranging (LiDAR), through which a 3D model of the
topology of the core and trays are generated, enabling the control and precision of a
continuous X-ray fluorescence (XRF) detector to generate XRF geochemistry and a
pseudo-SG measurement termed XSG. The generated 3D model constitutes the central
dataset that, apart from guiding the XRF detector, is also used to calculate the RQD,
measure structural features and derive bulk densities.
The geochemical dataset is generated by scanning drill samples with a high-intensity
2 cm × 1 mm XRF line beam in a continuous movement along the sample surface. This
data can then be quantified at any requested interval above 5 s of scanning time. A 3D
robotic system ensures that sample distance is maintained with a sub-millimetre preci-
sion. The elemental range spans between sodium (Na) and uranium (U) depending on
settings, environment and X-ray tube configuration.
The samples are cleaned and dried before entering the Minalyzer core scanner. Once
loaded into the chamber, the core is pre-scanned with a digital camera and a 3D LiDAR
scanner. The 3D camera produces a point cloud (.pcd file) which is used to make a 3D
model of the core tray and used by the instrument software to define a path that the
X-ray beam will take over the core.
Drill hole BRDB0004 was scanned at 1.5 mm/s, and a step size of 5 mm, with a
silver anode X-ray tube at 30 kV and 24 mA. The results were initially quantified by
fundamental parameters (FP), which were calibrated on two different pressed pellets,
lacking a binding agent.
5.1 Calibration
The special needs of this project for 1 cm quantified data required that the project core
was scanned at 1.5 mm/s. These small intervals were required due to the unconvention-
ally thin sections of mineralisation. Therefore, a new method for calibrating had to be
developed.
Normally the Minalyzer calibration procedure is to produce XRF geochemistry data
from two OREAS certified reference materials, through a fundamental parameters
method. This is termed primary data because it has not been matrix matched. Primary
data, on the same intervals as the lab assays from different ores and metallurgical sam-
ples, were used to find a calibration curve that was applied to all scanned intervals, and
the data are then termed as calibrated. There are two issues that we are trying to over-
come for Northern Minerals that arise from the normal calibration process:
1. The REEs are extremely enriched within the small veins, over 100 times higher than
the country rock. This means that if we were to take more than the veins for calibra-
tion purposes, the sample will have low values when compared to the concentrations
in the 1 cm data. Using this to calibrate normally, we would be extrapolating, rather
than interpolating the data when calibrating.
2. It is expensive, hard and subjective to assay on 1 cm sample interval. Even a small
special difference between the assayed sample and the scanned sample could be on
totally different grades.
A solution was to make artificial core out of Northern minerals concentrate from
various ore types and metallurgical samples (Fig. 12). This allowed the full range of
concentration to be covered on laboratory assayed samples. The artificial core was then
scanned by the Minalyzer CS. Figure 13 shows the Primary Minalyzer data vs. the
Fig. 12. Artificial core.

Fig. 13. Calibration graphs for the uncalibrated (left) and calibrated (right) Minalyze data vs. concen-
trate assay values.
concentrate assay values used to generate calibration curves in the left column. The
graphs in Fig. 13 show a strong correlation of the Minalyzer calibrated data vs. concen-
trate assays. All show a strong correlation with lab assays assuring that the results are a
fair representation of area that was scanned.
6 Modelling a Perfect Sort from XRF Scanning
Yttrium in NML deposits is contained in xenotime, a phosphate mineral that also con-
tains the heavy rare earth elements. Xenotime contains ~60% yttrium, and yttrium can
be accurately measured by XRF, so XRF is an ideal analytical method to track the valu-
able components of NML deposits.
The XRF analyses of Y, determined by XRF scanning, through the entire depth of
diamond drill hole BRDB0004 at Browns Range Banshee deposit are shown in Fig. 14.
The block marked in red is an interval proposed to be mined and was used to produce a
composite for ore sorting and later metallurgical test work.
There are areas of very low grade that may not be mined and there is high variation
of yttrium within the areas that have high grade, on the scale of rocks that will be sorted,
so that there will be individual rocks of high grade and low grade, making this deposit
a good candidate for ore sorting.
Fig. 14. BRDB0004 XRF Scan Y grade (ppm).
Fig. 15. BRDB0004 XRF Scan Y grade (ppm) over mining interval.
Looking at the proposed mining interval used to make the composite in more detail,
it is clear that there is significant variation in grade of Y on a 10 mm scale as shown in
Fig. 15.
If these data are rearranged from high grade to low grade, a distribution of the grades
in this interval can be plotted as shown in Fig. 16.
By selecting cumulative numbers of intervals, corresponding to mass recovery inter-
vals, the grade of select, rejected material and cut-off grade can be estimated for a per-
fect sort at the 10 mm scale.
Fig. 16. Distribution curve of Y in composite interval.
Fig. 17. Predicted Y recovery, grade and tail grade for a perfect sort.
These data are shown graphically in Fig. 17 as continuous curves.
7 X-Ray Attenuation and Ore Sorting
Attenuation is the reduction in intensity of an X-ray beam as it transverses matter. The

attenuation coefficient is a measure of the quantity of the radiation absorbed by a given
thickness of absorbing material.
The mass attenuation coefficient is strongly dependent on the molecular weight of
elements in the material, and the coefficients are shown for xenotime and the host rock
as shown in Fig. 18. This variation in mass attenuation coefficient allows an ore sorter
Fig. 18. Mass attenuation coefficient: host rock and xenotime.
to identify particles containing the valuable HRE and select these from particles con-
taining mainly gangue elements.
In practice the Steinert ore sorter has an X-ray source below the belt on which the
particles travel. The generated X-rays pass through the rocks and the remaining X-rays
are measured by a dual energy X-ray detector. Differences in the thickness of the
absorbing rocks are in this specific application dealt with by using data from the 3D
laser sensor that gives a 3D image of each particle. This allows the separation of
particles across both size fraction ranges (i.e. 10–25 mm and 25–75 mm) to be sorted
simultaneously by exactly the same sort program/algorithm.
8 Comparing Tests and Modelling
Core from two Banshee drill cores BRBD0004 and BRBD0003 were used to make
three composites for ore sort testing. These included the composite from BRBD0004
called BRDD0004 Upper that was used to illustrate the method of estimating the results
of a perfect sort.
The composite was produced by crushing the PQ drill core to −50 mm, with any
discs of material being returned to the crusher for further breaking. The material was
Fig. 19. Comparison of Predicted and Test Data Y Recovery.
Fig. 20. Comparison of Predicted and Test Data Select Y Grade.
then scrubbed and wet screened into fractions −50 to +25 mm, −25 to +10 mm and a
fines fraction of −10 mm. The two sortable fractions were then separately passed
through an XRT sorter at Steinert’s ore sorting test unit, and several cuts were made at
increasing recovery and reducing grade to determine a sorting curve. These data are
plotted on the recovery and grade curves estimated in Fig. 19.
The Y recovery curve matches the predicted recovery curve at low mass recovery for
both size fractions, but the recovery falls off for the −25 mm +10 mm fraction at higher
mass recoveries.
The Y2O3 grade curve corresponds well to the test data for both size fractions as
shown in Fig. 20.
9 Conclusion
The small-scale laboratory methodology of XRF scanning and X-Ray radiodensity

measurements on core and hand specimens appears to provide a viable way of estimat-
ing the sortability of NML xenotime ore at pilot scale. This implies this methodology
can be used to directly estimate modifying factors in Ore Reserve estimates where suf-
ficient volume of material for pilot-scale test work cannot be obtained.
The XRF scanning process has provided sufficient data to model ore sortability
based on XRF scanning. The data from XRF scanning of Banshee ore have correlated
well with laboratory ore sorting trials. Small-scale laboratory core sorting at vendor
laboratory can confirm the sorting predictions from XRF scanning.
Work will continue on developing this methodology to develop Ore Reserve estima-
tion parameters for the deeper deposits at Browns Range.
References
1. The JORC: Code 2012 Edition – Australasian Code for Reporting of Exploration Results, Mineral
Resources and Ore Reserves (jorc.org).
2. Blomdahl A, Artursson M, Veno K, Halonen M, Bragsjö N, Smajic A, Sernevi C. Device and method
for analysing samples when prospecting (International Publication No. WO2011/146014A1).
World Intellectual Property Organization; 2011.
Catalytic Dissolution of Metals Using Deep Eutectic
Solvents
Rodolfo Marin Rivera1 (*), Phil Hunt1, Guillaume Zante1, Jennifer M. Hartley1,
Karl S. Ryder1, and Andrew P. Abbott1
1
School of Chemistry, University of Leicester, Leicester, UK
ramr1@le.ac.uk
Abstract. Due to their high metal content, printed circuit boards (PCBs) are an attractive
resource for metal recovery. Until now, the dissolution and subsequent recovery of metals
from PCBs have considered either pyro- and/or hydrometallurgy. Such processing routes
are typically very energy intensive and are not exempted from producing a vast amount of
solid and liquid residues. Consequently, there is a need to develop more energy-efficient
and environmentally compatible processes. In the current study, the use of iodine/iodide
was investigated as a redox catalyst to solubilise copper, nickel and gold in the eutectic
system Ethaline with different water content. It was observed that the dissolution rates of
metals tended to increase when increasing the water content. Alternatively, FeCl3 and
CuCl2 were used as catalysts for selective metal recovery from PCBs in an eutectic sol-
vent formed from calcium chloride hexahydrate and ethylene glycol. The catalysts selec-
tively etched copper, allowing gold and nickel to be recovered by simple filtration.
Leaching rates determined with an optical profiler showed that CuCl2 enabled faster
leaching kinetics compared to FeCl3.
Keywords: Electrocatalysis · oxidising agent · deep eutectic solvents
1 Introduction
Printed circuit boards (PCBs) are essential for most electrical and electronic devices.
The growth in consumer electronics fueled by an increase in ownership and reduced
lifespan of electronic devices has led to a significant global increase in the volume of
electronic waste. It is estimated that ca. 55 Mt of e-waste is generated every year, while
an increase of ca. 30% is expected by 2030. The content of base metals in PCBs (such
as copper, tin, led, cadmium, chromium, zinc, nickel and manganese) can be as high as
30–40%, while gold and silver are also present at low but potentially significant con-
centrations. These figures render PCBs as an attractive ‘urban mineral resource’ for
metal recovery [1].
Pyro- and hydrometallurgy are often the processing routes used to recover metals
from waste PCBs. Yet, material processing operations are initially required to separate
reusable and hazardous component from these materials. This process is carried out
manually due to the diversity of connections mounted on PCBs, e.g. socket pedestal,
https://doi.org/10.1007/978-3-031-17425-4_71
584 R. M. Rivera et al.
through-hole device, surface mounted device, screw joint and rivet. Pyrometallurgy
allows processing PCBs at high temperatures (>300 °C) and, consequently, produces
hazardous gases due to the presence of mercury, brominated flame retardants and
(hydro)chlorofluorocarbons as part of the substrate [2]. High-temperature processes are
very energy intensive and face economical limits when treating low-grade raw materi-
als. Meanwhile, hydrometallurgical processes entail the consumption of large volumes
of lixiviants, and commonly used mineral acids, such as H2SO4, HCl and HNO3, have
poor selectivity, requiring further purification steps [3].
Deep Eutectic Solvents (DESs) can provide an alternative approach to recover met-
als from e-waste, as metals can be directly recovered from solid matrices by following
a similar principle to the conventional hydrometallurgical route, but with higher selec-
tivity and lower toxicity. Electrochemical oxidation has already been demonstrated as
an alternative methodology for solubilising metal oxides in DESs [4, 5]. However, the
methodology demands a high volume of chemicals. The use of a redox catalyst can
circumvent this issue, as it makes use of a chemical oxidising agent that can be recov-
ered in situ by an electrochemical reaction.
Initially, the use of iodine/iodide was investigated as a redox catalyst to solubilise pure
metals (in the form of wires with a defined diameter, ∅), such as copper (∅ 1.25 mm),
gold (∅ 0.25 mm), nickel (∅ 1.0 mm) and silver (∅ 1 mm) in the DES Ethaline (a 1:2
molar ratio of choline chloride and ethylene glycol, ChCl: EG) with varying water con-
tents. Metal dissolution tended to increase with increasing water content, due to the
reduction of the viscosity and the combined oxidative effect of both [I2Cl]− and [I3]−
species. Optical profilometry (Fig. 1) was used to measure the dissolution rate of met-
als, and images were constructed by determining the features of an image that are in
focus at different heights. These were then analysed to produce a reconstructed 3D
topography of the surface of the PCB block terminal. Line profiles were measured
across the surface, where the dissolution rate was measured using a flat surface of an
insoluble phase, e.g. the substrate holding the metallic deposit or an undissolved metal,
to provide a reference height.
Copper and gold had the fastest dissolution rate, particularly when the water content
was high (Fig. 2a and b, Table 1). On the contrary, a high water content tended to be
detrimental for nickel and silver solubilisation (Fig. 2c and d ), as their dissolution rate
tended to decrease when the water content was increased (Table 2), presumably because
the two metals tended to form passivating films. The dissolution of gold (Fig. 2b), in
particular, compares very favourably to cyanidation. For instance, the gold dissolution
rate under dry condition was 1.1 × 10−4 mol m−2 s−1 and it was further enhanced with
the addition of water (Table 2). Thus, the dissolution rate of gold in our system can be
40 times greater than the maximum gold dissolution rate in cyanide determined by
Wadsworth et al. (2000) of 0.69 × 10−5 mol m−2 s−1 [6]. Although the dissolution rate of
copper and gold was the highest (Table 1), extraction yields were less than 5% due to
mass transport limitations, as the solvent was gently stirred just to maintain a homoge-
Catalytic Dissolution of Metals Using Deep Eutectic Solvents 585
Fig. 1. 3D reflected light images of copper, nickel and gold after 60 min of etching in contact with
0.1 mol L−1 iodine as oxidising agent in ChCl: EG at 50 °C. The dark area corresponds to the epoxy
resin (substrate) holding the corresponding metal. All axes are in μm.
(a) (b)
50 20
50 vol.% 50 vol.%
25 vol.% 25 vol.%
40
10 vol.% 15
Dissolved, m
Dissolved, m
Dry
30 Dry
10
20
5
10
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time, min Time, min
(c) (d)
15 15
50 vol.% 50 vol.%
25 vol.% 25 vol.%
Dissolved, m
Dissolved, m
Dry
10 10 Dry
5 5
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time, min Time, min
Fig. 2. Etch depth of (a) copper, (b) gold, (c) nickel and (d) silver, over time with different concentra-
tions of iodine as oxidising agent in the ChCl: EG eutectic system at 50 °C with different water content.
The corresponding trend lines are also shown.
Table 1. Etching rate (in μm min−1) of pure metals in contact with iodine and different water contents
in ChCl: 2EG at 50 °C.
Water content
Vol.% Copper Gold Nickel Silver
50 0.597 0.139 0.028 0.033
25 0.370 0.105 0.037 0.021
Dry 0.166 0.065 0.083 0.073
Table 2. Molar dissolution rate (in mol m−2 s−1) of pure metals in contact with iodine and different
water contents in ChCl: 2EG at 50 °C.
Water content
Vol.% Copper Gold Nickel Silver
50 1.3 × 10−3 2.3 × 10−4 7.1 × 10−5 5.3 × 10−5
25 8.0 × 10−4 1.7 × 10−4 9.4 × 10−5 3.4 × 10−5
Dry 3.6 × 10−4 1.1 × 10−4 2.1 × 10−4 1.2 × 10−4
neous and continuous flow. Further improvement in extraction yields can be achieved
with intense agitation. It must be noted, however, that the use of the micro-leach optical
profiling technique aims solely to study the dissolution progress of solid/metal
compounds.
Alternatively, FeCl3 and CuCl2 were used as catalysts for selective metal recovery
from printed circuit boards (PCBs) in an eutectic solvent formed from calcium chloride
hexahydrate and ethylene glycol (CaCl2∙6H2O: EG at 1:1 molar ratio). This mixture has
higher conductivity, lower viscosity and lower glass transition temperature compared to
traditional DESs based on quaternary ammonium salts [7]. In addition, CaCl2 is pro-
duced on a large scale as a by-product in the Solvay process for the production of
sodium carbonate. FeCl3 and CuCl2 demonstrate a significantly high solubility in
CaCl2∙6H2O: EG, and both display reversible redox behaviour. These properties allowed
the use of both inorganic salts as strong, electrocatalytic oxidising agents [8].
The PCB studied in this research did not have any additional components, e.g. RAM,
chips and/or integrated circuits soldered to the substrate (Fig. 3). In order to evaluate
individual metal etching, kinetics were studied by considering the cross section of a
square-shaped (10 × 10 × 0.047 mm) block terminal. Copper was the major component,
followed by nickel and gold. The thickness of copper was 47 ± 2 μm, which is thin
enough to permit the transmission of electrical signals. Nickel had a thickness of
7.0 ± 0.1 μm and acted as protective barrier layer to prevent the migration and/or diffu-
sion of copper (located underneath) and gold (located on top) in case of oxidation. Gold
was located on top of the block terminal to provide contact points and edge connectors.
The thickness of the gold layer was approximately 1 μm.
The catalysts selectively etched copper, allowing the simultaneous release of nickel
and gold layers into the solution. Yet, gold and nickel were not dissolved and could be
easily separated from the solution by filtration. Optical profilometry was used to mea-
sure the dissolution of copper from the PCB. As shown in Fig. 4, the reddish-brown
area describes the copper layer and the white-greyish area describes the nickel layer.
The dark area corresponds to the substrate. Gold appears between the nickel layer and
the substrate.
Fig. 3. Gold-coated printed circuit board with block terminals of circular (in red circles) and square
(in green squares)-shaped geometries.
Fig. 4. 3D reflected light image of the cross section of a square-shaped block terminal (a) without
oxidising agents and with 1 mol dm−3 of (b) CuCl2 and (c) FeCl3 in CaCl2·6H2O: EG for 1 h. All the
axes are in micrometres (μm).
The dissolution rate of copper significantly changed with the concentration of

oxidising agent (Fig. 5). Thus, etching with CuCl2 took less time and was thus a more
effective oxidising agent than FeCl3 (Table 3). Nickel did not show any signs of dissolu-
tion and is assumed to passivate due to the relatively high water content in the eutectic
mixture.
The methodologies hereby presented offers a potentially autocatalytic method to
recycle electric and electronic waste. However, process optimisation is still required to
determine DES consumption and an optimal concentration of oxidising agent. Under
optimal conditions, both the solubilisation of metals and the auto-regeneration of the
catalyst could be possible by using oxygen from the surrounding air, preferably.
(a) (b)
80 80
with 1 mol L-1 with 1 mol L-1
with 0.01 mol L-1 with 0.02 mol L-1
60 60
Dissolved, m
Dissolved, m
R2 = 0.9712
R2 = 0.975 y = 1.213x
40 y = 0.639x 40
20 20 R2 = 0.988
R2 = 0.90
y = 0.091x
y = 0.094x
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time, mins Time, mins
Fig. 5. Etch depth of copper over time from the cross section of a square-shaped block terminal with
two different concentrations of (a) FeCl3 and (b) CuCl2 as oxidising agents in the CaCl2∙6H2O: EG
eutectic system at 50 °C. The corresponding trend lines are also shown.
Table 3. Copper etching rate (in μm min−1) in contact with different concentrations of FeCl3 and
CuCl2 in CaCl2∙6H2O: EG at 50 °C.
Oxidising agent Concentration (mol L−1) Rate (μm min−1) Molar dissolution rate (mol m−2 s−1)
FeCl3 1 0.64 1.4 × 10−3
FeCl3 0.01 0.09 2.0 × 10−4
CuCl2 1 1.21 2.6 × 10−3
CuCl2 0.01 0.09 2.0 × 10−4
Acknowledgements This project has received funding from the European Union’s Horizon 2020
research and innovation programme under the Marie Sklodowska-Curie grant agreement number
101026159 and also from the Faraday Institution (Faraday Institution grant codes FIRG005 and
FIRG006, project website https://relib.org.uk) and the UKRI Interdisciplinary Circular Economy
Centre for Technology Metals, Met4Tech project (EP/V011855/1).
References
1. Zeng X, Mathews JA, Li J. Urban mining of E-waste is becoming more cost-effective than virgin
mining. Environ Sci Technol. 2018;52:4835–41. https://doi.org/10.1021/acs.est.7b04909.
2. Cui J, Zhang L. Metallurgical recovery of metals from electronic waste: a review. J Hazard Mater.
2008;158:228–56. https://doi.org/10.1016/j.jhazmat.2008.02.001.
3. Binnemans K, Jones PT. Solvometallurgy: an emerging branch of extractive metallurgy. J Sustain
Metall. 2017;3:570–600. https://doi.org/10.1007/s40831-017-0128-2.
4. Abbott AP, Harris RC, Holyoak F, Frisch G, Hartley JM, Jenkin GRT. Electrocatalytic recovery of
elements from complex mixtures using deep eutectic solvents. Green Chem. 2015;17(4):2172–9.
https://doi.org/10.1039/C4GC02246G.
5. Abbott AP, Frisch G, Gurman SJ, Hillman AR, Hartley J, Holyoak F, Ryder KS. Ionometallurgy:
designer redox properties for metal processing. Chem Commun. 2011;47(36):10031–3. https://doi.
org/10.1039/C1CC13616J.
6. Wadsworth ME, Zhu X, Thompson JS, Pereira CJ. Gold dissolution and activation in cyanide
solution: kinetics and mechanism. Hydrometallurgy. 2000;57(1):1–11. https://doi.org/10.1016/
S0304-386X(00)00084-0.
7. Hartley JM, Allen J, Meierl J, Schmidt A, Krossing I, Abbott AP. Calcium chloride-based sys-
tems for metal electrodeposition. Electrochim Acta. 2021;402:139560. https://doi.org/10.1016/j.
electacta.2021.139560.
8. Rivera RM, Zante G, Hartley JM, Ryder KS, Abbott AP. Catalytic dissolution of metals from
printed circuit boards using a calcium chloride–based deep eutectic solvent. Green Chem.
2022;24:3023–34. https://doi.org/10.1039/D1GC04694B.
Neodymium(III) and Lanthanum(III) Separation
by a Flat-Sheet Supported Liquid Membrane System
Lin Li1, Krystal Davis1, Aaron King1, Mauro Dal-Cin1, Andrzej Nicalek1,
and Ben Yu1 (*)
1
National Research Council of Canada, Ottawa, ON, Canada
Ben.Yu@nrc-cnrc.gc.ca
Abstract. The interest in neodymium (Nd) is increasing due to its vital role in high-tech
applications, such as permanent magnets (Nd2Fe14B alloys). The extraction of Nd from
ores and recycling of Nd from waste products, such as batteries, demand effective and
efficient Nd separation from other rare-earth elements. Our research used a supported
liquid membrane with different extractants to extract and separate Nd and La. The effect
of type of extractant, extractant concentration, and the auxiliary agent on Nd/La extrac-
tion and separation were determined and are discussed. The results show effective extrac-
tion of Nd and promising selectivity of Nd against La using P507 as the extractant
compared to P204 and TBP. The extractant concentration played a significant role in
enhancing Nd extraction and separation. When the initial concentration of La(III) and
Nd(III) was 1000 mg/L, 40/80 (v/v)% P507/D80, initial feed solution pH = 3, strip solu-
tion of 3 M HCl, after 6 h 99.8% of Nd(III) was extracted, while the extraction of La(III)
was 22.2%. The tests at 5% (v/v) of P507 showed the highest Nd(III)/La(III) separation
factor of 44.25.
Keywords: Supported liquid membrane · Solvent extraction · Neodymium(III) ·

Lanthanum(III) · Extractant
1 Introduction
Rare-earth elements (REEs) have widespread applications in catalysis, metallurgy,

ceramics, medicine, and electronics [1–3]. Recently, rapidly increasing high-tech appli-
cations have further increased the demand for REEs with their unique physical proper-
ties, including hybrid cars, wind turbines, and compact fluorescent lights [4–6]. Among
all the REEs, the interest in neodymium is particular risen due to the forceful demand
for commercial high-strength permanent magnets based on Nd2Fe14B alloy and solid-
state lasers used in the medical industry [7–9]. Besides REEs containing ore, Ni-metal
hydride (Ni-MH) batteries are a potential secondary resource of neodymium [10–12].
In one example, spent Ni-MH batteries were subjected to acid leaching. After removing
the Ni(II), Co(II), and Mn(II) either by electrodeposition or by solvent extraction, La
and Nd were left in the leaching solution ready for separation [13].
https://doi.org/10.1007/978-3-031-17425-4_72
592 L. Li et al.
However, separating and purifying REEs can be difficult due to their similar chemi-
cal and physical properties [14]. The solvent extraction method has been commercially
employed as the principal technique for separating REE. Various extractants have been
reported to extract and separate REEs [8, 15–18]. Among the extractants, acidic organo-
phosphorus extractants have been studied the most, such as di-2-ethylhexyl phosphoric
acid (DEHPA; P204) [19] and mono-(2-ethylhexyl) phosphoric acid (EHEHPA; P507)
[15, 19–21], as well as neutral extractants, such as tri-n-butyl phosphate (TBP) [13, 22,
23], and tri-n-octyl phosphine oxide (TOPO) [13, 23]. Disadvantages of conventional
solvent extraction include third phase formation, extractant loss, and poor selectivity
[13]. In addition, the separation requires strict control of the phase variables.
The supported liquid membrane (SLM) separation techniques have attracted atten-
tion due to the superiority of low solvent inventory, ease of operation, and simultaneous
extraction and stripping [24–26]. REEs separation using the SLM method can be
straightforwardly seen as a simultaneous three-step process: (1) molecule extraction
from feed phase to the SLM; (2) diffusion through SLM; and (3) stripping process to
the received phase. The selective transport is based on the different permeability of
cations. Flat-sheet supported liquid membrane (FSSLM) is chosen in this research for
Nd extraction from La/Nd binary solution with the different acidic organophosphorus
extractant, including P204 and P507. Neutral extractant TBP was also employed to
investigate the Nd extraction. The effect of type of extractant, extractant concentration,
feed solution acidity, and auxiliary agent on La/the extraction and separation of La/Nd
have been studied in detail.
2 Experimental
2.1 Reagents and Materials
Chemical reagents used were A.R. grade. La2O3 (≥99.9%, Sigma-Aldrich) and Nd2O3
(99.9%, Sigma-Aldrich) solid powder to prepare REEs salts solution by dissolving in a
certain amount of acid (HCl, H2SO4, and HNO3) at 80 °C and cooling down to room
temperature. A microporous polypropylene membrane, 5 × 5 cm2 effective surface area,
55% porosity, 0.209 × 0.054 μm2 pore size, and 25 μm thickness, was used as the sup-
port. The extractant was P507, P204, and TBP (Kopper Chemical), and the diluent was
D80 (Exxsol™ D80).
2.2 Extraction Experiment
The FSSLM experiments were carried out in a two-compartment cell. The hydrophobic
polypropylene membrane was immersed in P204 (P507; TBP)/D80 mixture with a pre-
determined % (v/v) and treated with an ultrasound bath for 10 min. The membrane
pores were filled with carrier solution by capillary action. The membrane was then
impregnated in the organic solutions and kept in the petri dish for 12 h before the
Neodymium(III) and Lanthanum(III) Separation by a Flat-Sheet Supported Liquid… 593
extraction experiment. After removing from the carrier solution, the excess carrier solu-
tion on the surface of the membrane was then wiped out with tissue papers. Subsequently,
the loaded membrane was clamped between the two compartments and fixed firmly
using a gasket to avoid possible solution leakage from one compartment to the other.
Finally, the feed and strip solutions were added into the two separate chambers. The
feed solutions were 250 mL acidic solution containing La(III) and Nd(III), and the strip
solutions were acidic solutions with higher acidities.
The solutions were stirred at 350 rpm for 6 h at room temperature (20 ± 2 °C).
Samples from both feed and strip solutions were taken at 0, 0.5, 1, 1.5, 2, 3, 4, 5, and
6 h for La(III) and Nd(III) concentration measurement. An amount of 250–1000 μL
solutions from both the feed and strip were withdrawn and diluted. Their concentrations
were determined by ICP-OES (Inductively coupled plasma optical emission spectrom-
etry) using a Perkin Elmer Optima 3000. Before and after the separation tests, the mem-
branes were subjected to elemental analysis via Portable X-Ray Fluorescence
Spectroscopy (PXRF). The P concentration from the PXRF measurement reflects the
content of P507 loaded on the membranes.
La and Nd extraction using acidic organophosphorus extractants (abbreviated as H2A2)

is a cation exchange process. The extraction of REEs from feed solution is accompa-
nied by the release of three hydrogen ions from the extractants, as shown in Eqs.
(1)–(2):
La 3 3( HA)2 LaA3 ( HA)3 3H (1)
Nd 3 3( HA)2 NdA3 ( HA)3 3H (2)
The Nd and La extraction efficiency can be quantified as the extraction fraction for a
single REE extraction, illustrating the percentage of Nd(III) or La(III) transported into
the strip solution from the initial feed solution. To simplify the process, the concentra-
tion at the membrane boundaries was neglected. The equations can be written as
La( III )Strip

Extraction%La ( III ) 100% (3)
La( III )Feed;0
Nd ( III )Strip
Extraction% Nd ( III ) 100% (4)
Nd ( III )Feed;0
where [La(III)]Feed; 0 and [Nd(III)]Feed; 0 represent the initial Nd(III) and La(III) concen-
tration in the feed solution, respectively. To define the selectivity for the Nd(III)/La(III)
594 L. Li et al.
binary mixture, and therefore evaluate the feasibility of separating La(III)/Nd(III) using
SLMs, the separation factor was also determined:
Extraction% Nd ( III ) Nd ( III )Strip La( III )Feed;0

a Nd ( III )/ La ( III ) (5)
Extraction%La ( III ) La( III )Strip Nd ( III )Feed;0
3.1 Effect of Type of Extractant
Extraction and separation of Nd/La were first investigated as the function of different
extractants: P204, P507, and TBP. The initial pH of the feed solution was 3 with
1000 mg/L La(III) and Nd(III). The strip solution was a 3 M HCl solution. The extrac-
tor was 20% (v/v) of P507 or P204 or 60% (v/v) TBP in D80 diluent. The extraction and
separation of Nd/La results over a 6 h test are shown in Fig. 1.
Figure 1a, b shows extraction percentage of Nd and La by different extractants over
6 h, respectively. It is evident that the Nd extraction rate has the order of
Fig. 1. Effect of extractant on (a) Nd and (b) La extraction and (c) separation factor.
Feed: 1000 mg/L La(III) and Nd(III) in pH = 3 HCl solutions
Extractant: 20% (v/v) P507, P204, and 60% (v/v) TBP; Strip: 3 M HCl solution.
P507 > P204 > TBP. After 6 h, the Nd extracted were 88.8%, 63.5%, and 34.2%, respec-
tively. On the other hand, the efficiency of La transfer was the opposite for the 3 extract-
ants, as the order of TBP > P204 > P507, with La extracted being 13.7%, 20.8%, and
35.0%, respectively. Interestingly, P507 has the highest extraction rate for Nd and the
lowest for La. The separation selectivity for the binary Nd/La mixture was then calcu-
lated. The separation factor at 6 h was calculated based on Eq. (5), as shown in Fig. 1c.
The results indicate P507 had the highest separation factor of 6.5, while the separation
factor of P204 TBP was 3.1 and 1.0. Since P507 had the highest Nd extraction and
selectivity, it was chosen for subsequent tests.
3.2 Effect of P507 Concentration
The extractant concentration plays a significant role in metal ions transportation for the
SLM process. The effect of the phosphorus extractant (P507) concentration on separa-
tion of Nd from Nd–La mixed solution was investigated in the P507-HCl system: 5%,
10%, 20%, and 40% (v/v) of P507/D80. The feed solution was at initial pH of 3,
1000 mg/L La(III) and Nd(III) containing HCl solution. The strip solution was a 3 M
HCl solution. Figure 2a, b shows the extraction % of Nd and La within 6 h. Figure 2c
shows the selectivity of Nd extraction.
From Fig. 2, the extraction of both Nd and La increased with higher P507 concentra-
tions. When the P507 concentration was increased from 5% to 40% (v/v), the extraction
of Nd(III) and La (III) significantly increased from 51.8% to 94.1% and 1.2% to 28.2%,
respectively. The selectivity analysis also found that the extractant concentration sig-
nificantly affected the selective extraction of Nd. The separation factors, α(Nd/La),
shown in Fig. 2c, depict the high dependence on the P507 concentration. The highest
value of α(Nd/La) was found to be 44.25 at 5% (v/v) P507 and drastically reduced to
only 6.39 and 3.23 for 20% and 40% (v/v) of P507, respectively. The phenomena pro-
vide an opportunity for the deep separation of Nd from La.
In order to partially quantify the concentration of P507 loaded on the membrane, a
PXRF scan was conducted to determine the P concentration on the loaded membrane.
This is reasonable since P507 is a phosphorus-based extractant. The results of P wt. %
from scanning the membrane are shown in Table 1. The table shows the P mass percent-
age increased from 0.5% at 5% (v/v) P507 to 5.1% at 40% (v/v) P507 concentration.
The P concentration in the membrane only slightly decreased during the tests which
suggests that negligible loss of the extractant. For example, the 20% P507 test had an
initial P content of 2.81% and decreased to 2.62 wt.% after 6 h of REEs
transportation.
596 L. Li et al.
Fig. 2. Effect of P507 concentration on (a) Nd and (b) La extraction and (c) Nd/La separation
Feed: 1000 mg/L La (III) and Nd(III) in pH = 3 HCl solutions.
Extractant: 5%, 10%, 20%, 40% (v/v) P507; Strip: 3 M HCl solution.
Table 1. Weight percentage of P of the P507 loaded membrane after the REEs transportation.
Membrane 5% P507 10% P507 20% P507 40% P507
P (wt.%) 0.54 1.29 2.62 5.05
3.3 Effect of Auxiliary Agent
In order to tentatively explore the role of auxiliary agents, citric acid was used in the
SLM REEs recovery process. The feed solution was a pH 3, 1000 mg/L La(III) and
Nd(III) chloride solution with 0, 0.2, and 0.5 M of citric acid. It is noticed that the addi-
tion of citric acid works as a buffer solution as well. During the SLM tests, the pH was
almost constant at 3. The strip phase was a 3 M HCl solution. The extractant was 20%
(v/v) P507. The extraction and separation results are shown in Fig. 3.
It was found that the addition of citric acid significantly reduced the transfer rate of
Nd(III). With the increase of the citric acid concentration from 0 to 0.5 M, the Nd recov-
ered was reduced from 88% at 0 M to 34.5% at 0.5 M. Meanwhile, the addition of citric
acid increased the recovery of La. The La recovery rates after 6 h were 13.7%, 29.7%,
and 15.8% for 0, 0.2, and 0.5 M citric acid addition. As for the selectivity, it was found
Fig. 3. Effect of citric acid on (a) Nd and (b) La extraction and (c) separation factor.
Feed: 1000 mg/L La(III) and Nd(III) in pH = 3 HCl solution with 0, 0.2 and 0.5 M citric acid
Extractant: 20% (v/v) P507; Strip: 3 M HCl solution.
that citric acid reduced the selectivity for Nd extraction from Nd/La mix solution. The
separation factor decreased from 5.9 with no citric acid to 2.2 and 2.2 with 0.2 and
0.5 M citric acid, respectively.
4 Conclusions
This work tested Nd/La extraction and separation using a flat-sheet supported liquid
membrane. A microporous polypropylene film was used to support the liquid mem-
brane comprising different commercially available extractants. Multiple parameters
were tested to determine the optimal Nd separation from Nd/La mixed solution, includ-
ing the types of extractant, extractant concentration, and auxiliary agent. The results
suggest P507 is the most efficient extractant for Nd extraction and separation. The sepa-
ration efficiency increased with decreasing P507 concentration. The highest separation
factor of 44.25 was obtained at the low P507 concentration of 5% (v/v). The results will
be used as guidance for the upscale of Nd separation from other REEs using hollow-
fiber supported liquid membrane.
598 L. Li et al.
Acknowledgments The authors wish to thank Dr. Paramee Kumkrong, Judy Kung, and Cindy Jiang
from Energy, Mining, and Environment Research Centre, National Research Council Canada for the
ICP-OES analysis. Financial support of the Office of Energy Research and Development (OERD) of
Natural Resources Canada (NRC-20-07) is greatly appreciated.
References
1. Hou Z, Pei W, Zhang X, Zhang K, Liu Y, Deng J, Jing L, Dai H. Rare earth oxides and their sup-
ported noble metals in application of environmental catalysis. J Rare Earths. 2020;38:819–39.
2. Yahiro H, Eguchi K, Arai H. Electrical properties and reducibilities of ceria-rare earth oxide sys-
tems and their application to solid oxide fuel cell. Solid State Ionics. 1989;36:71–5.
3. Wang L-M, Lin Q, Yue L-J, Liu L, Guo F, Wang F-M. Study of application of rare earth elements
in advanced low alloy steels. J Alloys Compd. 2008;451:534–7.
4. Balaram V. Rare earth elements: a review of applications, occurrence, exploration, analysis, recy-
cling, and environmental impact. Geosci Front. 2019;10:1285–303.
5. Omodara L, Pitkäaho S, Turpeinen E-M, Saavalainen P, Oravisjärvi K, Keiski RL. Recycling and
substitution of light rare earth elements, cerium, lanthanum, neodymium, and praseodymium from
end-of-life applications-a review. J Clean Prod. 2019;236:117573.
6. Dushyantha N, Batapola N, Ilankoon I, Rohitha S, Premasiri R, Abeysinghe B, Ratnayake N,
Dissanayake K. The story of rare earth elements (REEs): occurrences, global distribution, genesis,
geology, mineralogy and global production. Ore Geol Rev. 2020;122:103521.
7. Kim D, Powell L, Delmau LH, Peterson ES, Herchenroeder J, Bhave RR. A supported liquid mem-
brane system for the selective recovery of rare earth elements from neodymium-based permanent
magnets. Sep Sci Technol. 2016;51:1716–26. https://doi.org/10.1080/01496395.2016.1171782.
8. Ni’am AC, Wang YF, Chen SW, Chang GM, You SJ. Simultaneous recovery of rare earth ele-
ments from waste permanent magnets (WPMs) leach liquor by solvent extraction and hollow fiber
supported liquid membrane. Chem Eng Process – Process Intensif. 2020;148:107831. https://doi.
org/10.1016/j.cep.2020.107831.
9. Kim D, Powell LE, Delmau LH, Peterson ES, Herchenroeder J, Bhave RR. Selective extraction of
rare earth elements from permanent magnet scraps with membrane solvent extraction. Environ Sci
Technol. 2015;49:9452–9. https://doi.org/10.1021/acs.est.5b01306.
10. Müller T, Friedrich B. Development of a recycling process for nickel-metal hydride batteries. J
Power Sources. 2006;158:1498–509.
11. Vander Hoogerstraete T, Binnemans K. Highly efficient separation of rare earths from nickel and
cobalt by solvent extraction with the ionic liquid trihexyl (tetradecyl) phosphonium nitrate: a pro-
cess relevant to the recycling of rare earths from permanent magnets and nickel metal hydride bat.
Green Chem. 2014;16:1594–606.
12. Yun XIA, Liansheng X, Jiying T, Zhaoyang L, Li Z. Recovery of rare earths from acid leach solutions
of spent nickel-metal hydride batteries using solvent extraction. J Rare Earths. 2015;33:1348–54.
13. Anitha M, Ambare DN, Kotekar MK, Singh DK, Singh H. Studies on permeation of Nd (III) through
supported liquid membrane using DNPPA + TOPO as carrier. Sep Sci Technol. 2013;48:2196–203.
https://doi.org/10.1080/01496395.2013.793199.
14. Zhang J, Zhao B, Schreiner B. Separation hydrometallurgy of rare earth elements. Springer; 2016.
https://doi.org/10.1007/978-3-319-28235-0.
15. Jiang F, Yin S, Srinivasakannan C, Li S, Peng J. Separation of lanthanum and cerium from chloride
medium in presence of complexing agent along with EHEHPA (P507) in a serpentine microreactor.
Chem Eng J. 2018;334:2208–14. https://doi.org/10.1016/j.cej.2017.11.173.
16. Gras M, Papaiconomou N, Chainet E, Tedjar F, Billard I. Separation of cerium(III) from
lanthanum(III), neodymium(III) and praseodymium(III) by oxidation and liquid-liquid extrac-
tion using ionic liquids. Sep Purif Technol. 2017;178:169–77. https://doi.org/10.1016/j.
seppur.2017.01.035.
17. Habibpour R, Dargahi M, Kashi E, Bagherpour M. Comparative study on Ce (III) and la (III) sol-
vent extraction and separation from a nitric acid medium by D2EHPA and Cyanex272. Metall Res
Technol. 2018;115:207. https://doi.org/10.1051/metal/2017083.
18. Huang XW, Li JN, Zhang YQ, Long ZQ, Wang CM, Xue XX. Synergistic extraction of Nd
and Sm3+ from sulfuric acid medium with D2EHPA-HEHEHP in kerosine, Zhongguo Youse
Jinshu Xuebao/Chinese. J Nonferrous Met. 2008;18:366–71. https://doi.org/10.19476/j.
ysxb.1004.0609.2008.02.030.
19. Chang H, Li M, Liu Z, Hu Y, Zhang F. Study on separation of rare earth elements in complex sys-
tem. J Rare Earths. 2010;28:116–9. https://doi.org/10.1016/S1002-0721(10)60270-0.
20. Chen K, He Y, Srinivasakannan C, Li S, Yin S, Peng J, Guo S, Zhang L. Characterization of the
interaction of rare earth elements with P507 in a microfluidic extraction system using spectro-
scopic analysis. Chem Eng J. 2019;356:453–60. https://doi.org/10.1016/j.cej.2018.09.039.
21. Wu W, Zhang F, Bian X, Xue S, Yin S, Zheng Q. Effect of loaded organic phase containing mix-
tures of silicon and aluminum, single iron on extraction of lanthanum in saponification P507-HCl
system. J Rare Earths. 2013;31:722–6. https://doi.org/10.1016/S1002-0721(12)60348-2.
22. Xie F, Zhang TA, Dreisinger D, Doyle F. A critical review on solvent extraction of rare earths from
aqueous solutions. Miner Eng. 2014;56:10–28. https://doi.org/10.1016/j.mineng.2013.10.021.
23. Banda R, Jeon H, Lee M. Solvent extraction separation of Pr and Nd from chloride solution contain-
ing la using Cyanex 272 and its mixture with other extractants. Sep Purif Technol. 2012;98:481–7.
https://doi.org/10.1016/j.seppur.2012.08.015.
24. Smith RC, Taggart RK, Hower JC, Wiesner MR, Hsu-Kim H. Selective recovery of rare earth
elements from coal Fly ash leachates using liquid membrane processes. Environ Sci Technol.
2019;53:4490–9. https://doi.org/10.1021/acs.est.9b00539.
25. Xu D, Shah Z, Sun G, Peng X, Cui Y. Recovery of rare earths from phosphate ores through sup-
ported liquid membrane using N,N,N′,N′-tetraoctyl diglycol amide. Miner Eng. 2019;139:105861.
https://doi.org/10.1016/j.mineng.2019.105861.
26. Prasad R, Sirkar KK. Hollow fiber solvent extraction: performances and design. J Membr Sci.
1990;50:153–75.
Process Development for Selective Recovery
of Lithium from Black Mass of Spent LiFePO4
Batteries
Tianyu Zhao1 (*), Harshit Mahandra1, Rajashekhar Marthi1, Michael Traversy1,

Yeonuk Choi1, and Ahmad Ghahreman1 (*)
1
Hydrometallurgy and Environment Laboratory, The Robert M. Buchan Department of Mining,
Queen’s University, Kingston, ON, Canada
tianyu.zhao@queensu.ca; ahmad.g@queensu.ca
Abstract. Lithium iron phosphate batteries are extensively used in automobile industries
as a source of electricity in electric/hybrid electric vehicles and are considered as a high-
grade secondary resource for lithium at the end of life. In addition, huge amount of waste
generated leads to environmental problems if left untreated. In this work, a method for
selective recovery of lithium from black mass of spent LiFePO4 batteries has been devel-
oped. Effect of different parameters i.e., pulp density, reaction time, formic acid and H2O2
concentration, and reaction temperature have been studied. Experimental results showed
that from lithium iron phosphate cathode powder under the optimized conditions i.e.,
30 min reaction time, 10% pulp density, 1.0 mol/L HCOOH, 10% H2O2, and 30 oC,
99.98% lithium was selectively leached. However, 96.32% lithium can be leached from
black mass procured from a local industry at 30 oC, 10 mL/g liquid-to-solid ratio, 0.5
mol/L HCOOH, and 1.5% H2O2 concentration and 2 h reaction time (Or 2.5% H2O2
concentration with 40 min reaction time). At the same time, the leaching rates of iron,
copper, aluminum, nickel, cobalt, and manganese were negligible.
Keywords: Lithium · Battery · Leaching · Selectivity · Black mass
1 Introduction
Due to their high safety standards, high energy density, no memory effect, and lower
environmental impact of mining the raw materials, lithium iron phosphate (LFP) batter-
ies have been widely used for electric vehicles and energy storage [1–6]. However, with
the large-scale application of LEP batteries, there has been an increase in the number of
discarded batteries. On one hand, LFP batteries can cause environmental problems if
not disposed of properly. On the other hand, it contains valuable secondary resources,
such as lithium, phosphorous, and carbon.
Given the low recovery values of iron, currently, lithium is recovered from the active
cathode materials of LFP batteries via selective leaching [7]. The core idea of this
method is to oxidize the ferrous ions in lithium iron phosphate to ferric ions. Since the
charge neutrality of the compound (LiFePO4) must be maintained, the monovalent lith-
ium ions will be leached into the solution. As shown in Fig. 1 [8], selective leaching
https://doi.org/10.1007/978-3-031-17425-4_73
602 T. Zhao et al.
Fig. 1. E–pH diagram of the Li–Fe–P–H2O system and the relationship between the electrode poten-
tial of redox couples and pH (25 °C) [8]. (Copyright 2022 The Royal Society of Chemistry).
often requires a weakly acidic environment. Therefore, the active components of the
leaching solution used are generally acids and oxidants. Some commonly used acids
are citric acid [9], oxalic acid [10], and sulfuric acid [11], and the common oxidants
used are hydrogen peroxide [9–11] and air [8].
Formic acid does not contain elements, such as N, P, and S, that can make water
eutrophic, and it is easily biodegradable. Moreover, formic acid is the strongest organic
acid which can help in achieving higher lithium leaching rates. Therefore, formic acid
can be considered as an efficient and an environmentally friendly reagent [12, 13]. In
this work, we chose formic acid as the lixiviant and hydrogen peroxide as the oxidant
to conduct an experimental study on the selective leaching of lithium.
To verify the feasibility of the proposed method, we first conducted selective leaching
experiments on a relatively pure cathode powder LiFePO4/C. In brief, a certain amount
of pure LFP/C powder was added to a solution of formic acid and hydrogen peroxide
and stirred magnetically for a certain amount of time. Various experimental parameters
such as pulp density, reaction time, formic acid and H2O2 concentration, and reaction
temperature were controlled to optimize the leaching conditions. After leaching, the
solids and liquids were separated using filtration, and the leachate was analyzed 4200
MP-AES (Agilent Technologies).
Process Development for Selective Recovery of Lithium from Black Mass of Spent… 603
Table 1. Composition of the pure LFP/C and industrial waste LFP/C black mass.
Element Li (%) Fe (%) Al (%) Cu (%) Ni (%) Co (%) Mn (%) C (%) P (%)
Pure LFP 4.35 32.50 1.42 18.05
Black mass 1.65 10.72 3.12 0.82 0.14 0.17 0.10 37.10 7.08
To test the feasibility of our proposed method on an actual industrial waste sample,
spent LiFePO4 battery black mass powder was procured from a local industry. The
black mass powder was ground and sieved with 28 mesh sieve series (600 μm) prior to
the leaching experiments. Table 1 shows that the graphite in the anode is the main com-
ponent of the black mass powder (~37 wt%), while the valuable lithium is less than
2 wt% of the black mass powder. Unlike the pure LFP/C powder, the industrial-grade
sample contains some additional metals, such as Al and Cu, which come from the con-
ducting foils. Ni, Co, and Mn are present in the black mass as these metals were used
as dopants to improve electrochemical properties of the cathode material. The industrial-
grade LFP powder was subjected to leaching in a mixture of formic acid and H2O2,
similar to the pure LFP/C powder. The leaching conditions were optimized by
controlling various parameters, such as pulp density, reaction time, formic acid and
H2O2 concentrations, and the reaction temperature.
Table 2 shows the leaching rates of pure LFP/C and industrial LFP/C black mass pow-
ders under the optimum. It can be seen that 99.98% of lithium was leached from the
pure LFP/C powder at the optimum conditions, i.e., 30 min reaction time at 10% pulp
density, 1 mol/L HCOOH, 10% H2O2, and 30 °C. Only 0.24% of Fe was leached under
these conditions indicating high selectivity of the lixiviant and the oxidizing agents
toward lithium over iron. The selectivity of leaching can be attributed to the oxidation
of Fe2+ to Fe3+ by H2O2. While the formic acid provides H+ ions for efficient lithium
leaching, H2O2 accepts electrons from Fe2+ present in LiFePO4. The overall reaction can
be represented as [14]
LiFe II PO4 s HCOOH aq ½ H 2 O 2 aq nFe III PO4 s

HCOO aq Li aq H 2 O aq . (1)
In practical applications, it is very cumbersome and difficult to obtain pure cathode

active materials, which undoubtedly increases the production cost and prolongs the
Table 2. The leaching rate of lithium and impurity metals for industrial LFP/C under optimal
conditions.
Element Li (%) Fe (%) Al (%) Cu (%) Ni (%) Co (%) Mn (%)
Pure LFP 99.98 0.24 - - - - -
Black mass 96.32 0.07 0.42 0.93 0.27 0.44 0.38
604 T. Zhao et al.
process. Therefore, in this work, lithium was selectively leached from the industrial
waste LFP/C black mass without separating the anode and cathode powders. Moreover,
this industrial waste contains additional impurities, such as the dopants and Al and Cu
conducting foils. From Table 2 it can be observed that through the combined action of
formic acid and H2O2, the lithium leaching rate can reach more than 96% under the
optimum condition, while the impurity metals are hardly leached (<1%). The optimum
condition for leaching industrial waste LFP/C black mass is 30 °C, 10% pulp density,
0.5 mol/L HCOOH, 1.5% H2O2 concentration, and 2 h reaction time (Or 2.5% H2O2
concentration with 40 min reaction time).
4 Conclusion
To sum up, lithium can be leached selectively and efficiently using synergistic effects
of formic acid and hydrogen peroxide from waste lithium iron phosphate batteries
black mass powder. The composition of pure LFP/C and LFP/C waste acquired from
local industry is significantly different. A high selectivity toward the lithium can be
attributed to the oxidation of Fe2+ to Fe3+ in pure LFP/C system. The proposed method
provides a novel idea for the lithium battery recycling industry, and it has several advan-
tages, such as high lithium leaching rate, good lithium selectivity, wide practicability,
environmentally friendliness, and economy.
Acknowledgments The authors express gratitude to NSERC I2I and Queen’s University for funding
support.
References
1. Georgi-Maschler T, Friedrich B, Weyhe R, et al. Development of a recycling process for Li-ion

batteries. J Power Sources. 2012;207:173–82.
2. Etacheri V, Marom R, Elazari R, et al. Challenges in the development of advanced Li-ion batteries:
a review. Energy Environ Sci. 2011;4(9):3243–62.
3. Tang Y, Zhang B, Xie H, et al. Recovery and regeneration of lithium cobalt oxide from spent
lithium-ion batteries through a low-temperature ammonium sulfate roasting approach. J Power
Sources. 2020;474:228596.
4. Chen Y, Kang Y, Zhao Y, et al. A review of lithium-ion battery safety concerns: the issues, strate-
gies, and testing standards. J Energy Chem. 2021;59:83–99.
5. Zeng X, Li J, Liu L. Solving spent lithium-ion battery problems in China: opportunities and chal-
lenges. Renew Sust Energ Rev. 2015;52:1759–67.
6. Mauger A, Julien CM. Critical review on lithium-ion batteries: are they safe? Sustainable? Ionics.
2017;23(8):1933–47.
7. Beaudet A, Larouche F, Amouzegar K, et al. Key challenges and opportunities for recycling elec-
tric vehicle battery materials. Sustainability. 2020;12(14):5837.
8. Jin H, Zhang J, Wang D, et al. Facile and efficient recovery of lithium from spent LiFePO 4 bat-
teries via air oxidation–water leaching at room temperature. Green Chem. 2022;24(1):152–62.
Process Development for Selective Recovery of Lithium from Black Mass of Spent… 605
9. Kumar J, Shen X, Li B, et al. Selective recovery of Li and FePO4 from spent LiFePO4 cath-
ode scraps by organic acids and the properties of the regenerated LiFePO4. Waste Manag.
2020;113:32–40.
10. Yang Y, Meng X, Cao H, et al. Selective recovery of lithium from spent lithium iron phosphate
batteries: a sustainable process. Green Chem. 2018;20(13):3121–33.
11. Li H, Xing S, Liu Y, et al. Recovery of lithium, iron, and phosphorus from spent LiFePO4 batteries
using stoichiometric sulfuric acid leaching system. ACS Sustain Chem Eng. 2017;5(9):8017–24.
12. Tobiszewski M, Namieśnik J, Pena-Pereira F. Environmental risk-based ranking of solvents
using the combination of a multimedia model and multi-criteria decision analysis. Green Chem.
2017;19(4):1034–42.
13. BASF: Formic acid-general. 2020. https://insights.basf.com/files/pdf/Formic_acid_general.pdf.
14. Mahandra H, Ghahreman A. A sustainable process for selective recovery of lithium as lithium
phosphate from spent LiFePO4 batteries. Resour Conserv Recycl. 2021;175:105883.
Dynamic Simulation of Uranium Ion Exchange
Processes
1
KWA Kenwalt Australia, Canadian Office, Saskatoon, SK, Canada
kevin.heppner@syscad.net
Kevin Heppner1 (*)
Abstract. Nuclear energy is a low carbon form of energy seen broadly as a means to
reduce carbon emissions associated with energy generation. A number of uranium depos-
its around the world are amenable to extraction via in situ recovery. Pregnant liquors
produced from these deposits are commonly processed via ion exchange processes using
commercially available resins.
The utility of process simulation for design and optimization of a uranium ion
exchange plant is demonstrated in this work. A dynamic SysCAD model of an ion
exchange loading, washing, and elution circuit is presented. Ion exchange is a
sequential batch process circuit operating within an overall continuous process. As a
result, significant surge tank capacity is required for operation of connected continu-
ous circuits. This feature of uranium ion exchange plants makes dynamic analysis
important for circuit design, optimization, and debottlenecking. The presented model
uses open literature data as input to calculate circuit capacity and equipment utiliza-
tion as a function of changes in operating conditions.
Dynamic models are also important for understanding impacts of equipment reli-
ability and maintenance scheduling. By calculating equipment utilization, the criti-
cality of various equipment in the process can be assessed. This is useful both for
design, e.g. identifying where backup units are required, and operations, e.g. under-
standing the cost and benefits of maintenance schedules.
Keywords: Uranium · Ion exchange · Dynamic simulation · Reliability · Circuit

capacity · Utilization · Resin
1 Introduction
In situ recovery of uranium is the process whereby uranium leaching is done in situ by
direct introduction of a lixiviant. There are in situ recovery operations in the United
States, Kazakhstan, Uzbekistan, Australia, China, and Russia. Ion exchange is the pre-
ferred technology for in situ recovery due to its lower operating cost compared to sol-
vent extraction for this application [1].
Due to the batch sequential nature of a uranium ion exchange process, dynamic
modelling is needed to estimate plant capacity, identify bottlenecks, understand process
performance, etc. In this work, the chemical process simulator SysCAD is used to per-
https://doi.org/10.1007/978-3-031-17425-4_74
608 K. Heppner
form a dynamic simulation of a number of lead–lag–polish ion exchange columns

being serviced by a single elution circuit. Various operational scenarios are tested to
determine the impact of process changes on production rate and equipment utilization.
2 Chemistry Model
2.1 Loading Model
Uranyl sulphate solutions are known to form anions of uranyl sulphate in a sulphuric
acid solution, i.e.
2 n 1
UO22 nSO 42  UO2 SO 4 n . (1)

Uranyl sulphate anionic complexes can therefore be loaded onto a strong base anion
resin. SysCAD Reactions used for uranyl sulphate loading are presented below:
R 4NH 2SO4 s UO2SO4 aq  R 4NH 2 UO2 SO4 2 s . (2)
Dunn [2] conducted a series of tests to determine equilibrium and maximum loading
characteristics for Dowex® 21K strong base anion resin. It was determined that maxi-
mum loadings varied between 0.76 and 1.12 mM U/g resin (dry basis, chloride form).
Dunn also found that the loading characteristics could be adequately fitted to a Langmuir
isotherm. Equilibrium was modelled using the Langmuir isotherm:
KCU
q qmax . (3)
1 KCU

Parameters for the Langmuir isotherm are taken from [2] for 0.1 molar sulphate
mmol
solution ( qmax = 1.0 , K = 2.0 L/mmol).
g resin
2.2 Elution Model
Elution was performed using a mixture of sodium nitrate and nitric acid. Concentrations
of the eluent were taken from Roberts et al. [3]. Elution of the loaded resin is modelled
using the following reactions:
R 4NH 2SO4 s 2 NaNO3 aq  2 R 4NHNO3 s Na2SO4 aq , (4)
R 4 NH 2 UO 2 SO 4 2 s 2 NaNO3 aq  2 R 4 NHNO3 s Na 2 UO 2 SO 4 2 aq . (5)

Dynamic Simulation of Uranium Ion Exchange Processes 609
The model assumes full replacement of uranyl sulphate and sulphuric acid adsorbed
to the resin. This is a conservative assumption because it increases both nitrate reagent
consumption and elution/conditioning cycle times.
The resin was conditioned using a 5% sulphuric acid solution. Conditioning of the
eluted resin to return the resin to a sulphate form is modelled using the following
reaction:
R 4 NHNO3 s 0.5 H2SO4 aq  0.5 R 4 NH 2SO4 s HNO3 aq . (6)

Both the elution and conditioning reactions were assumed to proceed to
completion.
3 Process Model Description
3.1 Process Overview
The process consists of a number of three column loading trains and a single elution
column. The loading columns are used in a lead–lag–polish configuration. Redirection
of the flow of pregnant solution changes the order of the columns, rather than physical
movement of the resin. Column discharge is also redirected such that lead discharge
goes to lag, lag discharge goes to polish, and polish discharge goes to barren solution.
The elution column is a separate processing circuit which receives resin from the
lead column of the loading trains. Once eluted, washed, and conditioned, the resin is
returned to the loading train. The freshly eluted resin is used in the polish stage for that
loading train. The overall process is illustrated in Fig. 1.
3.2 Ion Exchange Loading Train
The ion exchange loading circuit is presented in Fig. 2. Once the lead column becomes
loaded, the resin is removed for elution. The eluted resin is returned to the same col-
umn, the former lead column, which now becomes the polish column. The former lag
column becomes the lead column, and the former polish column becomes the lag col-
umn. In this way, resin is subjected to three loading cycles of pregnant solution as it
moves from polish to lag to lead position before being eluted.
There are three identical ion exchange loading trains in this model, all being ser-
viced by one elution column.
3.3 Elution Column
The elution column is presented in Fig. 3. Resin is received from the loading train being
serviced. In this model configuration, the elution column is separated from the loading
column trains, and thus, the resin is physically moved. The resin is then eluted with a
610 K. Heppner
Fig. 1. Block flow diagram of overall process.
Fig. 2. Process flow diagram for lead/lag/polish ion exchange loading circuit for uranium.
mixture of sodium nitrate and nitric acid, followed by washing to remove residual elu-
ate. The column is then conditioned with sulphuric acid to ensure that the resin is in the
sulphate form. The sulphate form resin is then returned to the loading train to be reused
for the next loading cycle. The freshly eluted resin is used in the polish column.
Fig. 3. Process flow diagram for elution column for uranium.
3.4 Column Model
Ion exchange columns were modelled using a finite element analysis with the columns
divided into 50 slices along the column length. Equilibrium was assumed within each
slice, with the distribution of uranium in solution and adsorbed to the resin determined
by a Langmuir isotherm. Chemical concentration profiles were produced as a result of
convection of production fluids through the column along the column length and equi-
librium conditions within each column slice.
3.5 Sequencing
Once the lead column is loaded, a loading train must wait until the elution column is not
being used by another loading train. There are no priorities are defined in the sequenc-
ing of loading trains for elution.
3.6 Implementation of Control Logic
3.6.1 Column Controller Class
The column controller is implemented as a reusable PGM (Programmable Module)

class in SysCAD. This class is used by all columns, both loading and elution, through-
out the model. The column class has the following states: Operating, Draining, Washing,
612 K. Heppner
Conditioning, AwaitingEmptying, Emptying, AwaitingFill, Filling, and Idling. The

descriptions of each of these states are presented below.
Operating The column is actively working. For loading columns, it is either receiving
pregnant solution for the loading of a resin or being washed to remove residual barren
solution. For elution columns, it is either receiving eluent, being washed to remove
residual eluate, or being conditioned.
Draining Following operating, the column is allowed to drain to a specified liquid
fraction in the column.
Washing Once drained, process water is used to wash out any residual solution from
the resin.
Conditioning This step only applies for elution. The eluted resin is treated with sulph-
uric acid solution.
AwaitingEmptying The column is awaiting availability of the downstream column to
receive its resin. For example, another column may be being eluted, and thus, the col-
umn must wait until the elution cycle of the other resin is completed.
Emptying The column is discharging resin to the downstream equipment. This is done
until a minimum resin level is reached in the column being emptied.
AwaitingFill The column is empty and is awaiting its resin to be returned. For the elu-
tion column, this is the state it remains for a significant portion of time as the loading of
the columns proceeds.
Filling The column is receiving resin from the upstream unit. Once it is filled, its state
is set to idling.
Idling The column is ready to operate. The managing class for the column will start
the flow to the column, which will trigger the column to enter the operating state.
A flowchart of the column control logic is presented in Fig. 4. The state of the col-
umn is determined by the measuring of inputs and outputs. The inputs and outputs to
the column are set by the controller which oversees the column. This could be either a
loading train controller or an elution system controller. The logic forms a closed loop,
enabling the repeated cycling of a column without any reset logic required.
3.6.2 Loading Train Class

This is a management class for three columns in a lead, lag, and polish configuration. It
manages the movement of resin as well as proper direction of the pregnant solution.
The pregnant solution is always directed to the lead column. Each column receives
three cycles of loading, first as a polish, then as a lag, and finally as a lead column.
The class is used by each of the three ion exchange loading circuits in this model.
The conditions and actions resulting from these conditions are presented in Table 1.
3.6.3 Elution System Class

The elution system class manages the operation of the elution column. The conditions
and actions of the elution system class are presented in Table 2.
Fig. 4. Individual column control logic.
Table 1. Loading train class trigger conditions and actions.

Condition Action
Is columns 1, 2, or 3 awaiting Lock the elution column for this train to ensure no other train
the return of resin? tries to use it
Is columns 1, 2, or 3 operating? Release the elution column if this train currently has a lock on it
Is columns 1, 2, or 3 idling? Start pregnant solution flow to the lead column. Adjust splits on
each column for lead, lag, polish operation
Table 2. Elution system class trigger conditions and actions.

Condition Action
Has a loading train locked me Set the barren resin return splitter to return to the correct loading
for service? train
Is the elution column draining? This state occurs when the column is fully eluted (see Fig. 4). Turn
off flow of eluent to the column
Is the elution column idling? Set the liquid splitter to report to loaded eluate. Start the eluent
flow to the column
Is the elution column Set the liquid splitter to report to spent conditioning
conditioning?
614 K. Heppner
4 Baseline Capacity Analysis
4.1 Input Parameters
Input parameters for loading and elution in the baseline analysis are presented in Tables
3 and 4, respectively.
The elution column volume was slightly larger than the loading columns to ensure
no spillage of resin. The elution column was initially empty.
Table 3. Input parameters for loading.
Parameter Value
Pregnant solution flow (t/h) 200
Pregnant solution U content (mg/L) 650
Maximum resin loading (mol U/kg dry resin, 0.7
chloride form)
Liquid fraction after draining (%) 5
Wash flowrate (t/h) 100
Wash cycle time (h) 3
Resin flowrate (t/h) 100
Number of loading trains 3
Resin charge in column (m3) 10
Loading column volume (m3) 12
Resin bed depth (m) 1.5
Table 4. Input parameters for elution.
Parameter Value
Eluent flowrate (t/h) 20
Liquid fraction after draining (%) 5
Wash flowrate (t/h) 20
Wash cycle time (h) 3
Conditioning flowrate (t/h) 20
NO3 limit after conditioning (%) 0.5
Resin flowrate (t/h) 100
Number of elution columns 1
Elution column volume (m3) 12
4.2 Baseline Simulation Results and Discussion
4.3 Calculations of Averaged Values
In this analysis, as well as the sensitivity analysis, all time-averaged values reported are
arithmetic mean time-averaged values calculated over the duration of the analysis
period. For each calculation, the simulation time was approximately 23 days (2 × 106 s).
Table 5. Baseline capacity assessment.
Parameter Value
Loading Train 1 Utilization (%) 73
Elution System Utilization (%) 73
Production Rate (kg/d) 6000
This simulation time was chosen to ensure that any change in uranium inventory in the
circuits between the start and end of the averaging period was negligible compared to
the throughput of uranium over the averaging period.
4.3.1 Capacity Assessment Summary

A summary of the results of the baseline capacity assessment is provided in Table 5.
In Table 5, utilization is defined as.
WaitingTime
Utilization % 1 100%. (7)
TotalTime
Waiting time includes the following:

• Waiting for resin to be transferred to a column, i.e. when a loading column is await-
ing the return of eluted resin or when an empty elution column awaits resin to elute;
• Waiting for downstream services to allow transfer of resin, i.e. when a loading col-
umn awaits the elution column to be available.
Thus, the utilization gives an indication of the amount of time a piece of equipment
is doing useful work.
4.3.2 Loading Dynamics
The column loading process is shown in Fig. 5. This figure shows the pore liquid ura-
nium concentration in the column as a function of time and depth into the resin bed. As
the column loads, a profile advances with time along the length of the resin bed. The
shapes of these curves are consistent with expected behaviour for the loading column.
In Fig. 5, each profile is separated by 1 h of operating time. For example, after 4 h
(fourth line from left to right), the concentration of uranium in the pore liquid is approx-
imately 300 mg/L 0.2 m below the top surface of the resin bed. It is noted that this
model assumes equilibrium only, and thus neglects mass transfer resistance. Thus, this
model may overpredict the loading capacity of the column.
The transient concentration profile shown in Fig. 5 advances down the resin bed until
the end of the resin bed is reached. Resin breakthrough then occurs, which is the point
in time when uranium is measured in the column discharge. The predicted breakthrough
curve for uranium is shown in Fig. 6. As expected, the increase in uranium residual in
the column discharge remained low for most of the loading cycle and then increased
rapidly once the column was loaded. Axial mass diffusion is also not considered, and
thus, the predicted time required for breakthrough may be slightly longer than actual.
616 K. Heppner
Fig. 5. Uranium Concentration Profile along the resin depth during loading (each profile is separated
by 1 h).
Fig. 6. Uranium concentration in an ion exchange loading column effluent.
4.3.3 Elution/Conditioning Dynamics
Similar to loading, elution is modelled using a finite element method. Eluent is intro-
duced at the top of the column and reacts with the resin until the sodium nitrate eluent
is consumed. Uranium-rich eluate drains from the bottom of the column. Similar to
loading, the model predicts a travelling profile of uranium in the column, see Fig. 7. In
this figure, each profile is separated by 10 min of operating time. Based upon Fig. 7,
elution of the column takes approximately 3 h (i.e. the column is nearly fully eluted
after 180 min, or 18 profile lines).
Fig. 7. Uranium concentration profile across the column during elution (each profile is separated by
10 min).
Fig. 8. Nitrate concentration profile across the column during conditioning (each profile is separated
by 10 min).
Once the uranium is eluted and the resin is washed, the resin is in a nitrate form. The
nitrate is then recovered by conditioning with sulphuric acid to denitrify the resin. This
dynamic process is shown in Fig. 8. Again, in Fig. 8, each profile is separated by 10 min
of operating time. The nitrate concentration in the pore liquid within the resin bed is
indicative of the extent of denitrification of the resin. This is because the sulphate
exchanges with the adsorbed nitrate on the resin to form nitric acid in solution and sul-
618 K. Heppner
Fig. 9. Uranium production profile from the elution column.
phate form resin, see Eq. (6). Comparison of Figs. 7 and 8 shows that the conditioning
process is much faster than the elution process. This is reasonable considering the reac-
tion stoichiometry of elution compared to that of conditioning. Upon comparison of
chemical reactions for elution (Eqs. 4 and 5) and conditioning (Eq. 6), 2 moles of
sodium nitrate is required per mole of uranium and acid eluted, while 0.5 moles of sul-
phuric acid is required per mole of nitrate displaced. Therefore, it is expected that con-
ditioning will occur much more quickly than elution.
4.3.4 Production Dynamics
The uranium production rate from the elution column as a function of time is presented
in Fig. 9. The sequential nature of the process is clearly seen with a square wave pattern
of production. It is noted that the production of uranium only accounts for a portion of
the elution time, due to the need for washing and conditioning of the resin after elution
is completed.
5 Sensitivity Analysis
The model was used to assess the effect of the following process changes on the overall
circuit capacity compared to baseline operating conditions as previously described
(Table 6):
1. Reduction of pregnant solution flow to loading train 1 from 200 to 100 t/h.
2. Reduction of uranium head grade in leach solution from 650 to 325 mg/L.
3. Reduction of eluent flowrate from 20 to 10 t/h with no change in eluent strength.
Table 6. Sensitivity analysis.

Scenario Scenario Scenario Scenario Scenario
Parameter Baseline 1 2 3 4 5
Loading Train 1 Utilization 73 81 83 68 77 81
(%)
(%)
(%)
Elution System Utilization 73 64 51 96 71 58
(%)
Production Rate (kg/d) 6000 5300 3600 5600 5700 6900
4. Increase of washing time for loading and elution columns from 3 to 6 h.

5. Increase of resin loading capacity by doubling of Langmuir constants qmax and K
from 1 and 2 to 2 and 4, respectively.
From these results, it can be seen how the model predicts changes in equipment
utilization and production capacity as a function of operational changes. Each of these
changes are explained below:
• Scenario 1: Decreasing the flow of pregnant solution to the first ion exchange train
results in a decrease in production rate. It also reduces the utilization of the elution
system.
• Scenario 2: Reducing the uranium head grade in the leach solution reduces the pro-
duction rate, but not fully proportionate to change in head grade. This is because the
utilization of the loading trains increases due to reduced utilization of the elution
column. Thus, the time available for loading of the columns is increased, i.e. more
pregnant solution is processed.
• Scenario 3: Reducing the flowrate of eluent to the elution system significantly
increases the elution system utilization due to an increase in elution cycle time.
There is only a small effect on production rate because the elution system in this
model is underutilized. The utilization of the loading columns also decreases due to
increased cycle times in elution.
• Scenario 4: Increasing the washing cycle time for both loading and elution columns
did not significantly affect the circuit. A slight reduction in production rate was
noted.
• Scenario 5: Increasing the resin maximum loading capacity by doubling the
Langmuir constants slightly reduced the elution system utilization. Production rate
increased due to higher resin capacity for loading capacity.
It is notable that the utilization of the first loading train is consistently lower than the
second and third loading train. This is due to the sequential nature of the model. Since
each of the loading trains are identical, they finish their loading sequence at around the
same time given the same feed solution. Thus, a pattern is established where the first
loading column is consistently serviced after the other two loading trains are serviced.
This pattern arises from the order of execution of the different units in the SysCAD
model and is not due to assigned priorities.
620 K. Heppner
6 Conclusions
The following conclusions can be made from this work:

• A dynamic model is a useful tool for capacity and utilization assessments for ion
exchange processes, and for gaining a better understanding of the process.
• Chemical profiles within process equipment can also be predicted. This is useful for
understanding column breakthrough behaviour.
• In all operational scenarios tested, the ion exchange loading train capacity limited
the overall production rate. Thus, changes affecting elution only increased the utili-
zation of elution, rather than affecting production rate.
• Programmable logic can be implemented in classes allowing for reuse of code for
multiple units of the same type.
Acknowledgements The author wishes to acknowledge members of the KWA Kenwalt Australia
team for helpful reviews and feedback during the preparation of this paper. Helpful comments and sug-
gestions from the Conference of Metallurgists reviewer are also greatly appreciated.
References
1. In Situ Leach Mining (ISL) of Uranium – World Nuclear Association (world-nuclear.org).

2. Dunn WE. Uranium anion exchange equilibria on Dowex 21K. ORNL report CF-59-6-132. 1959.
Available at: https://www.osti.gov/servlets/purl/4094885.
3. Roberts JT, Higgins IR. Elution of uranium sorbed on anion exchange resins from low-grade ore
leach liquors. ORNL report 1418. 1952. Available at: https://www.osti.gov/servlets/purl/4350238.
Enhanced Elution Efficiency of Scandium from Ion
Exchange Resins
Mehdi Mostajeran1 (*) and Rory Cameron1
1
Natural Resources Canada, Lands and Minerals Sector, CanmetMINING, Ottawa, ON, Canada
mehdi.mostajeran@nrcan-rncan.gc.ca;
rory.cameron@nrcan-rncan.gc.ca
Abstract. Atmospheric leaching of mineralized wastes, such as coal fly ash, with dilute
H2SO4 is often able to extract a significant fraction of the contained Sc and rare-earth ele-
ment (REE) components, producing low concentrations of the valuable metals and high
concentration of nuisance metals into the pregnant leach solutions (PLSs). Recovery of
parts per million (ppm) levels of valuable metals from a matrix containing several g/L
nuisance metals requires a very selective separation technique, such as ion exchange (IX).
This paper builds on previous work, in which two solvent-impregnated IX resins were
identified with remarkable selectivity for adsorption of Sc from coal fly ash PLS while
absorbing a minimum amount of REEs and the other nuisance elements. While 2 M
H2SO4 was basically unable to recover Sc from these IX resins, addition of some water
miscible organic additives (WMAs) substantially improved the elution efficiency of Sc to
as high as 79%. In continuation of the previous work, elution of Sc from these resins was
investigated in using columns. After nine bed volumes (BVs) of elution and under identi-
cal operating temperature and flow rates, nine times the mass of Sc was recovered with
the blended eluent compared to 2 M H2SO4. Moreover, the maximum concentration of Sc
in the eluted fractions was significantly increased to over 1800 ppm (with the blended
eluent) compared to 167 ppm (with 2 M H2SO4).
Keywords: Coal Fly Ash · Ion Exchange Resin · Scandium Recovery
1 Introduction
Scandium is among the most expensive elements and one of the strategic members of
critical metals. It provides superlative properties to the materials to which it is added,
even at very low concentrations. Addition of <0.5% of Sc to Al produces Al-Sc alloys
which are light weight, corrosion resistant, and more importantly weldable [1, 2]. Using
these alloys in manufacturing commercial aircraft leads to improved fuel consumption
and significant cost savings for airline operators. The Al industry in Canada is vertically
integrated. There is an existing supply chain from bauxite processing to commercial
aircraft production. As a result, having a reliable supply of Sc could offer Canada’s
aerospace industry a significant competitive advantage.
Rare-earth elements production from secondary sources, such as coal fly ash, is a
promising approach because some mine tailings and industrial wastes contain REEs
with considerable economic value. Moreover, this approach requires relatively low
https://doi.org/10.1007/978-3-031-17425-4_75
622 M. Mostajeran and R. Cameron
capital and operation expenditures (capex and opex) as there is no need for physical
preparation (such as drying, crushing, etc.), and the materials are often amenable to
leaching under mild conditions [3]. Moreover, in the case of Sc, production from sec-
ondary sources seems to be the only viable option due to the lack of known primary
ores for commercial production [4]. Additionally, coal fly ash has a potential to be used
as a cement replacement material in civil engineering applications [5]. This consider-
ation is very important and if the leached residue can be used for cement production,
then there is almost no waste remaining.
During acid leaching, the resulting PLS contains valuable metals, such as Sc3+ and
REEs (ppm levels). Additionally, it contains several orders of magnitude higher amounts
(g/L level) of potentially interfering elements including Fe3+, Al3+, and Si4+ [6].
Extraction of Sc and REEs from this complex matrix requires highly selective separa-
tion methods, such as ion exchange. In our previous investigation [6] we employed two
phosphorus containing solvent-impregnated IX resins, TP 272 and VP OC 1026 and
demonstrated efficient Sc recovery from acidic fly ash PLSs. These two resins showed
very high selectivity and capacity for Sc. Mostajeran and Cameron [7] significantly
improved the efficiency of H2SO4 eluents by mixing equal volume of several water
miscible organic additives (WMAs) with 2 M H2SO4 during shake flask experiments.
The reported elution efficiency of Sc for TP 272 and VP OC 1026 was remarkably
improved from 8.1% to 84.5% and 0.02% to 50.5%, respectively.
In this work, columns containing the same two IX resins, TP 272 and VP OC 1026,
were loaded with a synthetic PLS (sPLS) containing 350 ppm of Sc until the column
was saturated with Sc (54 BVs). Elution experiments were conducted with different
eluents and the amount of desorbed Sc in 9 BVs were selected as the comparison objec-
tive to assess and compare the performance of the eluting experiments.
2 Experimental
2.1 Materials
Scandium chloride (ScCl3, anhydrous, 99.9%), 0.5 M H2SO4, and the water miscible
organic additives were supplied by Fisher Scientific and used as received. TP 272 and
VP OC 1026 were provided by LANXESS Corporation [4].
2.2 Loading/Elution Under Column Conditions
To load the resins in columns, resin beads (3 g) were placed in a glass IX column and
conditioned with 1 M H2SO4 overnight. After rinsing with DI water (3 BVs), the load-
ing solution was passed through the column at a specified flow rate and fractions were
collected for analysis. Then the columns were rinsed with 2 BVs of DI water prior to
elution. The elution solutions were passed through the column at a pre-set flow rate and
Enhanced Elution Efficiency of Scandium from Ion Exchange Resins 623
the collected fractions were analyzed for their metallic contents. The loading efficiency
was calculated based on Eq. 1,
C0 Ct
LoadingEfficiency % 100 (1)
C0

where C0 and Ct is the concentrations (ppm) of Sc in the solution before and after load-
ing, respectively (Mostajeran, 2020). The elution efficiency was calculated by Eq. 2,
Celuent Veluent
ElutionEfficiency % 100 (2)
mmetal on resin

where Celuent is the concentration (ppm) of Sc in the eluent, Veluent is the eluent volume
(L) and mmetal on resin is the mass (mg) of loaded Sc on the resin which is calculated from
(C0 − Ct) × Vloading.
2.3 Analytical Method
An Orion Star™ A211 bench-top pH meter was used to measure the pH levels of solu-
tions. A Varian VISTA-Pro inductively coupled plasma with atomic emission spectros-
copy (ICP-AES) was used to analyze the liquid samples to determine the Sc
concentrations. Instrument linear response and quality control checks were performed
before and during each analysis (after every 15 samples) using commercially available
established standard and reference materials.
The feed solution for loading the resins was a sPLS containing 350 ppm of Sc in dilute
H2SO4 at pH = 2.3. Although the Sc concentration in coal fly ash PLS is significantly
lower [6], the higher concentration of Sc was used to fully saturate the columns in a
timely fashion. The resins were loaded with sPLS at a flow rate of 3 BV/h. As shown in
Fig. 1, the adsorption of Sc by both TP 272 and VP OC 1026 was initially rapid and it
plateaued after approximately 30 BVs. The amount of adsorbed Sc in the first 24 BVs
is more than 3 times the amount of Sc adsorbed in the rest of the loading step. Loading
TP 272 led to the adsorption of 25.2 mg Sc which indicates a capacity of 8.40 mg/g.
Meanwhile, 37.7 mg Sc was adsorbed by VP OC 1026 which corresponds to a capacity
of 12.6 mg/g. Higher capacity of the latter resin (VP OC 1026) was also observed dur-
ing shake flask screening experiments [6].
In the first 9 BVs of the eluted fractions from TP 272 with 2 M H2SO4, 11.3% Sc was
measured to be desorbed from the TP 272 column (Fig. 2a). The amount of desorbed Sc
was significantly increased to 78.7% with the blended eluent consisting of WMA-2 and
2 M H2SO4. Furthermore, the maximum concentration of Sc was increased from
167 ppm when using H2SO4 alone to 1846 ppm using the mixture. Less than 0.08% of
624 M. Mostajeran and R. Cameron
350
Sc (ppm) in the loading effluent

300
250
200
150
100 TP 272
VP OC 1026
50
0
0 10 20 30 40 50 60
# Bed Volumes
Fig. 1. Loading TP 272 and VP OC 1026 columns with a sPLS containing Sc at pH = 2.3.
2000 25
A 2M B 2M
1600 20
Sc (ppm) in the eluent
Sc (ppm) in the eluent
WMA-2 in 2 M WMA-1 in 2 M
1200 15
800 10
400 5
0 0
0 2 4 6 8 10 0 2 4 6 8 10
# Bed Volumes # Bed Volumes
Fig. 2. Elution of Sc from (a) TP 272 and (b) VP OC 1026 with eluents containing water miscible
organic additives and 2 M H2SO4.
the adsorbed Sc could be eluted from fully saturated VP OC 1026 with 2 M H2SO4
(Fig. 2b). Increasing the H2SO4 concentration to 6 M resulted in elution of 1.3% of the
adsorbed Sc (data not shown). The maximum concentration of Sc in this eluent was also
increased from 1.0 ppm to 14.9 ppm. When WMA-1 was used in the blended eluent
with 2 M H2SO4, 6.8% of the adsorbed Sc was desorbed from the column with a maxi-
mum concentration of 23.0 ppm.
Enhanced Elution Efficiency of Scandium from Ion Exchange Resins 625
4 Conclusions
Both TP 272 and VP OC 1026 IX resins exhibited high affinity for Sc. These resins
were able to adsorb Sc during column experiments quite rapidly in the first 24 BVs of
loading at room temperature. Solutions of 2 M H2SO4 were able to elute more Sc from
saturated TP 272 than VP OC 1026, 11.3% vs. 0.08%, respectively. Addition of WMA-2
to H2SO4 increased the elution efficiency of Sc from TP 272 seven fold (78.7%).
Similarly, addition of WMA-1 to H2SO4 improved the elution efficiency of Sc from VP
OC 1026 to 6.8%. However, the amount of recovered Sc was not satisfactory. The
future experimental work will focus on isolating the eluted Sc from the solutions and
preparing a saleable Sc product.
Acknowledgments The authors would like to thank Bryan Tisch and Nicolas Reynier from
CanmetMINING for constructive advice and the Analytical Service Group at CanmetMINING for
ICP-AES analyses. They also thank the anonymous reviewer who provided productive comments to
improve the paper. They thank LANXESS Corporation for providing resins and their technical infor-
mation. Mehdi Mostajeran thanks Natural Resources Canada’s Critical Minerals program for support-
ing his research.
References
1. Ahmad Z. The properties and application of scandium-reinforced aluminum. J Miner Metals Mater
Soc. 2003;55:35–9. https://doi.org/10.1007/s11837-003-0224-6.
2. Madhusudhan Reddy G, Mukhopadhyay AK, Sambasiva Rao A. Influence of scandium on weld-
ability of 7010 aluminium alloy. Sci Technol Weld Join. 2005;10:432–41. https://doi.org/10.117
9/174329305X29456.
3. Gaustad G, Williams E, Leader A. Rare earth metals from secondary sources: review of poten-
tial supply from waste and byproducts. Resour Conserv Recycl. 2021;167:105213. https://doi.
org/10.1016/j.resconrec.2020.105213.
4. Mostajeran M, Bondy JM, Cameron R. Turning waste to value: Selective recovery of rare earth
elements from coal fly ash with ion exchange technologies. Proceedings of the 59th conference
of metallurgists, COM 2020: The Canadian Institute of Mining, Metallurgy and Petroleum, 2020.
5. Rafieizonooz M, Mirza J, Salim MR, Hussin MW, Khankhaje E. Investigation of coal bottom ash
and fly ash in concrete as replacement for sand and cement. Constr Build Mater. 2016;116:15–24.
https://doi.org/10.1016/j.conbuildmat.2016.04.080.
6. Mostajeran M, Bondy JM, Cameron R. Mining value from waste: scandium and rare earth ele-
ments selective recovery from coal fly ash leach solutions. Miner Eng. 2021;173:107091. https://
doi.org/10.1016/j.mineng.2021.107091.
7. Mostajeran M, Cameron R. Optimization of an environmentally benign eluent for scandium with
significantly improved elution efficiency. Proceedings of the 60th conference of metallurgists,
COM 2021: The Canadian Institute of Mining, Metallurgy and Petroleum, 2020.
Processing Black Mass into a Mixed Nickel/Cobalt
Hydroxide Precipitate and Industrial-Grade Lithium
Carbonate: A Techno-Economic Analysis Perspective
Christopher P. Baxter1 (*), Andriy Plugatyr1, and Ben Yu1

1
National Research Council of Canada, Ottawa, ON, Canada
Christopher.Baxter@nrc-cnrc.gc.ca
Abstract. The lithium-ion battery (LIB) uptake is accelerating rapidly as part of the
global push toward energy decarbonization, particularly in the automotive sector.
Increased demand is expected to put a significant pressure on the mining industry to sup-
ply critical energy metals (e.g. Li, Ni, Co). In this regard, the rapid development and
deployment of efficient processes for LIB recycling as part of a circular economy is of
great practical importance for alleviating raw materials supply issues as well as reducing
the environmental footprint associated with mining, manufacturing, and end of life.
The development of effective LIB recycling practices presents unique obstacles
such as the complexity of both individual LIB cells and packs, the diversity of cath-
ode chemistries in use and under development, and the lack of standardization
between battery manufacturers. Additionally, effective comparison between pro-
posed processing flowsheets can be complicated by a lack of value quantification and
life cycle impact parameters, especially at early stages of development.
The following work presents a conceptual recycling flowsheet for the processing
of black mass (BM) obtained from a mixed LIB feed stream into a mixed hydroxide
precipitate (MHP) and industrial-grade lithium carbonate. The intent of this exercise
is to introduce an industrially proven and robust “Base Case” hydrometallurgical
processing route which recovers nickel, cobalt, and lithium in forms which are read-
ily integrated into existing material value chains. A Techno-Economic Analysis
(TEA) was performed, consistent with the level of AACE Class 5 capital and operat-
ing cost estimation. Results of this study can provide a basis for benchmarking alter-
native LIB recycling strategies.
The TEA is based on a hydrometallurgical process plant with an annual capacity
of 16,560 dry tonnes of BM, which corresponds to a nominal throughput of 2 dry
tonnes of BM per hour. In this study, BM composition is taken to be 5.1 wt.% Co,
23 wt.% Ni, and 4.0 wt.% Li. The plan is estimated to produce 8440 dry tpa of MHP
product containing 44 wt.% Ni and 9.4 wt.% Co, and 3150 dry tpa of industrial-grade
lithium carbonate. The estimated CAPEX and OPEX are 44 MUSD and 32.9 MUSD
/ year, respectively. The calculated payback and Net Present Value after 20 years
(NPV20) are 1.6 years and 220 MUSD, respectively.
Keywords: Lithium-ion battery · Recycling · Nickel · Cobalt · Lithium

Hydrometallurgy · Techno-Economic Analysis
https://doi.org/10.1007/978-3-031-17425-4_76
628 C. P. Baxter et al.
1 Introduction
Lithium-ion batteries (LIBs) have emerged as a keystone technology in the push to

lower global carbon emissions. This is due in large part to their potential to displace
non-renewable, carbon-intensive energy sources in numerous applications, particularly
in the automotive sector. Over the next decade, demand for LIBs for use in electric
vehicles (EVs), stationary storage, and consumer electronics is projected to increase
substantially [1]. Nevertheless, the net environmental benefits are not completely clear
when accounting for each stage of life [2, 3], for instance, the sourcing of the critical
metals contained in LIBs or the energy-intensive nature of cell manufacturing.
Additionally, if improperly handled or disposed of, LIBs pose several safety and envi-
ronmental hazards, including fire and release of toxic materials [4, 5].
Although remanufacturing and repurposing can help extend the useful life of LIBs
and may be feasible for some applications, ultimately, recycling is the only alternative
to disposal at end of life [6]. It would also displace a portion of the material preparation
needed for new batteries, which is a major contributor to LIB life cycle impacts [2].
Finally, recycling offers an alternative to the local supply of critical metals, including
lithium, cobalt, and nickel.
The recycling of lead–acid batteries from the automotive sector is frequently invoked
as a comparison. Indeed, it shares some of the challenges associated with LIBs, includ-
ing potentially harmful releases of toxic metals and acid if handled incorrectly. Lead–
acid battery recycling is an admirable success story in many jurisdictions, with for
example, a recycling rate of 99% in the United States, higher than tires, paper, alumi-
num, and glass [7]. Some of the key solutions, such as laws, regulations, battery stan-
dardization, and a valuable product (i.e. lead), may serve as inspirations to the case of
LIBs.
Nevertheless, LIB recycling presents a host of unique challenges. The composition
and structure of both individual cells and packs is significantly more varied [8].
Numerous cathode chemistries exist with differing mixes of nickel, manganese, cobalt,
iron, and aluminum [9–11]. Development of novel cathode and anode materials is pro-
ceeding rapidly, introducing a changing feed stream for any recycling operation [1].
With the current absence of standardization or consistent labelling, any potential recy-
cling process will encounter challenges with a variable feed stream and potential con-
tamination. Finally, given the rapid uptake of LIBs over the last few years, new recycling
capacity will need to come online quickly to meet demand for recycling [12, 13].
In this work, a conceptual hydrometallurgical flowsheet for the recovery of cobalt,
nickel, and lithium from a mixed cathode/anode residue, commonly referred to as black
mass (BM) is presented. The motivation for this study is to provide a basis for compari-
son of alternative routes for BM processing and LIB recycling, in general. The pro-
posed flowsheet relies largely on established commercial processes inspired by
similarities to processing streams seen in nickel laterite processing [14, 15], and
otherwise well-understood engineering processes (e.g. chemical precipitation, crystal-
lization). Thus, such an approach to recycling spent LIBs may be both implementable
on relatively short time frames, while simultaneously providing a comparison case for
demonstrating benefits of more novel or specialized processes.
Processing Black Mass into a Mixed Nickel/Cobalt Hydroxide Precipitate… 629
To this end, we present the results of a Techno-Economic Analysis of this Base Case
flowsheet, using capital and operating cost estimates consistent with AACE Class 5
(+50%/−50%).
2 Process Description
The Base Case flowsheet was developed to process the BM cake, consisting of mainly
anode (graphite) and cathode (metal oxides) material, to produce intermediate mixed
nickel/cobalt hydroxide (MHP) and industrial-grade lithium carbonate products.
Although this represents a “downcycling” of the recycled materials [16], these products
were selected due to simplicity of processing, larger tolerance for impurities in the
product, wider available market, and anticipated small-scale nature of the first facilities.
Notably, the proposed flowsheet is also compatible with further refining of MHP and
lithium carbonate, if so desired. Black mass was selected as the feed material for this
study as currently initial battery collection and disassembly is already demonstrated by
several commercial operations, and the cathode presently comprises a substantial por-
tion of the material value of a LIB cell [17, 18].
The process consists of reductive acid leach, primary and secondary slurry neutral-
ization, mixed nickel/cobalt hydroxide precipitation (MHP), scavenger precipitation,
manganese and magnesium removal, calcium carbonate precipitation, lithium carbon-
ate precipitation, and sodium sulphate decahydrate (i.e. Glauber Salt) crystallization
(Fig. 1) process areas. A mass and energy balance was constructed using SysCAD
process simulation software, and incorporate all internal recirculation streams shown in
green (see Fig. 1). Consideration of the complete water balance is crucial for gauging
overall process recoveries, sizing major equipment items, and providing an estimate for
life cycle inventories.
From the LIB manufacturing perspective, R&D efforts are nowadays focused on
developing higher energy density formulations as well as on reducing total cobalt con-
tent. The latter is mainly driven by relatively high cobalt costs due to its limited global
supply [19–21]. As a result, the share of nickel-cobalt aluminum (NCA) and nickel-
manganese-cobalt (NMC) formulations with high Ni/Co ratios, such as NMC (811), is
expected to increase significantly. This was accounted for in this study by considering
a black mass feedstock representative of a cathode chemistry mix expected in 2030 [1],
with an allowance for minor impurities (Table 1).
Global and regional recycling capacity demands projections for spent LIBs contain
inherent uncertainty due to unknowns about future LIBs sales, battery lifetimes, battery
compositions, collection and recycling of spent batteries, and differences in geographi-
cal markets and policies. In this study, a nominal throughput capacity of 2 dry tonnes of
black mass cake per hour was selected as a baseline. This corresponds to a Recycling
Process Plant annual throughput of 16,560 dry tonnes of black mass cake. For context,
according to forecasted battery metal demand growth [1], and assuming an approxi-
mate EV battery life of approximately 7 years, the above capacity corresponds to
roughly 1% of global LIB recycling demand per year expected by 2030, after adjusting
for the proportion of cathode contained in a LIB.
Fig. 1. Block flow diagram of BM recycling process.
Table 1. Elemental black mass feed composition.
Element Concentration in Feed, wt.%

Ni 23
Co 5.1
Mn 2.3
Li 4.0
Al 1.2
Cu 0.3
C 39
3 Results
The facility was modelled as a continuous operation with three 8-hour shifts and an
overall operating factor of 95%. Based on the mass balance, 1.85 t/h of MHP cake is
produced at a solid content of 55 wt.%, recovering 97% of Ni and 94% of Co.
Additionally, 390 kg/h of industrial-grade lithium carbonate crystals is produced,
representing an overall Li recovery rate of 91%. Although a relatively high recovery for
lithium, this is the result of the closed loop for process water in the flowsheet as well as
the high concentration of the metal in the black mass. Further test work will be needed
to verify the impact of impurities and the once-through recovery. Sensitivity to this
assumption is discussed later.
Total project capital cost is estimated at 44.4 MUSD, including direct, indirect, and
contingency costs (Table 2).
Total operating costs are shown in Table 3. Reagent costs were developed based on
the mass balance and include sulphuric acid, lime, sulphur dioxide, magnesia, sodium
hydroxide, and sodium carbonate. Cost of fresh water was based on posted Canadian
municipal water rates for industrial water use. Similarly, power cost was estimated
using the approximate midpoint of Canadian provincial rates. Labor costs were based
on a headcount of 83 personnel, including an allowance for plant management and
administration, floor operators and maintenance personnel, as well as support
services.
In order to account for the value of metals contained in the black mass, the current
scenario assumes that the BM cake produced by LIB recycling operations could, alter-
natively, be sold “as is” to smelters for further processing and refining. Due to high
levels of impurities typically contained in the black mass cake, the Net Smelter Return
factor is expected to be low (ca. 50%) and accounts for Ni and Co values only, as Li
reports to slag during pyrometallurgical treatment and is not recovered.
This study considered further processing of the black mass cake to improve added-
value and produce intermediate Ni/Co and Li products that can be sold to refining
Table 2. Total project capital cost summary.
Description Value, USD

Total Direct Cost [A] $ 26,280,000
Total Indirect Cost [B = 30% × A] $ 7,884,000
Total Cost [A + B] $ 34,164,000
Contingency [C = 30% × (A + B)] $ 10,249,000
Total Project Cost [A + B + C] $ 44,413,000
Table 3. Total project operating cost summary.
Description USD/y
Reagent $ 17,601,000
Shipping and Disposala $ 4,971,000
Labor $ 6,339,000
Otherb $ 978,000
Contingency $ 2,989,000
Total $ 32,878,000
a
Includes a flat rate for packaging and shipping of MHP and lithium carbonate products, and separate
local disposal of neutralization residues and Glauber salt. Shipping of black mass is outside the battery
limits of this study (a trade-off vs. toll smelting)
b
Includes power, maintenance, and supplies
facilities at a higher price. In other words, the present economic analysis considers the
incremental revenue generated from the sale of intermediate products and the associ-
ated incremental CAPEX and OPEX.
London Metal Exchange (LME) Ni and Co and battery-grade lithium carbonate
prices used in this work are shown in Table 4, and were obtained in July 2021. Value of
the black mass was calculated by discounting the contained Ni and Co values by 50%,
with no credit for lithium. Net Smelter Return values of 80% for the intermediate MHP
product were used in the present analysis. The price of industrial-grade lithium carbon-
ate was set at 80% relative to that of battery-grade lithium carbonate.
Net Present Value (NPV) analysis was performed for the Base Case using a discount
rate of 10% and a facility life of 20 years. The resulting NPV of 220 MUSD with
Internal Rate of Return (IRR) of 70% indicates that further processing of black mass
according to the Base Case is economically promising. Results of the economic analy-
sis are shown in Table 5. The NPV sensitivity analysis is shown in Fig. 2.
Revenue presents the greatest sensitivity for NPV, both due to the high price and rich
content of nickel, cobalt, and lithium in the black mass. Consequently, fluctuations in
metal price over project life may greatly impact process attractiveness. However, it
should be noted that this study is based on a cathode chemistry mix projection for 2030
(i.e. strong bias toward NCA and NMC811 chemistries) which is already depleted in
cobalt relative to the present market mix. Thus, this economic analysis can be consid-
ered a conservative economic scenario with respect to cobalt dependency.
Due to the attribution of value in feed and product streams, lithium is the dominant
driver of process incremental revenue, accounting for 50% of the total (Fig. 3). This
remains relatively stable for slight variations in cathode feed composition, whereas
nickel and cobalt revenue is dependent on the proportion of each metal in the black
mass.
Table 4. Economic Credit for Ni, Co, and Li.

Parameter Units Ni Co Li2CO3
Reference Price USD/kg $ 18.35 $ 50.50 $ 13.35
Black Mass Value Factor % 50 50 0
Black Mass Value USD/kg $ 9.18 $ 25.25 $–
Base Case Product Value Factor % 80 80 80
Base Case Product Value USD/kg $ 14.68 $ 40.40 $ 10.68
Table 5. Base case economic parameters.
Description Units Value

Total Project Cost USD $ 44,413,000
OPEX USD per annum $ 32,878,000
Incremental Revenue USD per annum $ 63,935,000
Net Present Value 20 years $ 219,989,000
Payback Period Years <2.0
Internal Rate of Return % 70
400
350
300
NPV (MUSD)
250
200
150
100
50
0
-30 -20 -10 0 10 20 30
+/- (%)
CAPEX OPEX Revenue Discount Rate
Fig. 2. Base case net present value sensitivity analysis.
Fig. 3. Comparison of value of recycling flowsheet feed and product streams.
The largest individual operating expense is labor, at nearly one-fifth of the total. This
is a consequence of the use of North American labor rates in the estimate as well as the
relatively small throughput. Collectively, the primary reagents sulphuric acid, sodium
carbonate, lime, and sulphur dioxide account for a further 45% of OPEX, and so effort
to reduce acid and base consumption would also improve process economics.
Overall, the flowsheet NPV is least sensitive to CAPEX. Costs are distributed rea-
sonably evenly across all process areas with the exception of Manganese and Magnesium
Precipitation (19.7% of direct cost) and Glauber Salt Crystallization (14.8% of direct
cost). The former is resultant from the precipitation of MHP with magnesia, selected for
its slow dissolution rate. This may enable improved control of the supersaturation
regime yielding product with desirable settling and filtration characteristics. Moreover,
alternative basic reagents, such as caustic, would necessitate their own removal from
the process. The latter area is required to control the accumulation of sodium sulphate
in the process water. It is notable that the two most expensive process areas for this
flowsheet are related not to primary leaching of metal recovery, but rather to impurity
removal. Thus, they may be overlooked when comparing alternative lithium ion battery
recycling flowsheets without careful consideration of the mass balance.
4 Conclusion
The rapid rise in LIB uptake demands the proactive development of effective end-of-
life processing for environmental, safety, and critical metal supply considerations.
However, reprocessing of spent batteries present unique challenges, such as the com-
plexity of battery cells and packs, as well as the variable compositions of any feed
material to recycling. Given the imminent influx of spent batteries in significant quanti-
ties, it is crucial that recycling strategies be deployed in a timely manner. In this work,
a “Base Case” approach was presented based on relatively straightforward processing
steps, including preliminary Techno-Economic estimates. It is intended that this study
will provide a useful baseline for the grading of other candidate spent LIB processing
research.
Acknowledgements This work was undertaken under the Advanced Clean Energy (ACE) Program of
the National Research Council of Canada, with funding from the Office of Energy Research and
Development (OERD).
References
1. BNEF. 2H 2019 battery metals outlook demand realities. Bloomberg New Energy Finance; 2019.
2. Sun X, Luo X, Zhang Z, Meng F, Yang J. Life cycle assessment of lithium nickel cobalt manganese
oxide (NCM) batteries for electric passenger vehicles. J Clean Prod. 2020;273:123003. https://doi.
org/10.1016/j.jclepro.2020.123006.
3. Ellingsen LA-W, Hung CR, Strømman AH. Identifying key assumptions and differences in life
cycle assessment studies of lithium-ion traction batteries with focus on greenhouse gas emissions.
Transp Res Part D: Transp Environ. 2017;55:82–90. https://doi.org/10.1016/j.trd.2017.06.028.
4. Lv W, Wang Z, Cao H, Sun Y, Zhang Y, Sun Z. A critical review and analysis on the recycling of
spent lithium-ion batteries. ACS Sustain Chem Eng. 2018;6(2):1504–21. https://doi.org/10.1021/
acssuschemeng.7b03811.
5. Zeng X, Li J, Liu L. Solving spent lithium-ion battery problems in China: opportunities and chal-
lenges. Renew Sustain Energy Rev. 2015;52:1759–67. https://doi.org/10.1016/j.rser.2015.08.014.
6. Chen M, Ma X, Chen B, Arsenault R, Karlson P, Simon N, Wang Y. Recycling end-of-life
electric vehicle lithium-ion batteries. Joule. 2019;3(11):2622–46. https://doi.org/10.1016/j.
joule.2019.09.014.
7. Gaines L. The future of automotive lithium-ion battery recycling: charting a sustainable course.
Sustain Mater Technol. 2014;1-2:2–7. https://doi.org/10.1016/j.susmat.2014.10.001.
8. Brückner L, Frank J, Elwert T. Industrial recycling of lithium-ion batteries—a critical review of
metallurgical process routes. Metals. 2020;10(8):1107. https://doi.org/10.3390/met10081107.
9. Nitta N, Wu F, Lee JT, Yushin G. Li-ion battery materials: present and future. Mater Today.
2015;18(5):252–64. https://doi.org/10.1016/j.mattod.2014.10.040.
10. Wang Y, An N, Wen L, Wang L, Jiang X, Hou F, Yin Y, Liang J. Recent progress on the recy-
cling technology of Li-ion batteries. J Energy Chem. 2021;55:391–419. https://doi.org/10.1016/j.
jechem.2020.05.008.
11. Schmuch R, Wagner R, Hörpel G, Placke T, Winter M. Performance and cost of materials for
lithium-based rechargeable automotive batteries. Nat Energy. 2018;3:267–78. https://doi.
org/10.1038/s41560-018-0107-2.
12. Frith J, Li D, Lu S. Lithium-ion battery recycling: 2 million tons by 2030. Bloomberg New Energy
Finance, 2019.
13. Larouche F, Tedjar F, Amouzegar K, Houlachi G, Bouchard P, Demopoulos GP, Zaghib K. Progress
and status of hydrometallurgical and direct recycling of Li-ion batteries and beyond. Materials.
2020;13(3):801. https://doi.org/10.3390/ma13030801.
14. Roy T. Preparing nickel and cobalt concentrates. Ind Eng Chem. 1961;53(7):559–66.
15. McDonald RG, Whittington BI. Atmospheric acid leaching of nickel laterites review: part
I. sulphuric acid technologies. Hydrometallurgy. 2008;91(1):35–55. https://doi.org/10.1016/j.
hydromet.2007.11.009.
16. Beaudet A, Larouche F, Amouzegar K, Bouchard P, Zaghib K. Key challenges and opportuni-
ties for recycling electric vehicle battery materials. Sustainability. 2020;12(14):5837. https://doi.
org/10.3390/su12145837.
17. Attwood J. Bottom up cost scenarios for lithium-ion batteries. Bloomberg New Energy
Finance, 2017.
18. Wang X, Gaustad G, Babbitt CW, Bailey C, Ganter MJ, Landi BJ. Economic and environmental
characterization of an evolving Li-ion battery waste stream. J Environ Manag. 2014;135:126–34.
https://doi.org/10.1016/j.jenvman.2014.01.021.
19. BNEF. Supply and demand. Bloomberg New Energy Finance, 2017.
20. USGS. Mineral commodity summaries 2019. U.S. Department of the Interior; U.S. Geological
Survey, 2019.
21. Harper G, Sommerville R, Kendrick E, Driscoll L, Slater P, Stolkin R, Walton A, Christensen P,
Heidrich O, Lambert S, Abbott A, Ryder K, Gaines L, Anderson P. Recycling lithium-ion batter-
ies from electric vehicles. Nature. 2019;575:75–86. https://doi.org/10.1038/s41586-019-1682-5.
Detailing of Mineral Processing Operations Within
Strategic Mine Planning Algorithm: A Quantitative
Cornerstone for Enabling Process Innovation
Aldo Quelopana1, 2, Javier Ordenes2, and Alessandro Navarra2 (*)

1
Universidad Católica de Norte, Antofagasta, Chile
2
McGill University, Montreal, QC, Canada
alessandro.navarra@mcgill.ca
Abstract. Critical materials are essential in a spectrum of products in diverse eco-

nomic sectors. Currently, there is an increasing demand for energy storage materials
(Li, Ni, Co, Mn, Vn, graphite, etc.) that suffer from supply risk. Consequently, inno-
vative technologies have emerged to enable the mining industry to satisfy the forecast
demand, ideally leading toward processes that are equitable and environmentally
sound. Nevertheless, quantitative methods are needed to determine whether promis-
ing technologies in the lab could translate into solutions at a specific mine site, taking
into account the geological uncertainty of how much of the material is in the orebody,
and under what form, and how the technologies and processes would be operated dif-
ferently in confronting this uncertainty. Any new process that is enabled by a technol-
ogy must be characterized in terms of system-wide operational modes; this establishes
how the technology is coordinated with the adjacent processes. Particularly for open-
pit mines, there are well-established strategic planning algorithms that determine the
economic viability of different processing options, under varying degrees of geologi-
cal uncertainty, over the mine life. However, these algorithms have tuning parameters
that obscure the representation of new technologies. Our work has addressed this
issue. Sample calculations are being developed based on an open-pit manganese
mine. Adaptation of the approach to underground mines is also discussed.
Keywords: Critical materials · Geostatistical simulation · Strategic mine planning

algorithm · Manganese
1 Introduction
Critical materials are those whose supply chain presents risks and potentially signifi-
cant consequences. This does not necessarily mean that there is a current crisis; how-
ever, it is a warning based on plausible scenarios [1]. Numerous reports have defined
the criticality of materials and listed them, creating an awareness of the increasing
demand for energy storage materials and their consequent supply risk. Among these
materials, we may find lithium (Li), nickel (Ni), cobalt (Co), manganese (Mn), vana-
dium (V), and graphite [2–6].
https://doi.org/10.1007/978-3-031-17425-4_77
638 A. Quelopana et al.
A key difference that distinguishes mining systems from other value-adding indus-
trial systems is the geological uncertainty of the orebody. This uncertainty is one of the
major causes of costs overruns [7, 8], which has been addressed by mine production
scheduling research [9, 10]. Within this context, the quantitative analysis to determine
a technology’s suitability requires the consideration of operational modes that define
how the overall system may “accommodate major and sustained changes in feed” [11].
It is through the formulation of operational modes that technology can be represented
within strategic mine plans. A conventional set of technologies, having available opera-
tional modes, is compared to an alternate set of technologies, having a different set of
operational modes. This is done with respect to the life of mine to determine if the
alternate technology is a justifiable risk.
2 Advances in Stochastic Open-Pit Strategic Mine Planning
For over 20 years, open-pit mine planning algorithms have incorporated geological
uncertainty [10]. These algorithms apply geostatistical techniques, such as sequential
Gaussian simulation to geological sampling data (drill cores) to generate a series of
equiprobable representations of an orebody; the resulting mine plan (Fig. 1) is then
expected to perform well for the entire distribution of possible scenarios [11]. Studies
have shown that this incorporation of geological uncertainty can increase the expected
net present value of mining projects by as much as 25%, which can correspond to hun-
dreds of millions of dollars [10]. This incorporation of equiprobable geological sce-
narios into the optimization is what distinguishes stochastic mine planning algorithms
from the deterministic algorithms, the latter of which are optimized based on a single
geological scenario (i.e., the “average” or “interpolated” scenario, [10]). Moreover, sev-
eral different extensions of the stochastic approach have incorporated degrees of detail
into the strategic mining planning algorithms, such as stockpiling policies [13], supply
uncertainty [14], and commodity price uncertainty [15].
However, most of the implementations of high-resolution stochastic open-pit mine
planning algorithms (i.e., in which tens or hundreds of thousands of blocks must be
Fig. 1. A distribution of possible orebody scenarios is represented by a countable number of equiprob-

able geological scenarios (left) for which the strategic mine plan (right) is optimized. The coloring in
the geological scenarios represent geological attributes, e.g., grade; the coloring on the mine plan
represents the pit over different periods. (Adapted from Navarra and Waters [12]).
Detailing of Mineral Processing Operations Within Strategic Mine Planning Algorithm… 639
sequenced) rely on metaheuristic formulations to obtain an approximately optimal solu-

tion, many of which require the specification of tuning parameters; such metaheuristic
approaches include Simulated Annealing [16], Tabu Search [17], and Ant Colony
Optimization [18]. It is generally unclear which adjustment of tuning parameters will
reliably lead to a convergent mine plan, and under which circumstances. In comparing
different technological choices at a mine, there is thus the danger of evaluating one design
with tuning parameters that are unintentionally favorable, and then evaluating an alternate
design with tuning parameters that are unintentionally disfavorable; stated differently,
these approaches do not guarantee a fair comparison between mining technologies.
The main exception is the Variable Neighborhood Descent (VND) algorithm originally
developed by Lamghari et al. [19, 20], which is deliberately formulated to avoid ambigu-
ous tuning parameters; this algorithm structure is also commonly associated with the
Variable Neighborhood Search (VNS), although VNS generally includes some random
alterations of incumbent solutions while VND is entirely deterministic. Because of the
avoidance of tuning parameters, Lamghari’s algorithm was selected as a starting point for
subsequent work by Navarra et al., which has led to a series of publications that detail
mineral processes within strategic open-pit mine planning algorithms [12, 21–24].
Concurrently to the metaheuristic-oriented work [10–24], specialized linear pro-
gramming techniques have been developed which also avoid the reliance on tuning
parameters, most notably the Bienstock-Zuckerberg (BZ) algorithm described for
example in Ref. [25]. In general, it is necessary to avoid the presence of ambiguous
computational parameters which do not represent actual aspects of the mine plan, and
can obscure the comparison of one mine plan versus another.
The current work describes the development of a quantitative method to analyze the
impact of new technology in open-pit mine planning. This approach includes short-
term decisions to alternate between available operational modes. Moreover, it avoids
the need for extraneous tuning parameters, since it is an extension of the VND formula-
tion originally developed by Lamghari [19, 20]. Sample calculations are now being
developed that are based on a manganese mine. This development has led to the formu-
lation of a mass-balancing linear program that is solved as a subproblem within the
overarching structure of Lamghari’s VND algorithm. The linear program determines
the portions of ore that are to be processed under the alternate operational modes; this
representation of operational modes supports detailed descriptions of technology and is
therefore a vital link between modern strategic mine planning and techno-economic
evaluations. Moreover, this linear program has a particular structure that is subject to
the Dantzig-Wolfe (DW) decomposition, [25–27] which is an enhancement over the
standard Simplex method.
3 Development of Sample Calculations for a Manganese

Open-Pit Mine
Manganese ore deposits are formed from different ore genesis processes (diagenesis,
metamorphism, hydrothermal alteration, and supergene alteration) and exhibit diverse
mineralogy, textures, and physical properties [28]. The literature presents various
classifications of manganese deposits. These widespread geological contexts can gener-

ate different geometallurgical challenges for Mn ore beneficiation, particularly as a
mine enters into later production phases [29]; the initial metallurgical plant may well
have been designed for its intended feed and would have functioned successfully for
many years. But as the shallower ore is depleted, there is an increasing portion of dif-
ferent classes of ore, for which the initial plant was not optimally designed.
In particular for manganese mineral processing, flotation circuits are configured to
treat: (1) ores with a high calcite gangue, (2) ores with mixed calcite and siliceous
gangue, and (3) ores with a highly siliceous gangue [30]. In this context, manganese
deposits may consist of two or more host rocks with different carbonate concentrations
(e.g., the Kalahari manganese field, where the host rock vary from a thick strata of
dolomites and dolomitic limestones, to possibly stromatolite limestones interstratified
with carbonatized clays and siliceous shales, or possibly lavas of basic and medium
composition, [29]), and this has represented challenges for mineral processing.
Yet these aging mines that are a source of critical materials present an opportunity
for innovative process upgrades; the choice can be either that the mine pilots an innova-
tive/customized approach that responds to the new ore classes that are dominant at
deeper levels, or that the mine may suffer premature closure. The work of Navarra et al.
[21] from 2017 titled Strategic evaluation of concentrator operational modes under
geological uncertainty demonstrated the design (or indeed the redesign) of the mineral
concentrator and the strategic mine plan are coupled, and should therefore be resolved
simultaneously within the same quantitative framework. Figure 2 shows an adaptation
of the results that are generated from the 2017 framework; the lower-left portion of the
ore is not well handled without an additional operational mode, i.e., without a technol-
ogy upgrade. Figure 2a depicts the mine plan with the un-upgraded concentrator, in
which the difficult ore is postponed, leading to a smaller overall pit and a shorter mine
life; Fig. 2b shows the upgraded concentrator that is more versatile leading to a more
balanced pit development, and ultimately a longer life of mine by several years. As
Years 1 & 2
Years 3 & 4
Years 5 & 6
Years 7 & 8
Years 9 & 10
(a)
Years 1 & 2
Years 3 & 4
Years 5 & 6
Years 7 & 8
Years 9 & 10
(b)
Fig. 2. Schematic view of multi-year mine plans confronting difficult ore in the lower-left region of
the orebody that is indicated with a hatched pattern, for (a) un-upgraded mineral concentrator which
performs comparatively poorly in the difficult ore, (b) upgraded mineral concentrator that is equipped
to process the difficult ore. (See Navarra et al. [21]).
described in Ref. [21], this also corresponds also to an increase in roughly $50 million
in the NPV of the mine.
The main issue with the 2017 framework [21] that is now being addressed in the
context of manganese open-pit mining is the detailed representation of operational
modes. The 2017 framework, as well as the subsequent 2018 and 2021 efforts [23, 24],
does not represent the transitionary periods in the mine life, in which balances of ore
classes (e.g., with varying degrees of calcite/siliceous gangue) are supported through
the strategic alternation of modes. The embedding of a mass-balancing linear program
is thus an advancement over the previous work, and can support quantitative aspects
that describe the mineral processing technology whose representation in Fig. 2 is cur-
sory, e.g., the results of pilot plant studies. The successful functioning of the mass-
balancing linear program depends on its particular block angular structure that is subject
to the Dantzig-Wolfe (DW) decomposition [25–27].
4 Reconciliation of the VND-DW Approach with Underground

Mine Planning
The main attractive feature of Lamghari’s VND algorithm is that it avoids tuning
parameters that affect the results of mine plan optimization [19, 20]. However, the orig-
inal VND implementation still was not adequate for enabling innovative mineral pro-
cesses since it contained the so-called “penalty” factors; these factors permit the mine
plan to violate operational constraints with a certain probability (i.e., in a certain por-
tion of the geological scenarios). Such violable constraints are known as “soft con-
straints” and have a tenuous connection with reality; if a long-term (strategic) decision
is made and later found to be untenable due to operational complications, the strategic
objectives may still be attainable albeit through the application of costly contingencies,
hence penalties. Yet, the exact value of these penalties is difficult to ascertain within the
formulation of Lamghari et al. [19, 20] as well as the preceding works of Dimitrikopoulos
et al. [10, 13–15, 17]; these penalty values are effectively like tuning factors, and like-
wise obscure fair comparisons between conventional and innovative processes. Thus,
the adaptation of Lamghari’s VND algorithm for enabling innovative processes has
required that these penalty factors (in addition to the aforementioned tuning parame-
ters) be eliminated from the formulation.
The VND frameworks of Navarra et al. [12, 21–24] were influenced indeed by oper-
ational dynamics of inventory systems [31] in formulating alternating operational
modes rather than relying on artificial penalties. Moreover, the notions of alternating
modes are applicable to underground mining equally as well as open pit. Recent work
by Órdenes et al. [32] considers the operational dynamics of an underground gold mine,
within narrow vein-type deposits, whose imbalances in ore types cause spikes in cya-
nide consumption; these spikes in cyanide consumption are indeed an operational “pen-
alty” as the leaching plant applies a contingency mode.
Following this work by Órdenes et al. [32] to simulate the short-term operational
timeframe, a VND approach to strategic underground mining is now conceivable (at
least for the vein-type deposits that were the focus of Ref. [32]). Although these types
of deposits are more geomechanically restricted than open-pit mines, the mine planning
can nonetheless be formulated through a set of rules and precedence constraints; it is
likely that the main underground mining methods can each be formulated through simi-
lar rules and precedence constraints, which will be the subject of future research.
Interestingly, the mass-balancing linear program that is being developed for the current
work is directly applicable to vein-type gold deposits described by Órdenes et al. [32].
Thus, the VND-DW approach has implications for to the techno-economic evaluation
of technology in the underground context, as well as the open pit.
5 Collaboration to Enable Process Innovation
The authors of this work are committed to collaborating with mines and metal produc-
ers, and related suppliers, consulting firms, and information systems specialists. There
is particular interest to bring interdisciplinary perspectives into mineral processing and
extractive metallurgical plants, to elicit new solutions to longstanding problems.
We have found that existing plants are particularly susceptible to technological stag-
nation, as even the most incremental change carries some amount of risk. Computational
experiments are essential for exploring and quantifying new operational practices, as
well as new technology. Often the technological concepts exist, but a lack of quantita-
tive methods prevents the innovation initiatives from reaching their potential. The
approaches described in this paper provide a direct link to life-of-mine NPV calcula-
tions, to enable high impact initiatives.
References
1. King A. Critical materials. Elsevier; 2021. https://doi.org/10.1016/B978-0-12-818789-0.00003-7.

2. U.S. Department of Energy. Critical materials strategy. 2011. https://www.energy.gov/node/349057.
Last accessed 20 Dec 2021.
3. British Geological Survey. Risk List 2015. 2015. https://www2.bgs.ac.uk/mineralsuk/download/
statistics/risk_list_2015.pdf. Last accessed 20 Dec 2021.
4. Deloitte Sustainability, British Geological Survey, Buerau de Recherches Géologique ey Minières,
and Netherlands Organisation for Applied Scientific Research. Study on the review of the list of
critical raw materials. European Commission. 2017. https://op.europa.eu/en/publication-detail/-/
publication/08fdab5f-9766-11e7-b92d-01aa75ed71a1/language-en. Last accessed 21 Dec 2021.
5. Fortier S, Nassar N, Lederer G, Brainard J, Gambogi J, Mccullough E. Draft critical mineral list –
Summary of methodology and background information – U.S. Geological Survey technical input
document in response to Secretarial Order (N° 3359). 2018. https://pubs.er.usgs.gov/publication/
ofr20181021. Last accessed 21 Dec 2021.
6. National Science and Technology Council. Assessment of critical minerals: updated application of
screening methodology. 2018. https://trumpwhitehouse.archives.gov/wp-content/uploads/2018/02/
Assessment-of-Critical-Minerals-Update-2018.pdf. Last accessed 21 Dec 2021.
7. Nature Sustainability. From the lab to real-world use. Nat Sustain. 2019;2:989. https://doi.
org/10.1038/s41893-019-0435-7.
8. Baum W. Ore characterization, process mineralogy and lab automation a roadmap for future min-
ing. Miner Eng. 2014;60:69–73. https://doi.org/10.1016/j.mineng.2013.11.008.
9. Newman A, Rubio E, Caro R, Weintraub A, Eureke K. A review of operations research in mine
planning. Interface. 2010;40(3):222–45. https://doi.org/10.1287/inte.1090.0492.
10. Dimitrakopoulos R. Stochastic optimization for strategic mine planning: a decade of develop-
ments. J Min Sci. 2011;47:138–50. https://doi.org/10.1134/S1062739147020018.
11. Navarra A, Rafiei A, Waters K. A sysems approach to mineral processing based on mathematical
programming. Can Metall Q. 2017;56:35–44. https://doi.org/10.1080/00084433.2016.1261501.
12. Navarra A, Waters K. Concentrator utilisation under geological uncertainty. Can Metall
Q. 2016;55(4):470–8. https://doi.org/10.1080/00084433.2016.1237062.
13. Lamghari A, Dimitrakopoulos R. Network-flow based algorithms for scheduling production in
multi-processor open-pit mines accounting for metal uncertainty. Eur J Oper Res. 2016;250:273–90.
https://doi.org/10.1016/j.ejor.2015.08.051.
14. Senécal R, Dimitrakopoulos R. Long-term mine production scheduling with multiple processing
destinations under mineral supply uncertainty, based on multi-neigbourhood Tabu Search. Int J
Min Reclam Environ. 2020;34(7):459–75. https://doi.org/10.1080/17480930.2019.1595902.
15. Rimélé M, Dimitrakopoulos R, Gamache M. A stochastic optimization method with in-pit waste
and tailings disposal for open pit life-of-mine production planning. Resour Policy. 2018;57:112–21.
https://doi.org/10.1016/j.resourpol.2018.02.006.
16. Sari Y, Kumral M. An improved meta-heuristic approach to extraction sequencing and block rout-
ing. J South Afr Inst Min Metall. 2016;116:673–80. https://doi.org/10.17159/2411-9717/2016/
v116n7a9.
17. Lamghari A, Dimitrakopoulos R. Hyper-heuristics approaches for strategic mine planning under
uncertainty. Comput Oper Res. 2020;115. https://doi.org/10.1016/j.cor.2018.11.010.
18. Sattarvand J, Niemann-Delius C. A new metaheuristic algorithm for long-term open-pit production
planning. Arch Min Sci. 2013;58:107–18. https://doi.org/10.2478/amsc-2013-0007.
19. Lamghari A, Dimitrakopoulos R, Ferland J. A variable neighbourhood descent algorithm
for the open-pit mine production scheduling problem with metal uncertainty. J Oper Res Soc.
2014;65:1305–14. https://doi.org/10.1057/jors.2013.81.
20. Lamghari A, Dimitrakopoulos R, Ferland J. A hybrid method based on linear programming
and variable neighborhood descent for scheduling production in open-pit mines. J Glob Optim.
2015;63:555–82. https://doi.org/10.1007/s10898-014-0185-z.
21. Navarra A, Menzies A, Jordens A, Waters K. Strategic evaluation of concentrator operational modes
under geological uncertainty. Int J Miner Process. 2017;164:45–55. https://doi.org/10.1016/j.
minpro.2017.05.009.
22. Navarra A, Montiel L, Dimitrakopoulos R. Stochastic strategic planning of open-pit mines with ore
selectivity recourse. Int J Min Reclam Environ. 2018;32:1–17. https://doi.org/10.1080/1748093
0.2016.1201380.
23. Navarra A, Grammatikopoulos T, Waters K. Incorporation of geometallurgical modelling
into long-term production planning. Miner Eng. 2018;120:118–26. https://doi.org/10.1016/j.
mineng.2018.02.010.
24. Quelopana A, Navarra A. Integration of strategic open-pit mine planning into hierarchi-
cal artificial intelligence. J South Afr Inst Min Metall. 2021;121(12):643–52. https://doi.
org/10.17159/2411-9717/1367/2021.
25. Muñoz G, Espinoza D, Goycoolea M, Moreno E, Queyranne M, Letelier O. A study of the
Bienstock-Zuckerberg algorithm: applications in mining and resource constrained project schedul-
ing. Comput Optim Appl. 2018;69:501–35. https://doi.org/10.1007/s10589-017-9946-1.
26. Winston W, Venkataramanan M. Introduction to mathematical programming. 4th ed. Pacific Grove:
Thompson; 2003.
27. Hillier F, Lieberman G. Introduction to operations research. 10th ed. New York: McGraw-
Hill; 2015.
28. Peterson M, Manuel J, Sarath H. Geometallurgical characterization of Mn ores. Appl Earth Sci.
2020;130:1–21. https://doi.org/10.1080/25726838.2020.1863013.
29. Kuleshov V. Genetic types, classifications and models of manganese ore formation. In: Maynard
J, editor. The origin and formation of manganese rocks and ores; 2017. p. 63–352. https://doi.
org/10.1016/B978-0-12-803165-0.00003-3.
30. Michaud LD. Manganese ore processing. Miner Process Metall. 2016. https://www.911metallurgist.
com/blog/manganese-ore-processing#manganese-ore-concentration-by-flotation. Last accessed
21 Dec 2021.
31. Navarra A, Alvarez M, Rojas K, Menzies A, Pax R, Waters K. Concentrator operational modes in
response to geological variation. Miner Eng. 2019;134(1):356–64.
32. Órdenes J, Wilson R, Peña-Grad F, Navarra A. Incorporation of geometallurgical input into gold
mining system simulation to control cyanide consumption. Fortschr Mineral. 2021;11(1023):1–16.
https://doi.org/10.3390/min11091023.
Potential Processes for Producing High-Purity
Lithium Hydroxide: A Critical Review
M. A. Halim1 (*), V. I. Lakshmanan1, J. Chen1, S. Kumar1, S. DasGupta1,

and M. Dehn2
1
Process Research Ortech Inc., Mississauga, ON, Canada
halim@processortech.com
2
United Lithium Corp., Vancouver, BC, Canada
Abstract. Demand of high-purity lithium hydroxide (LiOH) has increased significantly

with the rapid advancement of lithium batteries for clean technologies and energy storage.
Lithium hydroxide is commercially produced by the addition of lime into lithium carbon-
ate. While this process produces a low concentration of LiOH, a significant amount of
lithium is lost. To overcome these limitations, substantial studies have been conducted to
develop new technologies to produce battery-grade LiOH from lithium sulfate and chlo-
ride solutions by causticization and membrane electrolysis. Furthermore, the membrane
electrolysis processes have been shown to be a promising new technology for battery-
grade LiOH production. Therefore, this paper reviews these technologies with emphasis
on the process chemistry, key flowsheets, and challenges and potential improvement
opportunities of the technologies.
Keywords: Lithium hydroxide · Causticization · Membrane electrodialysis ·

Lithium sulfate · Lithium chloride · Evaporative crystallization · Process flowsheets
1 Introduction
The use of lithium batteries is consistently increasing in clean technologies, including

electric vehicles (EV), portable electronics, and energy storage, due to their large energy
charge and discharge capacity and no memory effects. High-purity (>99.3%) lithium
hydroxide monohydrate (LiOH.H2O) is primarily used as a raw material for the produc-
tion of cathode materials of lithium batteries. By 2030, LiOH is projected to account for
~57% of lithium compound demand, compared to 24% in 2019, while the consumption
of total lithium in the EV batteries is expected to account for 80% [1]. Currently, LiOH.
H2O is commercially produced by reacting lime with lithium carbonate (Li2CO3) [2–4].
This process, however, has a number of limitations including low concentration of lith-
ium ion in the product solution and excessive loss of lithium dragged by calcium car-
bonate as solid waste due to low solubility of calcium hydroxide [5–7].
A substantial number of studies have been conducted modify this commercial tech-
nology and to develop new methods for the production of battery-grade LiOH from
https://doi.org/10.1007/978-3-031-17425-4_78
646 M. A. Halim et al.
lithium carbonate, lithium sulfate (Li2SO4), and lithium chloride (LiCl) [3, 8–16]. It has
been reported that the battery-grade LiOH can be produced by causticization of lithium
sulfate obtained from hard rock, followed by sodium sulfate crystallization [8, 9].
Furthermore, the membrane electrolysis process utilizing anionic and cationic exchange
membranes has been shown to be a promising new technology for producing a high
concentration of LiOH under optimal conditions, including current density, pH, tem-
perature, and initial lithium concentration [1, 10–16]. Therefore, this paper is to criti-
cally review the existing and new technologies to produce battery-grade LiOH from
different feed materials.
2 Lithium Hydroxide Production
A number of methods, including causticization of Li2CO3, Li2SO4, and LiCl, and elec-
trolysis of Li2SO4 and LiCl have been developed for the production of high-grade lith-
ium hydroxide. Some of the potential technologies are discussed below.
2.1 Lithium Hydroxide Production by Causticization
Causticization of Lithium Carbonate

Lithium carbonate is mainly produced from brine. Lithium hydroxide is commercially
produced from Li2CO3 by the following metathesis reaction with calcium hydroxide
(CaCO3) [2–5, 7, 17]:
Li 2 CO3 Ca OH 2 2 H 2 O 2 LiOH.H 2 O CaCO3 . (1)
This process consists of a number of stages, including dissolution, conversion, thick-

ening, evaporation, crystallization, and drying. The features of the potential process
steps and a flowsheet are shown in Table 1 and Fig. 1, respectively. This process yields
solutions containing up to 3.5% LiOH [2]. Attempts to work at higher concentrations
lead to loss of Li2CO3 in the CaCO3 residue, because the solubility of Li2CO3 is reduced
in the presence of LiOH. This process is simple and produces high-purity LiOH.
H2O. However, it has a number of limitations, including a low concentration of LiOH
in the product solution and excessive loss of lithium dragged by calcium carbonate as
solid waste due to low solubility of calcium hydroxide [8].
Causticization of Lithium Sulfate
Hard rock consisting of spodumene is one of the potential sources for commercial lith-
ium production. Calcination of spodumene concentrate at ~1050 °C for ~30 min fol-
lowed by sulfuric acid roasting at ~250 °C for 30 min, water leaching, and impurities
removal by precipitation and ion exchange have become the main methods to produce
high-purity Li2SO4 solution. A number of studies have been carried out to produce
high-purity LiOH.H2O by causticization of Li2SO4 with NaOH [9, 18, 19]. Conversion
of Li2SO4 into LiOH.H2O with NaOH is shown by the following reaction:
Li 2 SO 4 2 NaOH 2 H 2 O 2 LiOH.H 2 O Na 2 SO 4 . (2)

Potential Processes for Producing High-Purity Lithium Hydroxide: A Critical Review 647
Table 1. Main features to produce LiOH.H2O from Li2CO3 with Ca(OH)2.

Process step Description
Dissolution and Preparing ~30 wt.% suspensions of Li2CO3 and Ca(OH)2
Conversion Mixing these suspensions at 60–70 °C to produce LiOH and precipitate CaCO3
Thickening Separation of LiOH solution from CaCO3 by counter-current four stages
thickening and washing
Solid-liquid separation of thickener underflow (U/F) and overflow (O/F)
consisting of ~40 wt.% and ~2 wt.% solids, respectively
Maintaining 6–8% LiOH in filtrate while washing with water with ~86% and
99% of washing efficiency for U/F and O/F, respectively
Evaporation and Two-stage evaporation of LiOH filtrate to concentrate LiOH
Crystallization Vacuum crystallization and separation to produce LiOH.H2O and recycling of
aqueous to the evaporators
LiOH Solution Concentrating LiOH by evaporation of the filtrate obtained from Na2SO4.10H2O
Evaporation crystallization step
LiOH.H2O Production of LiOH.H2O from LiOH concentrate solution by vacuum
Crystallization evaporative crystallization
Drying Drying product at 70–75 °C in a rotatory dryer in the presence of a mixture of
air and combustion gasses to reduce the moisture from ~4% to 0.05%
Fig. 1. Process flowsheet to produce LiOH.H2O from Li2CO3 with Ca(OH)2 [8].
The features of this process are summarized in Table 2. This process demonstrations
high yield of LiOH.H2O product. However, it has a number of limitations, including
difficulty in removing the highly soluble Na2SO4 and significant amount of lithium loss
due to the entrapment in the Na2SO4 crystals [8].
Liu and Azimi [8] have proposed a process flowsheet for producing battery-grade
LiOH⋅H2O from Li2SO4 solution with barium hydroxide (Ba(OH)2) as a causticizing
Table 2. Features of causticization process to produce LiOH.H2O from Li2SO4 with NaOH.
Causticization Production of LiOH from Li2SO4 solution with NaOH solution
Na2SO4.10H2O Crystallization of Na2SO4.10H2O in between −10 and −5 °C
Crystallization Separation of Na2SO4.10H2O by centrifuge
Na2SO4.10H2O Drying Evaporative crystallization and drying to produce anhydrous Na2SO4
LiOH Solution Concentrating LiOH by evaporation of the filtrate obtained from
Evaporation Na2SO4.10H2O crystallization step
LiOH.H2O Production of LiOH.H2O from LiOH concentrate solution by vacuum
Crystallization evaporative crystallization
LiOH.H2O Drying Drying of LiOH.H2O crystals under CO2 free air
Fig. 2. Flow diagram for producing battery-grade LiOH⋅H2O from Li2SO4 with Ba(OH)2 [8].
Table 3. Features of causticization to produce LiOH.H2O from Li2SO4 with Ba(OH)2.

Causticization Producing LiOH from Li2SO4 with Ba(OH)2 slurry
Removing residual SO42− with excess Ba(OH)2 at 70 °C for ~60 min
Vacuum filtration to separate BaSO4 from LiOH solution
Excess Ba2+ Removal Removing Ba2+ from LiOH solution with Li2CO3 at ~70 °C
Entrained Li+ Recovery Recovering entrained Li+ by washing BaSO4 cake with water
Recovered ~50% of entrained Li+ by four times washing
LiOH.H2O Evaporative Removing residual CO32− with BaCl2 from LiOH.H2O solution
Crystallization Evaporative crystallization in an inert atmosphere by vacuum
evaporation at ~90 °C to produce 99.6% pure LiOH.H2O
Ba(OH)2 Reproduction Conversion of BaSO4 into BaO in a box furnace at 1600 °C for 5 h
Production of Ba(OH)2 by adding water into BaO
agent (Fig. 2). The potential process design parameters are summarized in Table 3. The
causticization reaction is shown by the following reaction:
Li 2 SO 4 Ba OH 2 2 H 2 O 2 LiOH.H 2 O BaSO 4 (3)

The authors [8] reported that battery-grade LiOH.H2O (99.6%) can be produced
from Li2SO4 solution with Ba(OH)2. However, this process was evaluated based on a
synthetic Li2SO4 solution, which is much more pure compared to the real pregnant
leach solution (PLS). Furthermore, there is no information regarding other impurity
metals, such as aluminum (Al) sodium (Na), potassium (K), calcium (Ca), magnesium
(Mg), and manganese (Mn), which may present in the real PLS. Barium (Ba) may
report in the precipitate of these metals if Ba(OH)2 is used as a neutralizing reagent.
Causticization of Lithium Chloride
Sharma [12] reported that ~99.9% pure LiOH.H2O can be produced by causticizing of
lithium chloride (LiCl) with NaOH (Fig. 3). This process includes (i) precipitation of
LiOH from LiCl with stoichiometric addition of NaOH, (ii) LiOH.H2O filtration and
recycling of filtrate to the leaching stage, (iii) production of anhydrous LiOH by heating
LiOH.H2O at ~500 °C under nitrogen atmosphere, (iv) filtering the anhydrous LiOH
product by ceramic filters and recycling the solid residue to the leaching stage, (v)
quenching the anhydrous LiOH with water to produce LiOH.H2O crystals, and (vi) dry-
ing LiOH.H2O under vacuum at 45 °C.
2.2 Lithium Hydroxide Production by Membrane Electrolysis
Membrane technology is one of the potential processes which has been evaluated to
produce high-purity LiOH.H2O from LiCl and Li2SO4 solutions [20]. Highly purified
feed solutions used in the electrolysis/electrodialysis are obtained from brines, leaching
of hard rocks, and dissolution of Li2CO3 in acid medium. Electrodialysis can be carried
out by either a two-compartment or three-compartment cell (Fig. 4). A cation exchange
membrane (CEM) is used to separate an anode from a cathode in the two-compartment
cell in which the feed solution is placed in the anode compartment. In contrast, the
three-compartment cell consists of an anode, a cathode, an anion exchange membrane
Fig. 3. Flowsheet to produce high-purity LiOH.H2O from LiCl solution with NaOH [12].
Fig. 4. Three-compartment electrodialysis cell to convert Li2SO4 into LiOH.
(AEM), and a CEM in which AEM separates the anolyte from the feed solution and
allows anions, e.g., sulfate (SO42−) to transfer into the anolyte. Similarly, the CEM sepa-
rates the catholyte from the feed solution and permits lithium ion (Li+) to transfer into
the catholyte. In both of these electrodialysis cells, water reacts at the anode and cath-
ode according to the following reactions:
Anode : 2 H 2 O O2 4 H 4e (4)
Cathode : 2 H 2 O 4e H 2 2 OH (5)
The cathodic reaction produces hydrogen gas (H2) and hydroxyl ions (OH−), which
reacts with Li+ to form LiOH. This LiOH solution is used in the subsequent downstream
process steps, including evaporation and crystallization to make high-purity LiOH.
H2O. In contrast, oxygen gas (O2) and hydrogen ions (H+) are produced by the anodic
reaction. This H+ ion reacts with anion, e.g., SO42− to produce sulfuric acid (H2SO4).
Hydrogen and O2 gases are released to the atmosphere (Fig. 4).
Nemvok et al. [11] reported a waste-free electrolysis technology to produce high-
purity LiOH⋅H2O from technical-grade Li2CO3 and Li2SO4 with a two-compartment
cell. This electrolytic conversion was evaluated with CTIEM-1 and CTIEM-3 mem-
branes, which are analogs to Nafion 900 and Nafion 434, with a current density of 10
A/dm2, 25–50 g/L H2SO4 anolyte acidity, initial feed solution of 220 g/L Li2SO4, flow
rate of 80–160 mL/min, and 2 mm thickness of the electrode compartments. It was
found that the current efficiency of LiOH solution decreases linearly from ~70% to
~35% with its increasing concentration from ~1.0 M to ~4.0 M, respectively. It may be
due to the concentration of LiOH increasing the proportion of charges species to the
amount of water present becomes significant. Water molecules are required to form
solvent cages around the ions to keep them in solution. The increasing ion concentra-
tion can limit the conductivity of the solution.
Grageda et al. [7] evaluated the effects of current density, electrode material, electro-
lyte concentration, temperature, and cation exchange membrane on the recovery of
battery-grade LiOH from lithium brines. In this study, up to 99.93% pure LiOH.H2O
was achieved with a thicker membrane Nafion 117, nickel as the cathode material, at
75 °C, 2.30% LiOH initial concentration, average voltage of 5.15 V, 51.65% current
efficiency, and an energy consumption of 10.79 kWh/kg LiOH at 2400 A/m2. Moreover,
Ryabtsev et al. [20] used membrane electrolysis to obtain LiOH from Li2CO3 treated
with H2SO4 and managed to obtain an average of 45 g/L LiOH.
Bipolar membrane electrodialysis has recently been evaluated for the recovery of
LiOH from LiCl and Li2SO4 solutions. This technology consists of a cation exchange
layer, an anion exchange layer, and an intermediate zone that allows water catalysis of
dissociation into OH− and H+ ions without gas generation [21]. Jiang et al. [22] evalu-
ated bipolar membrane electrodialysis to produce LiOH from brines with
0.054 M–0.18 M Li2CO3 solutions. They achieved a current efficiency of 91.8–94.2%
and an energy consumption of 6.66 kWh/kg LiOH at 300 A/m2. Melnikov et al. [23]
produced up to 0.3 M LiOH from LiCl solution consisting of 1.8–59% organic solvents
and achieved a current efficiency of 60% and an energy consumption of 6.6 kWh/kg. In
contrast, Bunani et al. [24] studied simultaneous recovery of lithium and boron from
aqueous solutions by bipolar membrane electrolysis at different voltages, with >90%
recovery efficiency. It was observed that the current efficiency is decreased with increas-
ing applied voltage. González et al. [1] achieved ~77% current efficiency at initial con-
centrations of 0.5 wt% LiOH and 14 wt% LiCl and produced 3.34–4.35 wt% LiOH
with a solution purity of 96.0–95.4%. Furthermore, a membrane electrodialysis process
was developed for the recovery of LiOH from lithium-containing brine [25, 26]. Chen
et al. [27] reported that purified 1.5 mol/L Li2SO4 solution can be converted into
2.2 mol/L LiOH with a specific energy consumption of <10 kWh/kg LiOH.
3 Conclusion
Development of innovative processes for the production of battery-grade LiOH.H2O

from treated brines, process Li2SO4 solution, and Li2CO3 is critically important as the
demand of LiOH is consistently increasing for clean technologies and energy storage.
This paper reviews and summarizes the potential features of important processes on the
production of high-purity LiOH.H2O. Causticization of Li2CO3 with Ca(OH)2 is simple
and produces high-purity LiOH.H2O, while excessive loss of Li dragged by CaCO3 as
solid waste due to low solubility of Ca(OH)2 is the main disadvantage. Conversions of
Li2SO4 solutions into LiOH.H2O with NaOH and Ba(OH)2 are a potential alternative
process option. However, these causticization processes need to be tested in a continu-
ous pilot plant operation to optimize the process design parameters and to evaluate the
loss of Li in the Na2SO4 and BaSO4 crystals.
Membrane technology using either two-compartment or three-compartment and
bipolar membranes consisting of electrodialysis cells could be the potential alternative
to the conventional causticization processes for producing high-grade LiOH.
H2O. Electrodialysis can also produce acid, which can be recycled and reused in the
flowsheet. This review has discussed the potential operating parameters, recovery of
LiOH, current efficiency, and energy consumption. It has been reported that ~99.93%
Fig. 5. Proposed process flowsheet to produce high-purity LiOH.H2O from Li2SO4 solution.
pure LiOH.H2O can be produced with Nafion 117 at an average voltage of 5.15 V at
75 °C with a current efficiency of 51.65%, and energy consumption of 10.79 kWh/kg
LiOH at a current density of 2400 A/m2 by two-compartment electrodialysis. In con-
trast, 1.5 mol/L Li2SO4 can be converted to 2.2 mol/L LiOH with a recovery of ~73.3%
at a specific energy consumption of <10 kWh/kg LiOH by bipolar electrodialysis.
Although some of these studies have shown a better LiOH recovery with a higher cur-
rent efficiency, a significant process improvement can be performed to produce battery-
grade LiOH.H2O from crude Li2SO4 solution by recycling the process streams and
optimizing the design parameters by following the proposed flowsheet as shown in
Fig. 5.
References
1. González A, Grágeda M, Quispe A, Ushak S, Sistat P, Cretin M. Application and analysis of bipo-
lar membrane electrodialysis for LiOH production at high electrolyte concentrations: current scope
and challenges. Membranes. 2021;11:575.
2. Wietelmann U, Bauer RJ. Ullmann’s encyclopedia of industrial chemistry. Weinheim: Wiley-
VCH; 2000. https://doi.org/10.1002/14356007.a15_393.
3. Buckley D, Genders JD, Atherton D. Method of making high purity lithium hydroxide and hydro-
chloric acid. US 2011/0044882 A1. 14 Feb 2011.
4. Turan Z, Baloglu H, Ünveren E, Bulutcu AN. The behaviour of Nafion® 424 membrane in the elec-
trochemical production of lithium hydroxide. J Chem Technol Biotechnol. 2016;91(9):2529–38.
https://doi.org/10.1002/jctb.4853.
5. Grageda M, Gonzalez A, Quispe A, Ushak SA. Analysis of a process for producing battery grade
lithium hydroxide by membrane electrodialysis. Membranes. 2020;10:198.
6. Nemkov NM, Ryabtsev AD, Ìujin VV. Preparation of high purity lithium hydroxide monohydrate
from lithium-containing wastes of various industries. Izv Tomsk Polytech Univ. 2004;77:80–4.
7. Grágeda M, González A, Alavia W, Ushak S. Development and optimización of a modified process

for producing the battery grade LiOH: optimization of energy and water consumption. Energy.
2015;89:667–77.
8. Liu H, Azimi G. Production of battery grade lithium hydroxide monohydrate using barium hydrox-
ide causticizing agent. Resour Conserv Recycl. 2022;179:106115.
9. Nanping L, Jianmin G, Xiaohui J. Production process of lithium hydroxide monohydrate.
CN1214981C. 17 Aug 2005 (Chinese).
10. Gmar S, Chagnes A. Recent advances on electrodialysis for the recovery of lithium from primary
and secondary resources. Hydrometallurgy. 2019;189:105124.
11. Nemkov NM, Ryabtsev AD, Kotsupalo NP, Menzheres LT, Mamylova EV, Chayukova OI. Preparing
high-purity lithium hydroxide monohydrate by the electrochemical conversion of highly soluble
lithium salts. Theor Found Chem Eng. 2018;54(4):710–8.
12. Sharma Y. Production of lithium hydroxide. WO2016/070217A1. 12 May 2016.
13. Magnan J-F, Bourassa G, Pearse G, Symons P, Genders JD, Clayto G. Processes and system for
preparing lithium hydroxide. US10036094B2. 31 July 2018.
14. Magnan J-F, Bourassa G, Pearse G, Symons P, Genders JD, Bar D, Langevin M-E. Processes for
preparing lithium hydroxide. EP2971252B1. 30 Dec 2020.
15. Binnemans K. Method for producing battery grade lithium hydroxide monohydrate.
WO2021228936A1. 18 Nov 2021.
16. Mukul MS, Barton SB. Electrolysis process for making lithium hydroxide. WO2021212018A1.
21 Oct 2021.
17. Hader RN, Nielsen RL, Herre MG. Lithium and its compounds. Ind Eng Chem. 1951;1951:2636.
18. Yao K, Jin P, Huo L, Tu M, Dong H, Huang C, Xia G. Method for preparing battery-stage mono-
hydrate lithium hydroxide. CN 200710051016.5. 9 July 2007.
19. Jian G, Peng W, Yaoyao L, Min Z, Wenxuan L, Huan Z. Solid-liquid phase equilibria in the pro-
cess of producing lithium hydroxide from lithium sulfate. Chem Res Chin Univ. 2019. https://doi.
org/10.1007/s40242-019-9197-z.
20. Ryabtsev AD, Nemkov NM, Kotsupalo NP, Serikova LA. Preparation of high-purity lithium
hydroxide monohydrate from technical-grade lithium carbonate by membrane electrolysis.
Zhurnal Prikladnoj Khimii. 2004;77:1123–31.
21. Koter S, Warszawski A. A new model for characterization of bipolar membrane electrodialysis of
brine. Desalination. 2006;198:111–23.
22. Jiang C, Wang Y, Wang Q, Feng H, Xu T. Production of lithium hydroxide from lake brines through
electro-electrodialysis with bipolar membranes (EEDBM). Ind Eng Chem Res. 2014;53:6103–12.
23. Melnikov S, Sheldeshov N, Zabolotsky V, Loza S, Achoh A. Pilot scale complex electrodialysis
technology for processing a solution of lithium chloride containing organic solvents. Sep Purif
Technol. 2017;189:74–81.
24. Bunani S, Yoshizuka K, Nishihama S, Arda M, Kabay N. Application of bipolar membrane elec-
trodialysis (BMED) for simultaneous separation and recovery of boron and lithium from aqueous
solutions. Desalination. 2017;424:37–44.
25. Harrison S, Blanchet R. Processes for preparing highly pure lithium carbonate and other highly
pure lithium containing compounds. EP 2749535 A1. 16 Oct 2012.
26. Buckley D, Genders JD, Atherton D. Method of making high purity lithium hydroxide and hydro-
chloric acid. US Patent 2011/0044882. 24 Feb 2011.
27. Chen X, Ruan X, Kentish SE, Li GK, Xu T, Chen GQ. Production of lithium hydroxide by electro-
dialysis with bipolar membranes. Sep Purif Technol. 2021;274:119026.
The Revival of the French Rare Earth Industry
Alain Rollat1 (*)
1
Carester, Lyon, France
Abstract. In the 1980s and 1990s, France was one of the major countries for the Rare
Earth industry and the French RE separation technology was recognized as one of the
most advanced in the world. In the early 2000s, most of the French separation capacities
moved to China. Due to this situation and the development of end-use applications in
China, the French RE industry has gradually declined and is now very weak. It is the aim
of Carester to revive and develop the Rare Earth industry in France through two initia-
tives: An end-of-life magnet recycling unit and a Heavy Rare Earth separation hub (HRE
Hub).
The needs of magnet recycling in Europe have been highlighted by the EU and
Carester decided to participate in this ambition. The recycling unit will combine new
technologies allowing to improve the environmental footprint with the efficient sepa-
ration technology proposed and developed by Carester.
There is no doubt that in the coming years several plants dedicated to praseo-
dymium (Pr) and neodymium (Nd) production will start outside of China. Most of
these projects are based on Light Rare Earths (LRE) deposits with low HRE content.
Therefore, the HRE separation step is usually not economic for these companies and
not included in their projects. Based on its industrial experience Carester has the
ambition to propose to these companies a HRE Hub able to treat their HRE concen-
trate and give them back the pure rare earths they need, in particular terbium (Tb) and
dysprosium (Dy).
Keywords: Rare Earths · Magnet recycling · Solvent extraction
1 Introduction
In the 1980s and 1990s, France was one of the major countries for the Rare Earth (RE)
industry and the French RE separation technology was recognized as one of the most
advanced in the world. In the early 2000s most of the French separation capacities
moved to China. Due to this situation and the development of end-use applications in
China the French RE industry has gradually declined and is now very weak. It is the aim
of Carester to revive and develop the RE industry in France through two initiatives: An
end-of-life (EoL) permanent magnet recycling unit and a Heavy Rare Earth separation
hub (HRE Hub).
https://doi.org/10.1007/978-3-031-17425-4_79
656 A. Rollat
2 Recycling of End-of-Life Permanent Magnets
The needs for magnet recycling in Europe have been highlighted by the European
Union (EU) and Carester decided to participate in this endeavor. A market study done
with the help of the French Agency for Ecological Transition (ADEME) shows that the
resource is already huge within the EU (3400 t of Nd in 2020) and will reach 13,000 t
of Nd and 1300 t of Dy by 2040. Although the resource is large enough it is important
to assess the economically available part of it. The analogy with the geological concept
of cutoff grade is used to assess the economically realistic reserve within the whole
resources.
The recycling unit will have a capacity of 2000 t of magnets per year. This unit will
combine new technologies allowing an improvement in the environmental footprint
using the efficient separation technology proposed and developed by Carester.
At first, the ground EOL magnets are oxidized in very specific conditions with the
aim to improve the selective dissolution of RE vs. Fe. The oxidizing conditions aim to
avoid the formation of the mixed oxide NdFeO3 and to get the neodymium as Nd2O3
and the iron as Fe2O3. This target can be achieved by carefully controlling the oxidation
temperature. The oxidized magnet is then treated by nitric acid, and after a liquid/solid
separation the iron oxide is sent to waste. A commercial action is underway to check the
possibility to valorize this solid, for example, in the water treatment industry. The
nitrate solution obtained after the solid/liquid separation is treated by a novel solvent
extraction (SX) process to recover pure boron in the form of commercial borax. The
extraction of boric acid with 1,3,Diols [1] and aliphatic alcohols [2] has been studied in
the recent years. But these molecules present some industrial drawbacks due to high
viscosity or limited extraction efficiency at low B content. A new solvent formulation
with a specific extractant–modifier–diluent composition has been designed allowing to
overcome the thermodynamic and hydrodynamic limits of the systems described in the
literature.
The raffinate of this SX battery containing the RE nitrates is purified, separated by
SX, precipitated, and calcined leading to pure Pr, Nd, Tb, and Dy oxides which are sold
to metal makers and reintroduced in the magnet market. The SX design is based on
Carester’s expertise in RE separation. Carester experts have a long history in concep-
tion, design, and operation of RE separation units using either chloride route or nitrate
route. Both routes use the same mixer-settlers technology but differ in the chemistry
used. From a theoretical point of view, the real interest of the nitrate route is the possi-
bility to use a neutral molecule (tri-n-butyl phosphate or “TBP”) for the Pr and Nd
purification which does not need acid and alkali as are required by the chloride route
which uses an acidic extractant, such as PC88A. The nitrate route has been selected
rather than the chloride route due to its interest in terms of OPEX and environmental
footprint [3]. Indeed, the combination of this SX process with specific precipitating
chemicals allows to generate only ammonium nitrate solution as the effluent of the unit.
The ammonium nitrate solution can be sold as a fertilizer leading to a process without
any liquid waste. In contrast, the chloride route produces a copious effluent of dilute
NaCl-requiring disposal.
The Revival of the French Rare Earth Industry 657
3 Heavy Rare Earths Hub
Due to the increasing needs for NdFeB magnets, particularly for the market of e-mobility
and wind turbines, all the market studies show the necessity of opening new RE mines.
Most of the current RE mining projects are based on deposits containing mainly light
Rare Earths (LRE) minerals (in particular, bastnasite and monazite) with low heavy
Rare Earths (HRE) content. In this context, there is no doubt that in the coming years
several plants dedicated to LRE separation will start outside of China. All these projects
include a LRE separation unit focused on Pr and Nd purification, while the HRE are
usually not separated and produced as a concentrate. The reason for such a choice is
economic. For most of these companies a HRE separation unit would lead to a long
pay-back due to the low HRE content of their deposit. In the end, all these HRE con-
centrates produced by the future Pr and Nd producers will be sold as a mixed HRE to
the companies having a HRE separation unit, especially to the companies having Tb
and Dy purification units. Currently only Chinese companies have these capacities, and
the consequence of this situation is paradoxical. At the same time when the western
companies develop their independency from China for the Pr and Nd supply, they rein-
force their dependency on China for Tb and Dy supply.
Carester intends to propose to these companies a HRE Hub able to treat their HRE
concentrate and give them back the pure rare earths they need, in particular, Tb and Dy.
The Carester proposal is based on its expertise in RE separation. As was highlighted for
the magnet recycling process, Carester experts can design either chloride or nitrate
separation routes. When we consider HRE separation, TBP cannot be used since the
selectivity of TBP between the heavy lanthanides is very weak. However, another mol-
ecule that works in nitrate medium, Aliquat 336, presents an interesting selectivity
among the HRE. This molecule, a quaternary ammonium salt which is a kind of ionic
liquid, is similar to the neutral molecules and does not need alkali for loading the sol-
vent nor acids for stripping the solvent. Similar to LRE separation with TBP, HRE sepa-
ration with Aliquat 336 can lead to lower OPEX than PC88A typically used in the
chloride route. In the end, the decision to choose one or other route depends on specific
constraints, like the existence of a local ammonium nitrate market, the possibility to
release sodium chloride waste solution, the energy cost, and so on. In most of the cases
the nitrate route is more economic in terms of OPEX, but more expensive in terms of
CAPEX and the final decision is based on a tradeoff between OPEX and CAPEX.
In the specific case of HRE separation there is another point to consider for optimiz-
ing the separation process. Until 2010–2015, the HRE were mainly used in the phos-
phor application and for a small part in the magnet application. The phosphor application
needs Y, Eu, and Tb and the magnet application needs Sm, Tb and Dy as HRE. If we
add the Gd used in the nuclear and medical application, the overall HRE supply and
demand were quite well balanced. The current situation is totally different: The
phosphor market is dead, the NdFeB market exploded, and the Sm and Gd markets are
stable. In other words, the HRE market became a Tb and Dy market. The problem we
are facing is that in all the ores the Tb+Dy content represents only about 5–10% of the
total HRE for both the so-called LRE ores and the so-called HRE ores (Table 1).
658
Table 1. HRE composition of the main RE ores.

LRE ores HRE ores
A. Rollat
Country China USA Australia Malawi Australia Australia China Australia Sweden Brazil Australia
Company Northern MCP Lynas Mkango Hastings Iluka Southern ASM Leading Serra Northern
RE RE Edge Verde Minerals
Deposit Bayan Obo Mountain Mount Weld Songwe Yangibanaa Eneabbaa Middle Y Dubboa Norra Kärra Pela Browns
Pass Central zone Hilla Emaa rangea
Bearing Bastnasite Bastnasite Monazite Bastnasite Complex Monazite Ionic ore Complex Eudialyte Ionic ore Xenotime
mineral
Sm2O3 55.0% 67.4% 53.0% 22.4% 49.2% 26.5% 14.9% 8.6% 5.5% 9.0% 2.5%
Eu2O3 10.8% 10.6% 12.0% 6.0% 8.6% 1.5% 1.9% 0.3% 0.6% 0.5% 0.5%
Gd2O3 15.5% 12.8% 21.1% 14.1% 23.6% 17.0% 12.3% 8.6% 8.0% 8.2% 6.6%
Tb4O7 1.4% 0.9% 2.0% 1.7% 1.8% 2.2% 2.0% 1.2% 1.2% 1.4% 1.5%
Dy2O3 4.3% 2.5% 3.5% 8.6% 4.8% 10.3% 8.5% 7.8% 8.1% 7.6% 9.8%
Ho2O3 0.5% 0.3% 0.0% 1.5% 0.5% 1.7% 1.2% 1.5% 1.9% 1.6% 2.1%
Er2O3 0.5% 0.3% 0.4% 3.5% 0.6% 2.9% 4.1% 4.7% 5.5% 4.9% 6.0%
Tm2O3 0.0% 0.0% 0.0% 0.4% 0.0% 0.5% 0.9% 0.6% 0.8% 0.5% 0.8%
Yb2O3 2.4% 0.0% 0.0% 2.6% 0.4% 2.9% 3.2% 3.9% 5.0% 3.8% 4.9%
Lu2O3 0.0% 0.0% 0.0% 0.4% 0.0% 0.5% 0.6% 0.6% 1.0% 0.5% 0.7%
Y2O3 9.6% 5.2% 8.0% 38.9% 10.6% 33.8% 50.5% 62.0% 62.5% 62.0% 64.6%
TbDy/ 5.8% 3.4% 5.4% 10.3% 6.6% 12.6% 10.5% 9.1% 9.4% 9.0% 11.2%
HRE+Y
a
Project
The Revival of the French Rare Earth Industry 659
The HRE separation process schemes developed in the past are mainly based on the
needs for the phosphor market. These schemes need to be totally rethought to adapt
them to the new situation: purification of Tb and Dy.
The HRE Hub designed by Carester combines these 2 particularities: Nitrate route
and focus on Tb and Dy production for lowering the OPEX and the CAPEX. The plant
will have a nominal input capacity of 5000 t REO per year. This plant must be able to
treat HRE concentrates with very variable compositions. Indeed, as it can be seen in
Table 1, the HRE composition is very different depending on the REE-bearing mineral.
HRE concentrates from LRE minerals, like bastnasite or monazite, are mainly com-
posed of SmEuGd (SEG), while the typical HRE minerals (ionic ores and xenotime)
have an yttrium content up to 65%. A judicious choice of the sequence of separations
with different extracting molecules allows flexibility adapted to this variability. The
management of composition changes during operation will be possible thanks to the
dynamic software PAREX+ developed by Carester for the REs separation, in partner-
ship with the French Atomic Energy Commission (CEA). The final products will be
pure Tb4O7, pure Dy2O3 ready for metal makers, a mixture of SEG carbonate, and a
mixture of Ho-Lu+Y carbonate.
In the end, the HRE Hub will be integrated in a global RE platform including other
partners active in the magnet industry.
References
1. Karakaplan M, Tural S, Tural B, Turgut Y, Hosgören H. The solvent extraction of boron with syn-
thesized aliphatic 1,3-diols: stripping and extraction behavior of boron by 2,2,5-trimethyl-1,3-
hexanediol. Solvent Extr Ion Exch. 2004;22(6):897–911.
2. Xu Z, Su H, Zhang J, Liu W, Zhu Z, Wang J, Chen J, Qi T. Recovery of boron from brines with
high magnesium content by solvent extraction using aliphatic alcohol. RSC Adv. 2021;11:16096.
3. Berger C, Rollat A, Leveque A. Rare earths: how to optimize the process design? In: Proceeding of
the 59th conference of metallurgists, Toronto, Aug 2020.
Cobalt–Nickel Separations Using Supported Liquid
Membranes for End-of-Life Lithium-Ion Battery
Recycling
Mauro Dal-Cin1 (*), Krystal Davis1, Aaron King1, Lin Li1, Andrzej Nicalek1,
Gilles Robertson1, and Ben Yu1
1
Energy, Mining and the Environment Research Center, National Research Council of Canada,
Ottawa, ON, Canada
mauro.dal-cin@nrc-cnrc.gc.ca
Abstract. Lithium-ion battery recycling will be standard practice in the future as the
number of batteries used in electric vehicles and energy storage increase and eventually
reach their end of life. Cobalt and nickel are the two economic driving forces for battery
recycling. Solvent extraction (SX) further separates cathode materials if desired and
supported liquid membranes (SLMs) offer an alternate technique to apply SX. Separate
extraction, phase separation and stripping stages and possibly saponification of recy-
cled extractant in SX are replaced by a much simpler one-step approach. In this work,
Co2+ was recovered using flat sheet (FS)SLM from a synthetic feed with 10,000 ppmw
Co2+ and Ni2+ each, 0.5 M sodium acetate buffer and 20% H2SO4 strip phase. Three
Co2+ selective extractant systems (commercial phosphorus-based acids) were evaluated
based on Co2+ recovery rate and purity in the strip as a function of pH. The best case
scenario yielded a Co2+ purity of 99.5% in the strip phase at a recovery of 80%, with a
slight decrease to 99.3% when the recovery reached 90%. At 98% recovery the Co2+
purity decreased to 98%. Co2+ recovery rates decreased during all tests, from 10 to 12
gmol/m2/day to 2 gmol/m2/day. X-ray fluorescence-indicated extractants were depleted
in the membrane at the end of tests where 100% recovery was not achieved. The
FSSLM tests represent a challenging case; the feed contains ~0.051 moles of Co2+ and
the membrane 0.00008 moles of extractant, or a molar extractant:Co2+ ratio of ~1:630.
Hollow fibre (HF)SLM will have a more favourable ratio. HFSLM testing is expected
to commence this year with 12 m2 HF modules and 100 L of feed.
Keywords: Solvent extraction · Supported liquid membranes · Cobalt · Nickel

sulphate separation · Lithium-ion battery recycling
1 Introduction
The lithium-ion battery (LIB), with electric vehicles becoming an ever larger compo-
nent of their use, is expected to see secondary use in different applications, primarily in
stationary energy storage [1]. Currently, end of life for primary electric vehicle applica-
tion is at 80% state of health, meaning there is ‘plenty of life’ for secondary uses, such
https://doi.org/10.1007/978-3-031-17425-4_80
662 M. Dal-Cin et al.
as stationary storage coupled with renewable energy sources, such as solar or wind.
This would maximize the use of the batteries and increase the overall value [2]. When
such batteries have been exhausted and they are considered at their end of life, they are
to be recycled. The value of cobalt, in particular, is a primary driver for the economics
of battery recycling depending on what stage of the battery lifecycle it is at (i.e. before
complete exhaustion or after initial purpose ‘end-of-life’). Cobalt is defined as a critical
material [3] and the global supply is primarily from Congo (70%).
Cobalt–nickel separation is also important in primary nickel production. Currently,
stainless steel accounts for 71% of the world’s nickel consumption and batteries 4% [4].
The battery market will continue to increase as the estimated cumulative EV sales are
expected to reach 55–72 million vehicles by 2025 [5]. End-of-life LIB recycling will
become increasingly important in the future from the perspective of nickel recovery as
well as cobalt. Worldwide primary nickel production is from both sulphide and laterite
ore deposits with the latter representing an ever increasing source of nickel [6]. The
total nickel reserve is 296 million metric tonnes [7] of which 60% are in laterites
(oxides). Actual nickel production differs slightly, with laterites accounting for approxi-
mately 52% in 2010, 58% in 2014 and 68% in 2019 and 2020 [8]. Nickel recovery from
laterite ores uses several processes, including high pressure acid leaching or heap leach-
ing, which are both followed by solvent extraction (SX) and electrowinning. SX is used
after acid leaching to achieve Ni:Co mass ratios of 666:1 to meet LME Grade Nickel
[9] specifications. Not all laterite ores contain cobalt and some nickel sulphide-based
production uses SX to remove cobalt.
Cobalt–nickel separation is an intermediate step in an overall LIB cathode recycling
strategy. The batteries will have undergone numerous stages, including, discharge,
mechanical separation, digestion of the cathode in sulphuric acid, impurity removal and
partial neutralization. Numerous strategies have been proposed for the most efficacious
separation method, including SX or precipitation of pure or mixed metal hydroxides or
carbonates. Techno economic analysis (TEA) by Plugatyr et al. [10] showed a
precipitation-based process was viable over a wide range of plant capacity and cobalt–
nickel pricing scenarios. Hybrid processes using supported liquid membranes (SLMs)
and/or conventional SX and/or precipitation may also be viable.
Compared to conventional SX, SLM separations present several advantages, such as
reduced extractant inventory, no issues with phase separation, reduced evaporative
losses, oxygen-free operating conditions, modularity and less HVAC intensive opera-
tion in cold climates where separation ponds cannot be used [11]. A significant differ-
ence between SLM and conventional SX is that extraction and stripping are occurring
simultaneously on opposite sides of the membrane, eliminating the need for separate
saponification of the extractants before re-use in extraction. At the interface between
the feed and organic in the membrane, cobalt is preferentially complexed by the extract-
ant. The complex then diffuses towards the strip side of the membrane where cobalt and
protons are exchanged. The protonated extractant diffuses back towards the feed–mem-
brane interface. The extractant is de-protonated at the operating pH and complexes with
more cobalt, repeating the cycle. In the current FSSLM tests the pH is controlled via
addition of base to compensate for the release of protons into the feed.
Often discussed obstacles to SLM operation are membrane lifetime and membrane
module costs. Membrane lifetime refers to the integrity of the organic phase which is
Cobalt–Nickel Separations Using Supported Liquid Membranes for End-of-Life… 663
contained in the structure of the membrane, not the actual membrane itself. Dissolution
of the extractant(s) and diluents into the feed and stripper phases will eventually result
in loss of throughput and selectivity. This behaviour is expected, and from an engineer-
ing view point, can be accounted for. Deshmane et al. [12] reported several thousand
hours operating time of a hollow fibre SLM (HFSLM) with periodic replenishment of
the organic phase. Other options suggested in the literature include dispersion of the
organic phase in the feed, which, upon contact with the hydrophobic membrane, would
spontaneously wet/refill the microporous support [13]. The most robust operation
would be separate modules for extraction and stripping. Extraction would occur at the
outer surface of the hollow fibre membrane with feed in the shell side and organic phase
in the fibre lumen and pores. The exiting/loaded organic would proceed to the lumen
side of a second module with the stripping phase in the shell side at the outer surface of
the hollow fibre membrane [12].
Hollow fibre modules have construction similar to shell and tube heat exchangers. In
the case of HFSLM the tubes are hollow fibres with outer diameters between 100 and
500 μm. They generally have high packing density, ranging from 500 to 1300 m2/m3,
for 15–35 cm diameter housings, including the headers. These high packing densities
compensate for the transport rates and long experimental periods that will be noted in
the current FSSLM results. As an example, a 100-hour test in FSSLM which processes
300 g of feed is analogous to processing 100 L of feed with a 12 m2 module in 4–5
hours. HFSLMs can be operated in co-current or counter-current modes.
Historically, industrial applications of SLMs are rare, owing to membrane module
costs and primarily lifetime of the organic phase in the membrane. Module costs should
decrease in the future as demand increases. Currently a hollow fibre module with a
surface area of 57 m2 costs approximately $1150 USD (2020) and modules with larger
surface areas (up to 158 m2) are currently available. An industrial application of these
types of hollow fibre modules is gas–liquid contactors to saturate liquids with gases, or
the reverse process, to degas liquids. SLM operations also offer the ability to perform
separation with the extractants in an oxygen-free environment avoiding potential oxida-
tive degradation to which some extractants are susceptible [14].
The current work focused on the separation of cobalt and nickel from sulphate solu-
tions using supported liquid membranes (SLMs). Three extractant systems were com-
pared and ranked based on cobalt recovery rates and Co/Ni concentration ratios in the
stripper. Flat sheet membranes (25 cm2) were used in batch tests (300 g each of the feed
and strip phases) to evaluate the three extractant phases at various feed pHs between 4.0
and 6.0. Future work will extend to hollow fibre modules with 1–2, 12 and 57 m2 sur-
face areas and feed volumes of 100 L.
2 Materials
Test cells, Fig. 1, were constructed from acrylic. The effective membrane dimensions
were 5 cm × 5 cm, with nominal feed and stripper chamber volumes of 300 cm3 with
head space for addition of NaOH or KOH for pH control. Membrane samples were
installed between the two mirror-image half-cells. Gaskets were prepared from 1.59 mm
Fig. 1. Flat sheet supported liquid membrane test cell, feed cell on the left, strip on the right, nearing
completion of a cobalt separation from a 10,000 ppmw Co, 10,000 ppmw Ni initial feed.
thick Viton™ rubber sheet. Etatron DLX 0115 units were used for pH control and cali-
brated before each test. Mixing was provided using PTFE coated stirrers, Cowie
Technology Corp., with Cole-Parmer Ultra Compact mixers operating at 350 rpm.
All reagents were used as received. Cobalt sulphate heptahydrate and nickel hexahy-
drate were purchased from Alfa Aesar, both with 99.7% purity from their respective
certificates of analysis. Anhydrous sodium acetate was purchased from EM Science.
Assays for cobalt and nickel were performed using a Bruker Tracer 5i X-ray fluores-
cence spectrophotometer calibrated in house using standards in the range from 100 to
10,000 ppmw. Separate calibrations were used for the feed and stripper (20 wt% sulph-
uric acid) phases.
Membranes were hydrophobic polypropylene (Celgard 2500) with a thickness of
25 μm, porosity of 40% and Gurley permeability of 200 s. Pores are formed by con-
trolled stretching of isotactic propylene films. Extractants are phosphorus-based acids
available at commercial scale and were used as received at concentrations of
1 M. Exxsol™ D80, (Exxon Mobil), a de-aromatized hydrocarbon, was used as a dilu-
ent. Viscosity measurements of the organic phases were performed using an Anton-Paar
SVM 3000.
3 Methods
Feed solutions were freshly prepared with 10,000 ppmw each of divalent cobalt and
nickel, as sulphates, in reverse osmosis permeate water (conductivity 20 μS/cm).
Sodium acetate (0.5 M) was used as a buffer. The pH of the feed was adjusted using
20% sulphuric acid before introduction to the test cell. The stripper phase was 20%
sulphuric acid. The feed and stripper were added simultaneously to their respective
chambers to avoid generating a liquid head on the membrane.
Feed and stripper phase assays were performed by withdrawing 1.5 mL samples for
X-ray fluorescence. The addition of 2 N NaOH or KOH for pH control was monitored
to quantify nickel and cobalt mass balances for the duration of the test. Selectivity was
reported as the mass ratio of cobalt to nickel in the stripper.
Organic phases were prepared mixing the appropriate masses of each component,
according for the manufacturer’s stipulated purity to the equivalent of 1 M concentra-
tion. Organic phases with a commercially available synergist were based on 1 M of the
primary extractant. Membrane samples were prepared by immersion in the organic
phase taking care to wet only one side of the membrane first, followed by submersion.
Before installation in the test cells, membranes were blotted on lint-free tissue to
remove surface deposits of the organic phase, the cell assembled and the feed and strip-
per phases added in the minimum possible time to avoid evaporative losses of the
organic phase. Dry membrane samples weighed 0.040 g and ‘wetted’ membranes, as
installed, weighed 0.12 ± 0.01 g for a net organic phase mass of approximately 0.08 g.
4.1 Co–Ni Separations with System A
The organic phase system A was tested at pH points spanning 4.0 and 6.0 and cobalt
recoveries over time are shown in Fig. 2. The performance at pH 4.0 was poor, 100%
cobalt recovery was not achieved and the recovery rates were obviously slow compared
to higher pHs. The recovery rate was vastly improved at pH 4.6, and continued to
improve up to pH 5.5 and then decreased slightly at pH 6.0. The recovery was >98% for
pHs between 4.6 and 5.0.
The effect of pH on the cobalt recovery rate is examined further in Fig. 3 where the
time to achieve 80% recovery is shown as a function of the pH. The time for pH 4.0 is
omitted to avoid compression of results at other pHs. As will be seen later, a recovery
of 80% will be a desirable option to terminate extraction of cobalt and look at alternate
processing steps. The optimality of pH 5.5 is more evident in Fig. 3: pH 6.0 and 5.0
require 10–12% longer processing times, while pH 4.6 requires 24% longer processing
time. The pH dependence of the cobalt recovery rate is a critical parameter in modelling
the overall performance of a hollow fibre module, where pH changes occur within the
Fig. 2. Cobalt recovery as a function of time and pH using organic phase A.

Fig. 3. Time to achieve 80% cobalt recovery as a function of pH for organic phase A.
Fig. 4. Cobalt/Nickel ratio in the strip phase using organic phase A as a function of pH and cobalt
recovery.
length of a fibre. Base consumption for pH control in the feed was quantitative with
cobalt recovery: 2 moles of OH− was added for each mole of cobalt transferred to the
strip phase, which corresponded to proton transport from the strip to the feed.
The ratio of Co/Ni concentrations in the strip phase is used as an indication of the
separation efficiency, Fig. 4. Conventional selectivity values determined as (Costrip/
Cofeed)/(Nistrip/Nifeed) are misleading as they continuously increase as cobalt recovery
approaches 100% (3000 and higher), even though the cobalt purity in the strip phase is
decreasing. The cobalt purity is shown in Fig. 4 for cobalt recovery greater than 75%.
The Co/Ni ratios are higher, up to 600, at lower recoveries and exhibit significant scatter
at low nickel concentrations (<25 ppmw).
The Co/Ni ratio was at least 100 (0.99% nickel) at 90% recovery at pH 5.0 and
higher. At pH 4.6 the nickel impurity increased to 2%. At pH 4.0, 90% recovery was not
achieved and the nickel in the strip was >10%. The cobalt purity in the strip phase was
similar for pH 5.0 and 6.0. One of the two 6.0 tests yielded lower purity at high cobalt
recovery, although at up to 50% recovery, the Co/Ni ratio was at least 300, similar to all
the pH 5.0/5.5/6.0 tests.
4.2 Co–Ni Separations with System B
The cobalt recovery performance of organic Phase B is shown in Fig. 5 at pH 5.0, 5.5
and 6.0. Near complete recovery was achieved at pH 5.0 and 5.5 but at pH 6.0, only 90
and 75% cobalt recovery were achieved. Recovery at shorter times was similar for all
pHs but recovery at pH 6.0 lagged behind the others after ~2500 minutes and then pla-
teaued. The Co/Ni ratio, Fig. 6, at pH 6.0 was similar to other pHs below 60% cobalt
recovery but quickly decreased at higher recovery. At 80% recovery there was no
Fig. 5. Cobalt recovery as a function of time and pH using organic phase B.
Fig. 6. Cobalt/Nickel ratio in the strip phase using organic phase B as a function of pH and cobalt
recovery.
distinct difference between pH 5.5 and 5.0, with Co/Ni between 75 and 170. Co/Ni
decreased further, as expected, to 42–106 at 90% recovery, for both pHs.
4.3 Co–Ni Separations with System C
Cobalt recovery with organic phase C is shown in Fig. 7 as a function of time and
pH. As with organic phases A and B, the cobalt recovery rate at pH 6.0 was slower
compared to pH 5.0 and 5.5 and 90% recovery was not achieved. It is noted that at
40% recovery, the cobalt recovery rate at pH 6.0 was on par with pH 5.0 and 5.5.
Recovery rates at pH 5.0 and 5.5 were indistinguishable. The Co/Ni ratio at pH 6.0
was 400 and 550 at 40% cobalt recovery, on par with pH 5.0 and 5.5, (not shown) but
dropped quickly to 20 at 80% recovery, Fig. 8. The Co/Ni ratio at pH 5.5 was 200 at
Fig. 7. Cobalt recovery as a function of time and pH using organic phase C.
Fig. 8. Cobalt/Nickel ratio in the strip phase using organic phase C as a function of pH and cobalt
recovery.
80% recovery, the Co/Ni ratio at pH 5.0 varied considerably with values of 80–90 in
two tests and over 300 in a third test. At 90% cobalt recovery the Co/Ni ratio at pH 5.0
was 25 (where the recovery had plateaued), 50 and 100, with the pH 5.5 test also
achieving a ratio of 100.
4.4 Discussion
The ‘best’ organic phase for recovery of cobalt from a cobalt/nickel feed can be based
on recovery rate, Co/Ni concentration ratios in the strip phase or a combination of the
two. These two parameters are presented in Fig. 9, with the Co/Ni concentration ratio
in the strip phase plotted as a function of the time to recover 80% of the cobalt.
The error bars for the Co/Ni ratio are based on point values of the cobalt and nickel
concentrations. Small variations of the nickel concentration, in particular, can cause
large fluctuations in values of Co/Ni. However in some cases, such as organic phase C
at pH 5.0, large variations of Co/Ni were observed (Fig. 8), even when the cobalt recov-
ery rates were similar (Fig. 7).
From the presentation of results above for the three organic phases, the effect of pH
on recovery rate was insignificant at low cobalt recovery but became important when
the recovery was 80% or higher. At pH 6.0, organic phases B and C must be deemed
inadequate for large-scale application. At pH 5.0 and 5.5 all three organic phases exhib-
ited small variations in recovery time and Co/Ni selectivity. Organic phase A has the
benefit of maintaining high Co/Ni ratios and low recovery times at pH 6.0 offering the
broadest operating pH range without sacrificing stable performance. However, if the
feed pH decreases to 4.6 or lower, Co/Ni selectivity declines rapidly and processing
time increases. Decreasing the pH from 5.0 to 4.6 and 4.0 decreases the degree of
saponification/deprotonation at the membrane–feed interface by 2.5 and 10 times.
Fig. 9. Comparison of Co/Ni concentration ratios in the strip phase at 80% cobalt recovery, as a func-
tion of the time to achieve 80% recovery.
The lower cobalt transport at pH 6.0 noted for all three organic phases is not likely
due to nickel crowding out cobalt. The transport of nickel in the early stages of the
FSSLM test is 1/100th, or less, that of cobalt. The more probable cause of decreasing
cobalt transport at high pH is a viscosity increase of the organic phase if diffusion of the
cobalt-extractant complex is, or becomes, a rate-determining step. Increasing the pH
can increase the cobalt recovery rate by increasing the concentration of saponified or
deprotonated extractant at the membrane–feed interface. However, at some point the
viscosity will increase as cobalt loading becomes excessive, as in conventional SX,
which would reduce the cobalt transport rate [15, 16].
The viscosity of the native organic phases, Table 1, is important as a baseline of
expectations for mass transfer rates. However, they were poor predictors for relative
cobalt recovery rates. Organic phases B and C had similar viscosity, 3.2 cP, while phase
A was 33% higher, but had a faster cobalt recovery rate than phase B. Potentially, a bet-
ter indicator would be the viscosity of the organic phases during use. Carson et al. [17]
studied the effect of cobalt loading on the viscosity of Cyanex® 272 and Cyanex®272Y
in Isopar® M (Exxon Mobil). At high pH, the organic phase viscosity increased from
less than 5 to 280 cP as the cobalt loading increased to 90% and was attributed to the
formation of oligomers. Hereijgers et al. [18] studied Co/Ni separations in a flat-sheet
membrane contactor (SLM with feed and a bulk organic phase with no stripping) and
indicated that a 16 fold viscosity increase of the organic phase would account for devia-
tions of mass transport rates with their first-order kinetics model.
In addition to the increased viscosity of the organic phase via oligomer formation,
increasing the feed pH increases the solubility of most extractants as the degree of
saponification/deprotonation increases. The membranes are loaded with approximately
0.00008 gmoles (0.08 g of 1 M extractant) or approximately 0.024 g of extractant for a
nominal molecular weight of 300 g/gmole. This would yield a concentration of approx-
imately 76 mg/L if completely dissolved in the feed. As an example, Cyanex® 272,
di-2,4,4, trimethylpentyl phosphinic acid, has a reported solubility of 16 and 38 mg/L
at pH 2.6 and 3.7 respectively [19], and would be higher at the pHs used in this work.
Hence, the decreasing cobalt recovery rates after prolonged periods can also be attrib-
uted to the dissolution of the extractant into to the feed phase. This was seen by trace
colour build-up on the Teflon stir bar, similar in colour to that of the extractant loaded
with cobalt in shake-out tests. Losses to the strip phase were expected to be slower due
to the low pH and therefore near-zero saponification/deprotonation of the extractant.
Membranes scanned by X-ray fluorescence after completion of tests indicated a
decrease of >90% of the phosphorus compared to a membrane before testing.
Table 1. Organic phases and viscosities, standard errors on all viscosities were less than 0.25%.
Primary extractant
Organic phase concentration Viscosity
(Molarity) cP
D80 0.0 1.767
A 1.0 4.289
B 1.0 3.280
C 1.0 3.193
Dissolution of extractant and D80 occurred after long operating periods. This was
expected based on the known solubility of these components. Operating to the ‘failure’
point was purposely executed to determine operating limits of the system and how to
avoid them in HFSLM. Dissolution of the diluent and extractant would result in a ‘short
circuit’ between the feed and strip phases in some pores. This will form a path for diffu-
sion of the nickel rich feed phase, into the cobalt-rich strip phase via an aqueous path-
way. This was inferred from the appearance of potassium in the strip phase and a
concomitant increase of nickel, thereby decreasing the Co/Ni ratio, Fig. 10. The nickel
concentration in the strip phase was predicted by the potassium concentration increase as
a corresponding bulk transport and/or diffusion of all species in the feed across the mem-
brane in pores filled with the aqueous phase. Convective nickel transport was partly due
to a net liquid head on the feed side; a result of 2 N KOH added to the feed side for pH
control. Higher strength base could reduce the net head between the feed and strip but
caused some precipitation at pH 6.0. Under test conditions with high Co/Ni ratios in the
strip and fast cobalt transfer rates, K was not detected in the strip phase. There was a
2–4 cm liquid head in the feed over the strip phase; SLM can tolerate a liquid head before
the capillary forces are overcome. The rapid decay of the Co/Ni ratio may not occur in
HFSLM where the ratio of feed to membrane area is much lower than in FSSLM.
Further tests are required to replicate the desirable behaviour of organic phase A at
pH 5.0, 5.5 and 6.0. Likewise, the performance of organic phases B and C should be
extended to lower pHs, as these may simply have a different stable pH range.
Selection between the three organic phases is based on the cobalt recovery rate and
the Co/Ni ratio. Ultimately the priority of either performance parameter will depend on
flowsheet development. A flowsheet will incorporate multiple stages for cobalt recov-
ery with counter-current flow of the strip phase. In the current flat sheet tests, the pro-
cess is operating in the equivalent of co-current flow rather than counter-current flow.
In counter-current flow, in a hollow fibre module, or in a series of modules, both cobalt
recovery rates and Co/Ni ratios may improve.
Fig. 10. Measured and calculated nickel concentration and Co/Ni ratios in the strip phase, as a func-
tion of cobalt recovery.
Priority may be given to selectivity such that a high-purity cobalt phase can be pre-
cipitated. If selectivity is prioritized, organic phase A may be the better option, but
compared to organic phase B, the processing time is almost doubled, or the throughput
halved.
If a pure nickel phase for precipitation is the goal, a split stripper phase will be
required for further processing. Alternatively, a nickel selective SLM stage could be
used after 80–90% cobalt recovery to produce a high-purity nickel phase. This would
require a large SLM stage; purifying the nickel rich raffinate after cobalt extraction may
be more economic. These refinements and strategies for flowsheet development and
TEA studies will be conducted with performance data from HFSLM tests.
5 Conclusions
Supported liquid membrane separation of cobalt and nickel sulphates was evaluated
using three organic phases based on commercially available phosphorus acid-based
extractants in batch flat sheet tests. All three organic phases yielded very good Co/Ni
concentration ratios, 300–600, in the strip phase at cobalt recovery below 50%, decreas-
ing as the recovery increased. Phase A, at a cobalt recovery of 90%, yielded a Co/Ni
ratio of ~200 or 99.5% purity. The performance of this organic phase was favourable
from pH 5.0 to 6.0. Below this range, recovery rates and/or Co/Ni ratios degraded rap-
idly. Organic phases B and C both performed well at pH 5.0 and 5.5, but not at 6.0
where cobalt recovery was slower and did not reach 95%. These two phases should be
investigated at lower pH.
In general, cobalt recovery rates decreased for all the extractant systems from maxi-
mums of initial values of 10–12 gmol/m2/day to 2 gmol/m2/day at 100% recovery.
Decreasing cobalt recovery rates and Co/Ni ratios were attributed to losses of both
the extractant and diluent D80. When the recovery was extended beyond 80–90%, the
liquid height difference between the feed and strip sides may have contributed to bulk
feed transfer to the strip phase, thereby reducing Co/Ni ratios below what may be
encountered with hollow fibre SLM. In normal operation the SLM would be regener-
ated before allowing the organic phase to be depleted.
The flat sheet batch SLM tests suggest different possible strategies for flowsheet
design, offering a possibility of producing high-purity cobalt and nickel phases with
appropriate feed/raffinate and strip phase splitting. Future testing with hollow fibre
modules will provide data for flowsheet development, membrane area and reagent
requirements and other parameters for techno-economic analysis.
Acknowledgements This work was funded by the Office of Energy Research and Development,
Natural Resources Canada, NRC-19-109. The authors would like to thank Cindy Jiang of EME-NRCC
for viscosity measurements.
References
1. Faessler B. Stationary, second use battery energy storage systems and their applications: a research
review. Energies. 2021;14:2335. https://doi.org/10.3390/en14082335.
2. Hasan M, Standbridge CR. Post vehicle application lithium-ion batteryremanufacturing, repurpos-
ing and recycling capacity: modeling and analysis. J Ind Eng Manag. 2015;8(3):823–39. https://
doi.org/10.3926/jiem.1418.
3. Natural Resources Canada. https://www.nrcan.gc.ca/our-natural-resources/minerals-mining/
critical-minerals/23414. Accessed 16 Feb 2022.
4. Natural Resources Canada. https://www.nrcan.gc.ca/our-natural-resources/minerals-mining/
minerals-metals-facts/nickel-facts/20519. Accessed 20 May 2021.
5. International Energy Agency. Global EV Outlook 2021 – Analysis – IEA. Accessed 25 Feb 2022.
6. Dalvi AD, Bacon WG, Osborne RC. The past and the future of nickel laterites. In: PDAC 2004 inter-
national convention, trade show & investors exchange. Toronto: The prospectors and Developers
Association of Canada; 2004. p. 1–27.
7. Nickel Institute. The life of nickel. https://nickelinstitute.org/media/1190/thelifeofni.pdf. Accessed
21 May 2021.
8. Statista. https://www.statista.com/statistics/264642/nickel-mine-production-by-country/.
Accessed 21 May 2021.
9. Jones DL, McCoy TM, Mayhew KE, Cheng CY, Barnard KR, Zhang W. A new process for cobalt-
nickel separation. In: Proceedings ALTA. Perth: ALTA; 2010.
10. Baxter, C, Plugatyr, A and Yu, B. Lithium-ion battery recycling: a baseline flowsheet for processing
black mass in a mixed hydroxide precipitate and industrial grade lithium carbonate. published in
COM 2022, Montreal, Quebec, Canada, 21–24 Aug 2022.
11. Parhi PK. Supported liquid membrane principle and its practices: a short review. J Chem. 2013.
https://doi.org/10.1155/2013/618236.
12. Deshmane VG, Islam SG, Bhave RR. Selective recovery of rare earth elements from a wide
range of E-waste and process scalability of membrane solvent extraction. Environ Sci Technol.
2020;54(1):550–8. https://doi.org/10.1021/acs.est.9b05695.
13. Ho WSW, Sirkar KK. Membrane handbook. New York: WNR; 1992.
14. Barnard KR, Turner NL. LIX® 63 stability in the presence of Versatic 10 under proposed com-
mercial extract and strip conditions, part II: oxime isomer inter-conversion and the effect of oxime
degradation products on selected physical properties. Hydrometallurgy. 2008;91(1–4):11–9.
15. Swain B, Mishra C, Jeong J, Lee J-C, Hong HS, Pandey BD. Separation of Co(II) and Li(I) with
Cyanex 272 using hollow fiber supported liquid membrane: a comparison with flat sheet supported
liquid membrane and dispersive solvent extraction process. Chem Eng J. 2015;271:61–70. https://
doi.org/10.1016/j.cej.2015.02.040. ISSN 1385-8947.
16. Bukhari N, Chaudry MA, Mazhar M. Cobalt(II) transport through triethanolamine–cyclohexanone
supported liquid membranes. J Membr Sci. 2004;234:157–65. https://doi.org/10.1016/j.mem-
sci.2003.12.027. ISSN 0376-7388.
17. Carson I, Tasker P, Love J, Moser MT, Fischmann AJ, Jakovljevic B, Soderstrom M, Morrison
C. The supramolecular and coordination chemistry of cobalt(II) extraction by phosphinic acids.
Eur J Inorg Chem. 2018. https://doi.org/10.1002/ejic.201701473.
18. Hereijgers J, Vandermeersch T, Van Oeteren N, Verelst H, Song H, Cabooter D, Breugelmans
T, Malsche D. Separation of Co(II)/Ni(II) with Cyanex 272 using a flat membrane microcon-
tactor: extraction kinetics study. J Membr Sci. 2016;499:370–8. https://doi.org/10.1016/j.mem-
sci.2015.10.070. ISSN 0376-7388.
19. Solvay Cyanex 272 information brochure available by request from Solvay.
On the Solvometallurgical Extraction of Lithium
and Cobalt from Secondary Resources
Georgios Kolliopoulos1 (*) and Halimeh Askari Sabzkoohi1
1
Department of Mining, Metallurgical and Materials Engineering, Université Laval,
Quebec City, QC, Canada
Georgios.Kolliopoulos@gmn.ulaval.ca
Abstract. Energy, water, and metals are interconnected in a nexus shaped by the envi-
sioned electrification of our modern societies. Access to critical and strategic metals
(CSMs) is a vital part of this transition. In this work, we studied the solvometallurgical
extraction of lithium and cobalt from materials typically found in lithium-ion batteries,
using innovative green deep eutectic solvents (DESs). Both CSMs were leached in cho-
line chloride-based DESs with the extraction reaching more than 84% and 70% for lith-
ium and cobalt after 48 h, respectively.
Keywords: Sustainable processing · Solvometallurgy · Critical and strategic metals

· Urban mining · Waste valorization
1 Introduction
Lithium (Li) and cobalt (Co) are pivotal critical and strategic metals (CSMs) in the
transition of our modern societies to sustainable low-carbon economies. In 2020, the
World Bank estimated that by 2050, the demand for Li and Co will increase by 488%
and 460%, respectively [1]. Therefore, green and efficient urban mining will play a
significant role on our ability to alleviate supply chain disruptions caused by natural
disasters or geopolitical uncertainties, to cope with demand, and to ensure equitable
access to CSMs for all, which will in turn allow us to achieve the envisioned low-carbon
economies.
Recently, a solvometallurgical processing route for the sustainable extraction of
critical and strategic metals from both primary and secondary resources has been pro-
posed [2–7]. Sustainable processing of CSMs goes beyond the concept of CO2 emis-
sions and must also consider selectivity and water, energy, and waste minimization. The
solvometallurgical approach is similar to hydrometallurgical processing, as similar
operating conditions and equipment are used. However, there is one major difference:
solvometallurgical processing does not generate aqueous effluents since the consumption
https://doi.org/10.1007/978-3-031-17425-4_81
676 G. Kolliopoulos and H. A. Sabzkoohi
of water and acid/base reagents is limited, which could translate to a near zero-waste
processing approach [3, 8]. Instead, it relies on eutectic mixtures of common, low-cost,
non-toxic, and biodegradable salts to provide powerful green electrolyte-based solvents
with adjustable properties [9], namely, deep eutectic solvents (DESs). The DESs are
mixtures of salts, usually of a quaternary ammonium salt, such as choline chloride
(ChCl), and a metal salt or hydrogen bond donor (HBD) [10]. The use of different
HBDs leads to DESs with different properties, which may in turn affect metal solubility
and selectivity. The use of carboxylic acid HBDs in ChCl-based DESs has been found
beneficial to metal leaching compared to ChCl-based DESs containing amides and
polyols, such as urea and ethylene glycol, as the HBD [2]. However, ChCl:ethylene
glycol has been found to be more selective against iron [2], which is a significant con-
sideration when leaching iron-containing primary or secondary resources.
The DESs can be used effectively and efficiently in mineral and metal process cir-
cuits to selectively extract and recover metals while simultaneously minimizing water
and mineral acid consumption, which may in turn result in the elimination of aqueous
effluents. The aim of this work was to investigate the solvometallurgical extraction of
Li and Co from lithium cobalt oxide (LiCoO2), which is a common cathode material in
lithium-ion batteries. The effectiveness of two chemically stable ChCl-based DESs
with carboxylic acid HBDs, namely, levulinic acid (LA) and citric acid (CA), to extract
Li and Co from LiCoO2 was studied. Specifically, the extraction by leaching of both
metals was experimentally studied at a moderate temperature as a function of time in
ChCl:LA and ChCl:CA DESs.
Choline chloride (HOC2H4N−(CH3)3+Cl−; Sigma-Aldrich, 99%) was dried under vac-

uum for 24 h prior to being used. Levulinic acid (CH3COCH2CH2COOH, LA, Sigma-
Aldrich, 99%) and citric acid (HOC(COOH)(CH2COOH)2, CA, Sigma-Aldrich, ≥
99.5%) were used as received. Choline chloride was mixed with LA or CA at a 1:2 and
a 1:1 molar ratio, respectively. The mixtures were stirred at 80 °C until colorless homo-
geneous liquids, namely, the DESs, were formed. The DESs were then diluted with
20 wt% deionized (DI) water to lower their viscosity. The LiCoO2 (VWR, 97%) was
used as received. Leaching experiments were carried out in triplicates using 100-mL
glass flasks at 60 °C, 300 rpm, and a solid to liquid ratio of 20 g/L. The flasks were
placed in a heating shaking incubator for 48 h. Liquid samples were collected at pre-
determined time intervals, filtered, diluted with 5 wt% HNO3, and analyzed using
microwave plasma atomic emission spectroscopy (MP-AES, Agilent Technologies
4100+).
Metal extraction (%) was calculated using Eq. 1 [6]:

On the Solvometallurgical Extraction of Lithium and Cobalt from Secondary Resources 677
Cmetal VDES
Metal Extraction % , (1)
mLiCoO2 , init wt%metal

where Cmetal (g/L) is the concentration of Li and Co in the samples collected, as mea-
sured by MP-AES; VDES (L) is the volume of the DES; mLiCoO2 is the initial mass of the
LiCoO2 added in the DES; and wt%metal is the metal content in the LiCoO2: 0.07 for Li
and 0.60 for Co [6].
Figure 1 presents the metal extraction (%) for both Li and Co as a function of time,
while Fig. 2 presents the final extraction of Li and Co in both DESs tested. Metal extrac-
tion was relatively low in the ChCl:LA DES with a reported value of around 28.5% for
100
A B
80
Extraction (%)
60 Li - ChCl:LA
Co - ChCl:LA
40
20 Li - ChCl:CA
Co - ChCl:CA
0
0 1000 2000 3000 0 1000 2000 3000
Time (min) Time (min)
Fig. 1. The Li and Co extraction (%) in (A) ChCl:CA and (B) ChCl:LA. Leaching parameters: 60 °C,
300 rpm, and solid to liquid ratio of 20 g/L.
100
80
Extraction (%)
60
40
20
0
A
A
l:L
l:C
l:C
l:L
hC
hC
hC
hC
-C
-C
-C
o
Li
Li
C
C
Fig. 2. Final extraction of Li and Co in ChCl:CA and ChCl:LA after 48 h. Leaching parameters:
60 °C, 300 rpm, and solid to liquid ratio of 20 g/L.
678 G. Kolliopoulos and H. A. Sabzkoohi
both metals after 48 h. The extraction of both metals in ChCl:LA appears to have
reached a plateau after 24 h. However, after 48 h in the ChCl:CA DES, the extraction of
Li and Co exceeded 84% and 70%, respectively. The latter is a strong indication of the
suitability of the ChCl:CA DES for Li and Co leaching and may be attributed to the
presence of CA in the DES [6]. In fact, CA is a stronger acid than LA, which contrib-
uted to higher metal extractions in the ChCl:CA DES [6, 11, 12]. However, CA is a rela-
tively bulky molecule and its use in ChCl-based DESs increases the viscosity of the
resulting solvent. In fact, viscosity remains a challenge in solvometallurgy and often, to
alleviate this issue, the DESs are diluted with DI water. In this work, both DESs were
diluted with 20 wt% DI water, which resulted in less viscous solvents. This dilution
ensured that both solvents retained their fundamental DES properties, as it has been
found that most DESs retain their nanostructure in the presence of up to around 42 wt%
water [13].
Solvometallurgical processing promises higher selectivity in metal dissolution and a
near zero-waste mineral and metal processing approach by using DESs as the solvent
[3]. Deep eutectic protons are consumed during the leaching process before being
regenerated in the subsequent solvent extraction step, thus allowing the DES to be
regenerated and reused in leaching [6, 14]. The leaching process, although slower com-
pared to conventional hydrometallurgical processes, could be improved at higher agita-
tion speeds [6] or with dilution of the DES with water or ethylene glycol, which lowers
the DES viscosity. Finally, chloride anions are retained by hydrogen bonds in the DES
network [6] and as a result chlorine gas formation is prevented, as is the case in aqueous
leaching using hydrochloric acid.
4 Conclusions
Solvometallurgical processing seems capable to pave the path towards a new era of
sustainable mineral and metal processing, especially in urban mining. This work
focused on the extraction of lithium and cobalt from lithium-ion battery materials via
leaching with novel green deep eutectic solvents and minimal use of water. Both metals
were successfully leached in the solvents tested with more than 84% and 70% extrac-
tion for lithium and cobalt, after 48 h, respectively. Our results are an indication of the
potential success of solvometallurgy as a sustainable metal extraction and recovery
technique toward a green near zero-waste future of mineral and metal processing.
Acknowledgments The authors would like to acknowledge the Natural Sciences and Engineering
Research Council of Canada (NSERC) (Grant No. RGPIN-2020-04262) and the Fonds de recherche du
Québec–Nature et technologies (ERA-MIN 2 Grant; Project acronym: nanoBT) for the financial sup-
port of this research.
On the Solvometallurgical Extraction of Lithium and Cobalt from Secondary Resources 679
References
1. C.S.M.F. World Bank Group. Minerals for climate action: the mineral intensity of the, clean energy
transition. http://pubdocs.worldbank.org/en/961711588875536384/Minerals-for-Climate-Action-
TheMineral-Intensity-of-the-Clean-Energy-Transition.pdf. Consulted on June 15.
2. Rodriguez Rodriguez N, Machiels L, Onghena B, Spooren J, Binnemans K. Selective recov-
ery of zinc from goethite residue in the zinc industry using deep-eutectic solvents. RSC Adv.
2020;10:7328–35.
3. Binnemans K, Jones PT. Solvometallurgy: an emerging branch of extractive metallurgy. J Sustain
Metall. 2017;3:570–600.
4. Orefice M, Audoor H, Li Z, Binnemans K. Solvometallurgical route for the recovery of Sm, Co, Cu
and Fe from SmCo permanent magnets. Sep Purif Technol. 2019;219:281–9.
5. Önal MAR, Riaño S, Binnemans K. Alkali baking and solvometallurgical leaching of NdFeB
magnets. Hydrometallurgy. 2020;191:105213.
6. Peeters N, Binnemans K, Riaño S. Solvometallurgical recovery of cobalt from lithium-ion battery
cathode materials using deep-eutectic solvents. Green Chem. 2020;22:4210–21.
7. Sabzkoohi HA, Kolliopoulos G. Green zero-waste metal extraction and recycling from printed
circuit boards. Mater Proc. 2021;5:39.
8. Alonso DA, Baeza A, Chinchilla R, Guillena G, Pastor IM, Ramón DJ. Deep eutectic solvents: the
organic reaction medium of the century. Eur J Org Chem. 2016;2016:612–32.
9. Zhang Q, De Oliveira Vigier K, Royer S, Jérôme F. Deep eutectic solvents: syntheses, properties
and applications. Chem Soc Rev. 2012;41:7108–46.
10. Smith EL, Abbott AP, Ryder KS. Deep eutectic solvents (DESs) and their applications. Chem Rev.
2014;114:11060–82.
11. Abbott AP, Capper G, Davies DL, McKenzie KJ, Obi SU. Solubility of metal oxides in deep eutec-
tic solvents based on choline chloride. J Chem Eng Data. 2006;51:1280–2.
12. Rodriguez Rodriguez N, et al. Towards near-zero-waste recycling of mine tailings and metallurgi-
cal process residues through a novel solvometallurgical process based on deep eutectic solvents.
In: Jones PT, Machiels L, editors. Proceedings of the 4th international symposium on enhanced
landfill mining (ELFM IV), Mechelen (Belgium), 5–6 Feb 2018. p. 95–100.
13. Hammond OS, Bowron DT, Edler KJ. The effect of water upon deep eutectic solvent nano-
structure: an unusual transition from ionic mixture to aqueous solution. Angew Chem Int Ed.
2017;56:9782–5.
14. Tran MK, Rodrigues M-TF, Kato K, Babu G, Ajayan PM. Deep eutectic solvents for cathode recy-
cling of Li-ion batteries. Nat Energy. 2019;4:339–45.
The Beneficiation and Metallurgical Process
for the Production of Mixed REE Product
from Nechalacho Deposit
Kevin Bradley1 (*), Jing Liu1, Ray Anguelov2, and Tony Hadley2
1
Kevin.Bradley@sgs.com
2
Cheetah Resources Pty Ltd, Sydney, NSW, Australia
thadley@cheetahresources.com
Abstract. Cheetah Resources is Canada’s first rare earth producer (of modern times)
following development of its world-class rare earth Nechalacho project in the Northwest
Territories (NT) which commenced production in June 2021. Nechalacho’s North T Zone
hosts a high-grade resource at 9.01% LREO. The combined beneficiation and hydromet-
allurgical process developed for North T Zone is able to achieve high recovery of REE
from a parisite–bastnaesite ore body into a high-grade REE precipitate with relatively low
operating costs. In the process, the ore is first concentrated by ore sorting and then further
upgraded by dense media separation and Wilfley Table to produce a concentrate with a
40% TREO grade. The beneficiation concentrate is then subjected to calcination to pre-
treat the material prior to REE dissolution with hydrochloric acid which achieves partial
rejection of cerium relative to other REE. The REE-bearing pregnant leach solution is
purified by the removal of iron, thorium, uranium, and aluminum prior to precipitation of
a mixed REE carbonate of 98% purity. The overall Nd recovery from ore sorter concen-
trate to mixed REE carbonate is 82%. The resulting REE precipitate is further processed
to desired products by an off-take partner. This paper highlights some of the findings from
laboratory and pilot testing programs conducted to develop the process.
Keywords: Bastnaesite · Rare earth elements · Dense media separation · Wilfley

Table · Roasting · Leaching · Precipitation
1 Background
The Nechalacho REE deposit is located in Canada’s Northwest Territories. The Upper
Zone deposit contains a total resource (measured, indicated, and inferred) of 94.7 Mt at
1.46% REO. The high-grade North T Zone consists of 0.1 Mt at 9.01% LREO and min-
ing operations started on this deposit in June 2021 [1]. A photograph of starter pit min-
ing operations is included as Fig. 1 with the pit at top left, ore sorter at center right, and
the water settling pond at the bottom.
https://doi.org/10.1007/978-3-031-17425-4_82
682 K. Bradley et al.
Fig. 1. North T Starter Pit.
2 Sample Characterization and Mineralogy
One sample of ore sorter product (Bastnaesite Comp) weighing about 300 kg, from
Cheetah Resources’ Nechalacho REE Deposit, was received at SGS Lakefield for a
beneficiation and roasting pilot plant testwork program.
A representative subsample was split from the crushed Bastnaesite Comp (−2.0 mm)
and submitted for an ICP rare earth elements scan, whole-rock analysis, and particle
size analysis. As shown in Table 1, the head sample total rare earth oxide (TREO) con-
tent was 31.9% TREO. The key rare earth elements included light REE (LREO) cerium
(Ce), lanthanum (La), neodymium (Nd), praseodymium (Pr), and samarium (Sm). The
neodymium oxide (Nd2O3) grade was used in the following testwork to evaluate the test
performance. A ratio of 5.75 was used to estimate the grade of TREO from the Nd2O3
grade based on a previous test program employing similar material.
The head sample was subjected to semi-quantitative X-ray diffraction (SQ-XRD)
analysis which identified that the sample comprised primarily quartz, bastnaesite, and
parisite. XRD results are summarized in Table 2.
3 Beneficiation Testwork
The mineral processing testing included heavy liquid separation (HLS), dense medium
separation (DMS), and Wilfley Table testing.
The Beneficiation and Metallurgical Process for the Production of Mixed REE Product… 683
Table 1. Head assays of the Bastnaesite Comp.

Element Assay %, g/t Element Assay %
La2O3% 7.82 SiO2 47.7
CeO2% 16.0 Al2O3 0.88
Nd2O3% 5.54 Fe2O3 1.38
Pr6O11% 1.69 MgO 0.49
Sm2O3% 0.57 CaO 3.56
Eu2O3 g/t 408 Na2O 0.09
Gd2O3 g/t 1833 K2O 0.50
Tb4O7 g/t 113 TiO2 0.08
Dy2O3 g/t 277 P2O5 0.20
Ho2O3 g/t 43.2 MnO 0.03
Er2O3 g/t 29.8 Cr2O3 <0.01
Tm2O3 g/t 2.06 V2O5 <0.01
Yb2O3 g/t 9.45 LOI 12.9
Lu2O3 g/t 1.93 Sum 67.8
Y2O3 g/t <400
U3O8 g/t 17
ThO2 g/t 800
Sc g/t <70
TREO 31.9
Table 2. XRD results of Bastnaesite Comp.
Mineral Bastnasite Comp (Wt %)

Quartz 50.2
Parisite-(Ce) 26.7
Bastnaesite (Ce) 13.5
Neodymium Oxide Carbonate 5.7
Praseodymium Oxide Carbonate 2.0
Goethite 1.8
Total 100
3.1 Heavy Liquid Separation Testwork
Bench-scale heavy liquid separation (HLS) tests serve as a diagnostic tool to assess the
amenability of a sample to gravity separation by DMS or shaking table. It can also help
determine the optimum SG cut-points for the DMS testwork and to benchmark the
heavy media separation performance. A total of two samples with size fractions of
−2.0 + 0.5 mm and −0.5 + 0.15 mm were submitted for HLS tests.
HLS testing on each size fraction included four specific gravity cut-points (3.00,
2.90, 2.80, and 2.70) by immersing in a heavy liquid comprised of methylene iodide
diluted with acetone. The first pass was carried out at the highest specific gravity (3.00),
with each subsequent pass performed on the float fraction of the previous pass using a
heavy liquid with lower specific gravity. The sink product from each pass and the final
float product were weighed. Subsamples from the two fractions of each HLS product
were submitted for Nd2O3 and SiO2 assay.
The HLS results for the coarse size fraction (−2.0/+0.5 mm) and medium size frac-
tion (−0.5/+0.15 mm) are presented in Tables 3 and Table 4, respectively. Both HLS
tests demonstrated excellent results. The combined SG 2.90 sink products of the coarse
size fraction produced a 44.2% TREO concentrate with 95.8% REO recovery and a
61.7% mass yield. The HLS results indicated a great potential for upgrading the coarse
size fraction by dense media separation.
The combined SG 2.90 sink product of the medium size fraction produced a 51.8%
TREO concentrate with 97.4% REO recovery and a 56.8% mass yield. The medium
size fraction achieved slightly better TREO grade and recovery than the coarse size
fraction, likely due to better mineral liberation obtained at the finer size. The HLS
results indicated a good potential for upgrading the medium size fraction by conven-
tional gravity separation such as a Wilfley Table.
The combined HLS results of the coarse and medium size fractions at cut-off SG of
2.90 are presented in Table 5.
3.2 Dense Media Separation Pilot Plant Testwork
Based on the positive results obtained from the HLS tests, DMS pilot plant testing was
undertaken. The 170 kg of coarse size fraction (−2.0/+0.5 mm) of the Bastnaesite
Comp was processed through a DMS pilot plant to generate DMS concentrate.
DMS Pilot Plant Set-Up

The DMS pilot plant used was a DRA Americas pump-fed cyclone plant, fitted
with a 200 mm Multotec dense media cyclone. Sample pails containing the feed
were discharged into a vibratory hopper feeder. The discharge from the hopper
fed into a mixing box to allow for thorough mixing with dense media from the
circulating media sump. This mixture was then pumped to the dense media
cyclone, which separated particles based on their density relative to the SG cut-
point in the cyclone. Particles heavier than the SG cut-point reported to the
cyclone underflow, while particles lighter than the SG cut-point reported to the
cyclone overflow. The cyclone underflow and overflow were passed over parti-
tioned sections of a drain-and-rinse screen and rinsed with water to wash media
off the particles. Figure 2 shows the photo taken during the operation and depicts
the set-up of the DMS pilot plant.
The washed particles (screen oversize from the two screen sections) were collected as
the DMS sinks (cyclone underflow) and DMS floats (cyclone overflow), while the
drain-and-rinse screen undersize reported to a media recovery system. The DMS sinks
and DMS floats were finally fed to a fine aperture (~212 μm) circular vibrating screens
and rinsed with water to remove any remaining media from the particles.
Table 3. HLS test results of coarse size fraction (−2.0 + 0.5 mm).
HLS products Weight Assay, % Distribution, %
g % Nd2O3 TREO, Calc SiO2 Nd2O3 TREO SiO2
HLS Sink 3.00 1831 55.5 8.22 47.3 24.9 92.1 92.1 25.7
HLS Sink 2.90 203 6.2 2.94 16.9 66.1 3.7 3.7 7.6
HLS Sink 2.80 155 4.7 2.09 12.0 78.1 2.0 2.0 6.8
HLS Sink 2.70 302 9.1 1.04 6.0 88.8 1.9 1.9 15.1
HLS Float 2.70 809 24.5 0.07 0.4 98.3 0.3 0.3 44.8
Coarse Feed (Calc.) 3299 100 4.95 28.5 53.8 100 100 100
Combined Sinks Results
Combined HLS products Weight Assay, % Distribution, %
HLS Sink 3.00 1831 55.5 8.22 47.3 24.9 92.1 92.1 25.7
HL S Sink 2.90 2034 61.7 7.69 44.2 29.0 95.8 95.8 33.3
HLS Sink 2.80 2189 66.4 7.30 42.0 32.5 97.7 97.7 40.1
HLS Sink 2.70 2491 75.5 6.54 37.6 39.3 99.7 99.7 55.2
Table 4. HLS test results of medium size fraction (−0.5 + 0.15 mm).
Stage recovery-medium (−0.5 mm + 0.15 mm)
HLS products Weight Assay, % Distribution, %
HLS Sink 3.00 523 53.7 9.39 54.0 14.4 96.0 96.0 15.3
HLS Sink 2.90 31 3.1 2.41 13.9 53 1.4 1.4 3.3
HLS Sink 2.80 34 35 1.9 10.9 76 1.3 1.3 5.2
HLS Sink 2.70 52 5.3 0.9 5.2 89.3 0.9 0.9 9.4
HLS Float 2.70 335 34.3 0.06 0.3 98.2 0.4 0.4 66.7
Medium Feed (Calc.) 975 100 5.25 30.2 50.5 100 100 100
Combined Sinks Results
Combined HLS products Weight Assay, % Distribution, %
HLS Sink 3.00 523 53.7 9.39 54.0 14.4 96.0 96.0 15.3
HLS Sink 2.90 554 56.8 9.00 51.8 16.5 97.4 97.4 18.6
HLS Sink 2.80 588 60.3 8.59 49.4 20.0 98.7 98.7 23.8
HLS Sink 2.70 640 65.7 7.97 45.8 25.6 99.6 99.6 33.3
The dense media used in this project were a mixture of ferrosilicon (SG of 6.70) and
magnetite (SG of 5.00) in water. The proportions of ferrosilicon and magnetite used
were based on the desired media SG, while maintaining medium stability and viscosity
within acceptable ranges. In this DMS pilot plant campaign, the media were prepared
with 70% of ferrosilicon and 30% of magnetite. The media recovery system of the
DMS plant is designed to recover media for re-use in the system, and to maintain the
media density at the set-point. Tracer tests were conducted prior to, and subsequent to,
each operating shift to ensure that the cut-point was at the desired target.
Table 5. Combined HLS results of Bastnaesite Comp.

Products Weight Assay, % Distribution, %
% Nd2O3 TREO, Calc SiO2 Nd2O3 TREO SiO2
Coarse (−2.0 + 0.5 mm) 60.0 4.95 28.5 53.8 55.7 55.7 67.6
Medium (−0.5 + 0.15 mm) 17.7 5.25 30.2 50.5 17.5 17.5 18.8
Fine (−0.15 mm) 22.2 6.44 37.0 29.3 26.8 26.8 13.6
Feed (Calc.) 100.0 5.34 30.7 47.8 100.0 100.0 100.0
Feed (Direct) 5.54 31.9 47.7
Combined HLS products Weight Assay, % Global Recovery, %
% Nd2O3 TREO, Calc SiO2 Nd2O3 TREO SiO2
Coarse HLS 2.90 Sink (Calc.) 48.1 7.69 44.2 29.0 53.4 53.4 22.5
Medium (−0.5 + 0.15 mm) 23.0 5.25 30.2 50.5 17.5 17.5 18.8
Fine (−0.15 mm) 28.9 6.44 37.0 29.3 26.8 26.8 13.6
Roasting Feed Calc. 100.0 6.77 38.9 34.1 97.6 97.6 54.9
Dense Media Recovery

Cyclone
DMS Floats DMS Floats
Fig. 2. DMS Pilot Plant set-up.
DMS Test Results

Photos of sink and float products from the DMS 1st pass are shown in Fig. 3. The
overall results of the DMS pilot plant operation are summarized in Table 6 and
compared with the corresponding HLS results in Fig. 4. The DMS 1st pass sink
at SG 2.86 produced a high-grade concentrate (~44.6% TREO) with a high recov-
ery ~90.4%. The DMS results were similar to HLS results at SG 3.00. A second
DMS pass at a lower SG (2.81) was performed to improve the recovery.
Approximately 58% REO stage recovery was achieved in the second pass.
The combined DMS 1st pass and 2nd pass sinks produced a DMS concentrate
grading ~41% TREO at a recovery of ~96%. The combined DMS results at SG
2.81 were very similar to HLS results obtained at SG 2.90. The HLS concentrate
achieved a slightly higher grade compared to DMS concentrate, likely because
HLS is an ideal separation with minimum entrainment; however, DMS is a con-
tinuous operation with dynamic processing. Near-density material may report to
either the sink or floats, thus affecting the TREO grade and recovery. It is unclear
why the specific gravity used in DMS was about 0.09–0.14 lower compared to
HLS tests with similar results.
DMS Sinks DMS Floats
Fig. 3. Images of PP-01 DMS sinks and floats.
3.3 Wilfley Table Testwork
A representative subsample weighing 21 kg of the medium size fraction (−0.5/+0.15 mm)

of Bastnaesite Comp was screened at 0.25 mm. Each of the two size fractions
(−0.5/+0.25 mm, −0.25/+0.15 mm) was processed over the Wilfley Table. Figures 5
and 6 show the pictures of the operation and products of the Wilfley Table test. Good
separation could be expected from the clear color distinction for the concentrate, mid-
dlings, and tailings.
Table 7 presents the Wilfley Table test results. Both size fractions performed well
with the Wilfley Table. A very high-grade concentrate was achieved, ~62–64% TREO,
with 10–21% REO recovery. The combined Wilfley Table concentrate and middlings
produced a concentrate with good TREO grade (~40% TREO for the −0.5/+0.25 mm
fraction, ~52% TREO for the −0.25/+0.15 mm fraction) and good TREO recovery
(92% for the −0.5/+0.25 mm fraction, 85% TREO for −0.25/+0.15 mm fraction). The
mathematically combined products for the overall −0.5/+0.15 mm fraction provided a
45.5% TREO grade with 90.2% REO recovery.
Table 6. DMS pilot plant results.

Products Weight Assays % Distribution %
kg % Nd2O3 TREO, Cal SiO2 Nd2O3 TREO SiO2
DMS 1st Pass Sink 95.5 56.5 8.10 46.6 25.7 90.4 90.4 27.6
DMS 1st Pass Float 73.6 43.5 1.12 6.44 87.6 9.6 9.6 72.4
DMS Feed (calc.) 169.0 100 5.06 29.1 52.6 100 100 100
DMS Feed (Dir.) 4.95 28.5 53.8
DMS 1st Pass PP-02. SG at 2.81
DMS 2nd Pass Sink 13.1 18.8 3.39 19.5 64.5 57.9 57.9 13.9
DMS 2nd Pass Float 56.6 81.2 0.57 3.28 92.6 42.1 42.1 86.1
DMS 1st Pass Float (calc.) 69.6 100 1.10 6.32 87.3 100 100 100
DMS 1st Pass Float (Dir.) 1.12 6.44 87.6
Combined products
DMS 1st Pass Sink 95.5 56.5 8.1 46.6 25.7 90.4 90.4 27.6
DMS 1st & 2nd Pass Sink 108.5 64.6 7.51 43.2 30.6 95.9 95.9 37.6
DMS 2nd Pass Float 56.6 35.4 0.58 3.34 92.9 4.1 4.1 62.4
DMS Feed (calc.) 165.1 100 5.06 29.1 52.6 100 100 100
DMS Feed (Dir.) 4.95 28.5 53.8
Fig. 4. DMS results compared to HLS results.

Fig. 5. Photo of Wilfley Table test.
Concentrate Middling
Tailings
Fig. 6. Photos of −0.5 + 0.25 mm size fraction Wilfley Table products.
It is possible that the recovery of the −0.25/+0.15 mm fraction might be improved

with a scavenger stage. Some reddish-brown particles (suspected to be rare earth
minerals) were visually observed in the tailings. This needs to be further explored in
future testing.
Table 7. Wilfley Table test results.

Mid1 (−500 + 250 um) by Wifley Table
g % Nd2O3 TREO, Cal SiO2 Nd2O3 TREO SiO2
G-1 WT Conc 1618 10.3 10.7 61.5 7.77 21.2 21.2 1.6
G-1 WT Middlings 8313 52.8 6.96 40.0 34.9 70.8 70.8 36.0
G-1 WT Tails 5810 36.9 1.12 6.44 86.6 8.0 8.0 62.4
Head (calc.) 15,741 100 5.19 29.8 51.2 100 100 100
Mid2 (−250 + 150 um) by Wifley Table
G-2 WT Conc 265 5.0 11.1 63.8 5.59 10.5 10.5 0.6
G-2 WT Middlings 2317 44.1 8.99 51.7 17.2 74.2 74.2 15.6
G-2 WT Tails 2677 50.9 1.60 9.20 79.7 15.3 15.3 83.8
Head (calc.) 5259 100 5.33 30.7 48.4 100 100 100
Combined products Weight Assays % Distribution %
G1 & G2 WT Conc 1883 9.0 10.8 61.8 7.46 18.5 18.5 1.3
G1 & G2 WT Con & Middlings 12,513 59.6 7.91 45.5 15.6 90.2 90.2 32.4
G1 & G2 WT Tails 8.487 40.4 1.27 7.31 84.4 9.8 9.8 67.6
3.4 Combined DMS and Wilfley Table Results
The combined test results from the DMS and the Wilfley Table processing are presented
in Table 8. It is expected that the grade of combined products (combined DMS concen-
trates, Wilfley Table concentrate and middlings, and fines) would be ~41.5% TREO
with 96% TREO overall recovery.
4 Hydrometallurgical Processing
4.1 Flowsheet Development
Previous hydrometallurgical testing on samples of the Nechalacho deposit tended to

focus on the highly refractory zones of the deposit in which aggressive cracking meth-
ods, such as acid baking and alkaline cracking followed by acid leaching, were required
[2–4]. These methods were able to achieve high REE recoveries into pregnant leach
solution (PLS) but only with complex flowsheets.
With the REE in the North T Zone concentrates being in the mineral bastnaesite, less
aggressive extraction methods were tested alongside some of the more aggressive
extraction methods. A relatively simple oxidative roast followed by agitated acid leach-
ing was able to achieve the same levels of REE extraction, or better, than acid baking or
caustic cracking followed by acid leaching.
Table 8. Combined DMS and Wilfley Table test results.

Feed products Weight Assays % Distribution %
% Nd2O3 TREO, SiO2 Nd2O3 TREO SiO2
Cal
−2.0 + 0.5 mm fraction- DMS Feed 60.6 4.95 28.5 53.8 56.3 56.3 66.3
−0-5 + 0.15 mm fraction-WT Feed, Calc 17.4 5.25 30.2 50.5 17.1 17.1 17.8
−0.15 mm fraction, Calc 22.0 6.44 37.0 35.4 26.6 26.6 15.9
Feed (Calc.) 100 5.33 30.7 49.2 100 100 100
Feed (Dir.) 5.54 31.9 48.0
Stage Recovery with DMS & WT Calc on −2.0 + 0.15 mm fraction
Cal
−2.0 + 0.5 mm fraction- DMS 1st & 2nd 50.8 7.51 43.2 30.6 73.9 73.9 31.1
Pass Sink
−2.0 + 0.5 mm fraction Tailings (DMS 26.5 0.58 3.34 92.9 3.0 3.0 49.2
2nd Pass Float)
−0.5 + 0.15 mm fraction- WT Con & 13.6 7.91 45.5 15.6 20.8 20.8 4.2
Midd, Calc
−0.5 + 0.15 mm fraction-Tailings (WT 9.2 1.27 7.31 84.4 2.3 2.3 15.5
Tails), Calc
−2.0 + 0.15 mm fraction Concentrate 64.3 7.60 43.7 27.4 94.8 94.8 35.3
−2.0 + 0.15 mm fraction Tails 35.7 0.76 4.36 90.7 5.2 5.2 64.7
−2.0 + 0.15 mm fraction Feed (Calc) 100 5.16 29.7 50.0 100 100 100
Global Recovery on −2.0 mm fraction
Combined Products Weight Assays % Distribution %
Cal
−2.0 + 0.15 mm fraction Concentrate 49.0 7.60 43.7 27.4 69.6 69.6 29.7
−0.15 mm fraction 22.0 6.44 37.0 35.4 26.6 26.6 15.9
Combined Con + Fines 71.1 7.21 41.5 31.5 96.1 96.1 45.6
Tailings (DMS 2nd Pass Float, WT Tails) 27.2 0.76 4.36 90.7 3.9 3.9 54.4
The roast-leach and acid bake-water leach methods initially appears similar in that
both utilize a thermal treatment stage (such as rotary kiln) followed by an agitated leach
stage to extract the REE. Critical differences between the two methods are that the acid
bake method requires the beneficiation concentrate to be mixed with concentrated sul-
phuric acid, with the resulting mixture typically having a paste-like consistency. This
mixture is highly corrosive and abrasive and has poor flow characteristics which leads
to difficult thermal processing [5]. The feed to the thermal treatment component of the
roast-leach method is the product produced during physical beneficiation without any
chemical addition. In addition to not attacking roasting equipment, the beneficiation
concentrate does not have any of the physical handling issues encountered dealing with
an acid bake paste.
Caustic cracking of rare earth concentrates in atmospheric reactors does remove the
requirement of a thermal processing step and the resulting physical handling issues. A
downfall of the caustic crack method is that it uses sodium hydroxide in addition to
hydrochloric acid which adds additional reagent costs and complexity in the require-
ment to manage two liquor streams: the acidic REE-bearing PLS as well as the alkaline
caustic crack liquor.
Therefore, quite early on, the roast-leach method was selected as the extraction pro-
cess to develop for the North T zone concentrates. Initial testing evaluated roasting
followed by either sulphuric acid or hydrochloric acid leach flowsheets. Both acid types
were able to achieve acceptable REE extractions from the roasted product, with the
hydrochloric acid achieving slightly higher extractions (91% Nd vs. 85% Nd). Sulphuric
acid is typically significantly cheaper than hydrochloric acid and requires less expen-
sive materials of construction, which made that acid initially look more promising than
the hydrochloric route. However, during development of both flowsheets it was deter-
mined that the thorium content of a mixed REE product produced by double sulphate
precipitation and then metathesis was too high (292 g/t Th) and thus an additional re-
leach and impurity removal step would be required prior to precipitation of a suitable
mixed REE product. The hydrochloric acid liquor treatment was relatively straight for-
ward consisting of simple impurity precipitation and then precipitation of mixed REE
product, and was therefore selected as the route forward. This flowsheet is discussed in
detail in the next section.
4.2 North T Flowsheet
The hydrometallurgical flowsheet developed for the North T deposit consisted of sub-
jecting the beneficiation concentrate to roasting without any further particle size reduc-
tion, with the roasting feed having a K80 particle size distribution (PSD) of 1.2 mm.
Bench-scale testing determined optimum roasting conditions of 650 °C and 1-hour
retention time. The roasted product was then subjected to a hydrochloric acid leach fol-
lowing similar conditions used commercially to process bastnaesite in the historic
Mountain Pass operations [6]. For the North T samples, a final acidity of 20 g/L HCl
was able to achieve good levels of extraction of the desirable REE elements, such as
neodymium and praseodymium but with significant rejection of the un-desirable
cerium. The rejection of cerium occurs due the oxidization of cerium III to the insoluble
cerium IV during the roasting stage. The acid leach also rejects the majority of the tho-
rium and fluorine to the leach residue. Acid leaching was conducted without external
heat input, but the reaction of hydrochloric acid and sample was moderately
exothermic.
The HCl PLS contained 20 g/L HCl along with 70 g/L REE (predominantly 21.6 g/L
Nd, 27.8 g/L La and 10.9 g/L Ce) as well as impurity elements consisting of 1.7 g/L Fe,
0.2 g/L Al, 0.4 g/L F. The PLS contains very low levels of uranium (3.5 mg/L U) and
thorium (0.06 mg/L Th). Magnesium carbonate slurry is added to the PLS in two stages,
first stage is conducted around pH 2.8 and the second stage is conducted around pH 4.5.
The two-stage approach removes the iron and some of the aluminum in the first stage
while REE co-precipitation is kept as low as possible as this precipitate is sent to tail-
ings. The first stage liquor still contains some aluminum and fluoride which is precipi-
tated at the higher pH of the second stage, higher levels of REE co-precipitation occurs,
thus this second stage precipitate is sent upstream to be re-leached. The acidity of the
HCl PLS was determined to be sufficient to fully re-dissolve the REE.
Sodium carbonate is then added to the pH 4.5 liquor to precipitate a mixed REE
carbonate. This intermediate product is then sent to an off-take partner where it is
refined and separated into individual products [7]. The REE-depleted liquor is then
subjected to a radium removal stage prior to disposal. The hydrometallurgical flowsheet
is shown in Fig. 7.
Fig. 7. Hydromet flowsheet.

4.3 Results of Pilot Plant Operation
After developing the flowsheet at bench scale, the process was subjected to pilot plant
testing that consisted of continuous roasting, bulk batch leaching, continuous primary
impurity removal (Impurity Removal 1), bulk batch secondary impurity removal
(Impurity Removal 2), and continuous REE precipitation. Key parameters of each unit
operation plus the radium removal process are summarized in Table 9 and described in
detail in this section.
Pilot roasting was conducted in an electrically heated 165 mm rotary kiln (Fig. 8)
operated at 650 °C and a target retention time of 1 hour in the hot zone. The roasting
operations proceeded smoothly through the kiln with no material build-up issues
observed during the 72 hours of continuous operation, a photograph of kiln discharge is
shown in Fig. 9. XRD analysis of kiln discharge samples confirmed that the roasting
operations converted the REE fluoride carbonate mineral bastnaesite to REE oxides,
fluorite, and lanthanum fluoride.
Table 9. Roast-hydromet unit operations summary.

Impurity Impurity REE Radium
Stage Roast Leach removal 1 removal 2 precipitation removal
Temperature, °C 650 20–50 70 70 70 20
Reagent Heat, air HCl MgCO3 MgCO3 Na2CO3 IX Resin
Control – 20 g/L pH 2.5–3.0 pH 4–5 <100% stoich –
Parameter(s) HCl
Fig. 8. Pilot rotary kiln.

Fig. 9. Kiln discharge (kiln end plate removed to take picture).
The kiln discharge was leached with hydrochloric acid solutions in 50–100 L batches
to produce feed for continuous impurity removal and then REE precipitation circuits.
The batch leaching was able to achieve 95% neodymium extractions from the roasted
product, with cerium co-leaching ranging from 15% to 51%, with higher acidities (37 g/L
HCl) resulting in the higher cerium extractions than the lower acidities (~20 g/L HCl).
Continuous Impurity Removal 1 mini-pilot plant operations (Fig. 10, left) reduced
iron and thorium levels from 1650 mg/L Fe and 0.06 mg/L Th in the PLS to <2 mg/L Fe
and <0.03 mg/L Th in the discharge liquor with less than 2% co-precipitation of REE
with an average pulp pH of 2.6. Aluminum precipitation was 54% in Impurity Removal
1 but the Impurity Removal 2 stage reduced the aluminum in the REE bearing liquor to
<4 mg/L Al.
The impurity removal filtrate was then subjected to REE carbonate precipitation in a
continuous mini-pilot plant (Fig. 10, right). The purity of the mixed REE carbonate
precipitate was determined to be 98.2% by assuming impurity elements are present as
simple carbonates as well as factoring sulphur and fluorine assay. Detailed analysis is
provided in Table 10.
Radionuclide analysis of the post REE precipitation liquor determined that a radium
removal process was required because the sample analysis exceeded the unconditional
derived release limits for diffuse NORM sources. Samples of the post REE precipita-
tion liquor were treated in 25 mm column testing with Resin Tech RSM-50-HP (RSM)
ion exchange (IX) resin. The RSM resin is a sodium form microporous highly cross-
linked hybrid strong acid cation resin with a barium sulphate functional group which
Fig. 10. Continuous mini-pilot plants (Left: impurity Removal 1; Right: REE precipitation).
Table 10. REE precipitate analysis.

REO% REO(-Ce)% La2O3% CeO2% Pr6O11% Nd2O3% Other REO% LOI%
66.0 55.5 26.7 10.4 5.9 20.1 2.9 31.6
Th a/t U g/t CaO% Na2O% Other Oxide % F% S% Sum%
<0.5 1.9 0.5 0.1 0.1 0.02 0.03 98.3
exchanges radium and barium ions [8]. Testing reduced the combined radium 226 and
radium 228 content from 47 Bq/L to <0.05 Bq/L and <0.7 Bq/L, respectively.
Overall REE deportment excluding cerium to the REE precipitate in the hydrometal-
lurgical flowsheet was 89% when assuming full re-leaching of the Impurity Removal 2
precipitate is achieved (this was demonstrated to be the case during bench-scale test-
ing). Deportment of key analytes is given in Table 11.
5 Conclusion
The following conclusions may be drawn from the testwork performed on the sample
from the North T Zone of the Nechalacho deposit:
• The Bastnaesite Comp head sample contained a high concentration of light REE,
with a TREO grade of 31.9%. The main rare earth minerals contained in the sample
were parisite and bastnaesite, and the major gangue mineral was quartz.
Table 11. Percentage deportment in hydrometallurgical flowsheet.

Leach Impurity removal Spent REE REE
Steam residue precipitate liquor precipitate Sum
REE 28 1 3 69 100
REE (Ce 6 2 4 89 100
excl)
Ce 65 1 1 33 100
Nd 5 2 3 90 100
Th 100 0 0 0 100
U 30 60 1 9 100
Al 90 10 0 0 100
Fe 52 48 0 0 100
Mg 13 7 79 1 100
Ca 26 1 71 2 100
F 96 3 0 0 100
• The HLS results showed very good potential to separate rare earth minerals from
gangue by gravity separation. For the coarse size fraction (−2.0/+0.5 mm), the grade
of the SG 2.90 sink product was ~44% TREO with a recovery of ~96% and a mass
yield of ~62%; for the medium size fraction (−0.5/+0.15 mm), the grade of the SG
2.90 sink product was ~52% TREO with a recovery of ~97% and a mass yield of
~57%. The HLS results of the medium size fraction were slightly better than those
of the coarse size fraction, likely due to improved mineral liberation at the finer par-
ticle size.
• A DMS pilot plant campaign successfully processed 170 kg of the coarse size frac-
tion (−2.0/+0.5 mm) with two passes. The combined DMS 1st pass sink and 2nd
pass sink produced a ~43% TREO concentrate with 96% stage recovery.
• A Wilfley Table tested 21 kg of the medium size fraction (−0.5/+0.15 mm), pro-
duced a 45.5% TREO concentrate with 90.2% stage recovery.
• It is expected that a 42% TREO concentrate with 96% overall recovery can be pro-
duced using DMS and Wilfley Table.
• A roast-leach REE extraction flowsheet followed by impurity removal by precipita-
tion and then mixed REE carbonate precipitation achieved a product suitable for
separation with a purity of 98.2% with a neodymium recovery of 90% (roast feed to
REE carbonate).
References
1. Vital Metals Webiste, Resources-Vital Metals. Retrieved from https://vitalmetals.com.au/

resources/. Last accessed 27 Mar 2022.
2. Notzl H, Khan S, Verbaan V. Hydrometallurgical plant design parameters for the Avalon rare earth
process. Mining, metallurgy, and petroleum. In: Conference of metallurgists 2013 – rare earth ele-
ments. Canadian Institute; 2013.
3. Bradley K, Verbaan V, Notzl H, Marsh D, Goode J. The caustic crack route for processing con-
centrate from the Avalon Nechalacho deposit. In: Conference of metallurgists 2014 – rare earth
elements. Canadian Institute of Mining, Metallurgy, and Petroleum; 2014.
4. Avalon Rare Metals Inc. News Release – December 12, 2013: Avalon’s Nechalacho Project.
Metallurgical process optimization yielding improved REE recoveries. Retrieved from http://ava-
lonadvancedmaterials.com/_resources/news/2013/NR_13_15.pdf. 2013.
5. Verbaan V, Bradley K, Brown J, Mackie S. A review of hydrometallurgical flowsheets considered
in current REE Projects. In: Symposium on strategic and critical materials proceedings. Victoria,
British Columbia: British Columbia Ministry of Energy and Mines, British Columbia Geological
Survey Paper; 2015. 13–14 Nov 2015.
6. Duncan L. US Patent 3,812,233 process for separating cerium concentrate from ores. 1974.
7. Vital Metals. News Release – October 18, 2021: REETEC INCREASES OFFTAKE OF VITAL’S
REO BY 50%. Retrieved from https://wcsecure.weblink.com.au/pdf/VML/02436809.pdf. 2021.
8. ResinTech RSM-50-HP PRODUCT SPECIFICATION SHEET. Retrieved from https://1l8ojfeogc
732vtc03kye2i1-wpengine.netdna-ssl.com/wp-content/uploads/2021/06/RSM-50-HP-spec-1.0.pdf
The Beneficiation and Hydrometallurgical Process
Development for the Lofdal HREE Project
Jing Liu1 (*), Niels Verbaan1, Micheal Archer1, and Tassos Grammatikopoulos1,
Barbara Mulcahy2, and Rainer Ellmies2
1
Jing.Liu2@sgs.com
2
Namibia Critical Metals Inc, Halifax, NS, Canada
Abstract. The Lofdal REE project is one of the few rare earth deposits in the world
that contains mostly heavy rare earth elements (~75% HREO distribution). The min-
eralogy of the Lofdal project is complex, with xenotime as the primary rare earth
mineral, and silicate, calcite, and iron oxide as the main gangue minerals. The benefi-
ciation and hydrometallurgical flowsheets developed for the Lofdal project in previ-
ous studies were complicated and sub-economic. The recent development of the
beneficiation testwork largely simplified the flowsheet with improved concentrate
grade and recovery. The beneficiation process included ore sorting, magnetic separa-
tion, and flotation. The enhanced selectivity of xenotime against high reagent (acid,
neutralizing reagent) consuming gangue minerals (calcite and hematite) was benefi-
cial for the CAPEX and OPEX cost saving of the downstream hydrometallurgical
process. The original caustic crack-based flowsheet was converted into an acid bake-
based flowsheet treating the flotation concentrate capable of recoveries exceeding
95% of key HREE. Following several conventional liquor purification steps, a high-
purity rare earth oxide concentrate at >98% total rare earth oxide with low impurity
levels was produced. This paper highlights the findings from laboratory testing pro-
grammes conducted to develop the process.
Keywords: Heavy rare earth elements · Xenotime · Flotation · Acid bake ·

Leaching · Precipitation
https://doi.org/10.1007/978-3-031-17425-4_83
700 J. Liu et al.
Table 1. Head assays and REO distributions of the XRT SP.

Assays, TREO distribution Assays
Element %, g/t % Element %
TREO, % 0.60 100 SiO2% 40.3
Y2O3, % 0.37 61.4 Al2O3% 8.28
Dy2O3, g/t 442.1 7.3 Fe2O3% 8.79
La2O3, g/t 195.9 3.2 MgO% 0.84
Ce2O3, g/t 353.0 5.8 CaO% 19.3
Pr2O3, g/t 40.2 0.7 Na2O% 3.86
Nd2O3, g/t 137.2 2.3 K2O% 0.58
Sm2O3, g/t 66.0 1.1 TiO2% 0.67
Eu2O3, g/t 38.8 0.6 P2O5% 0.46
Gd2O3, g/t 232.2 3.8 MnO% 0.18
Tb2O3, g/t 57.2 0.9 Cr2O3% <0.01
Ho2O3, g/t 102.4 1.7 V2O5% 0.03
Er2O3, g/t 308.4 5.1 LOI% 16.7
Tm2O3, g/t 43.2 0.7 Sum% 100
Yb2O3, g/t 272.1 4.5
Lu2O3, g/t 37.6 0.6
Sc g/t <25
U g/t 17.2
Th g/t 626
1 Introduction
Namibia Critical Metals (Namibia) is developing the Lofdal xenotime-hosted heavy

rare earth deposit. In 2020, it engaged SGS Canada to re-develop the flowsheet for the
mineral beneficiation testwork and to produce a mineral concentrate suitable for further
hydrometallurgical upgrading. This paper describes the process development testwork
conducted to develop a flowsheet capable of producing a purified/mixed rare earth pre-
cipitate suitable for REE separation processing.
About 300 kg of sorted product (XRT SP) from Namibia Critical Metals’ Lofdal
deposit was received and used for the current testwork.
2 Sample Characterization
A head sample was taken and submitted for head analysis and mineralogy analyses.
As shown in Table 1, the total rare earth oxide (TREO) contents in the head sample
of XRT SP was 0.6% TREO. The most abundant rare earth element (REE) was yttrium
(Y), assaying at 0.37% Y2O3, which accounted for ~60% of the TREO. The second
most abundant rare earth element is dysprosium (Dy), grading 442 g/t Dy2O3, which
accounted for ~7% of the TREO. Other rare earth elements with decent contents
The Beneficiation and Hydrometallurgical Process Development for the Lofdal HREE… 701
included heavy REE erbium (Er), gadolinium (Gd), ytterbium (Yb), light REE cerium
(Ce), lanthanum (La), and neodymium (Nd). The yttrium oxide (Y2O3) grade was
tracked in the metallurgical testwork to evaluate the test performance and a ratio of 1.68
was used for the XRT SP sample to estimate the TREO grade from the Y2O3 grade.
3 Head Mineralogy
The mineralogical work was conducted with QEMSCAN (Quantitative Evaluation of

Materials by Scanning Electron Microscopy), electron probe micro-analysis (EPMA),
and X-ray diffraction. The purpose of this test programme was to determine the overall
mineral assemblage of the samples and the occurrence of REE minerals.
3.1 X-Ray Diffraction (XRD) Analysis
XRD analysis of the samples XRT SP indicated that it consisted mainly of albite
(36.9%), calcite (29.9%), and quartz (15.6%), with lesser hematite (5.3%), chlorite
(4.1%), and trace amounts (<2%) of muscovite, microcline, magnetite, goethite, anker-
ite, xenotime, and dolomite.
3.2 Grain Size
Fe oxides are the coarsest mineral with a D50 (50% passing value by mass from the
cumulative grain size distribution) of 40 μm in XRT SP, respectively. This is followed
by quartz/feldspars (38 μm), calcite (30 μm), F-C-REE (23 μm), biotite/chlorite (9 μm),
xenotime (19 μm), ankerite (16 μm), and thorite/Th-Y-silicates (17 μm). The D50 of the
Particle is 35 μm. The term “Particle” refers to both liberated and middling particles,
monomineralic and polymineralic.
3.3 Liberation and Association of Xenotime
Rare earth minerals (REM) (collectively accounting for 1.13%) included mainly xeno-
time (0.80%), as well as trace amounts (<0.2%) of synchysite/parisite, monazite, bast-
naesite, columbite/pyrochlore, and zircon (0.32%).
Liberation (free, ≥95% of the total particle area, and liberated, ≥80% of the total
particle area) for xenotime accounted for 37.6% in XRT SP, at a P80 of ~100 μm. The
remainder occurred mainly as complex particles (35.4%), and as binaries with zircon
(11.1%), quartz/feldspars (4.9%), and calcite (8.9%). Xenotime liberation increased
from <0.2% to 47% from the +106 μm to the −38 μm fraction.
The calculated exposure of xenotime for the head samples was 29% for a xenotime
exposure of >80% (Figs. 1, 2, and 3).
702 J. Liu et al.
Xenotime Association
100
90
80
70
60
Mass (% Xenotime)
50
40
30
20
10
0
XRT SP
Complex 35.4
Xnt:Other 0.91
Xnt: Oxides 0.90
Xnt: Carbonates 9.05
Xnt: Silicates 5.07
Xnt: Zircon 11.1
Lib Xenotime 14.9
Free Xenotime 22.7
Fig. 1. Summary of liberation (mass%) of xenotime calculated for the XRT SP head sample.
Xenotime Exposure
100
90
80
70
Mass (% Xenotime)
60
50
40
30
20
10
0
XRT SP
<20% Exposed 22.3
<80>20% Exposed 48.4
>80% Exposed 29.3
Fig. 2. Summary of exposure (mass%) of xenotime for the XRT SP head sample.
Mineral Liberation
100
90
80
70
Mass (% Mineral)
60
50
40
30
20
10
0
Combined +106um -106/+38um -38um
Xenotime 37.6 0.17 28.8 46.8
Thorite/Th-Y-Silicates 37.5 0.00 29.1 44.5
F-C-REE 36.8 23.0 3.19 55.5
Ankerite 4.0 0.1 0.9 7.3
Fe-Oxides 83.7 69.6 87.7 87.2
Fig. 3. Summary of liberation (mass%) of xenotime, thorite/Th-Y-Silicates, F-C-REE, ankerite and

Fe oxides for the XRT SP head sample.
3.4 Mineral Chemistry
Electron probe micro-analyses (EPMA) were conducted on the various REE minerals
to quantify their chemical composition. The data show that xenotime contains yttrium
and significant HREE, while light rare earth minerals (LREMs which include bastnae-
site, synchysite, and parisite) consist mainly of varied proportions of LREE. The com-
position in average of rare earth minerals is presented in Table 2.
• Xenotime averages Y2O3 44.59%, Dy2O3 5.47%, Er2O3 3.76%, Yb2O3 3.08% and
Gd2O3 2.67%, and ThO2 0.35%, while other REEs are generally low (<1.5%) or
below the detection limit of the instrument.
• Y-Th-silicate grains average Y2O3 6.65%, and ThO2 54.0%, while REE concentra-
tions are generally low (<1%).
• Tentatively identified hingganite averages Y2O3 35.07%, Nd2O3 1.72%, Gd2O3 2.3%,
Dy2O3 4.13%, Er2O3 2.59%, Yb2O3 1.84%, while other REEs are below 1.5%. It is
noted that BeO is assumed in the analyses.
• Bastnaesite averages Ce2O3 31.86%, Nd2O3 10.85%, La2O3 21.22%, Pr2O3 3.19%,
Sm2O3 2.08%, Gd2O3 1.56%, Y2O3 0.82%, and ThO2 0.85%.
704 J. Liu et al.
Table 2. Mineral chemistry by electron probe micro-analysis.

Element/Oxide, % Xenotime Th-silicate Higganite Bastnaesite Parisite Synchysite
P2O5 35.68 4.58 0.11 – – –
V2O5 0.05 0.87 0.00 – – –
As2O5 0.06 0.15 0.01 – – –
SiO2 0.24 14.81 26.99 – – –
ZrO2 0.20 3.39 0.01 – – –
ThO2 0.35 54.00 0.08 0.85 0.59 0.88
Y2O3 44.59 6.65 35.07 0.82 1.19 2.47
La2O3 0.02 0.08 0.25 21.22 15.08 13.44
Ce2O3 0.02 0.33 1.25 31.86 27.08 21.96
Pr2O3 0.02 0.05 0.21 3.19 2.68 2.10
Nd2O3 0.09 0.24 1.72 10.85 9.94 6.67
Sm2O3 0.40 0.40 1.07 2.08 1.75 1.24
EU2O3 0.37 0.21 0.51 – – –
Gd2O3 2.67 0.92 2.30 1.56 1.37 1.51
Tb2O3 0.70 0.18 0.49 – – –
Dy2O3 5.47 1.30 4.13 0.29 0.30 0.49
Ho2O3 1.17 0.18 0.87 – – –
Er2O3 3.76 0.62 2.59 0.02 0.06 0.17
Tm2O3 0.50 0.04 0.26 – – –
Yb2O3 3.08 0.45 1.84 0.03 0.06 0.09
Lu2O3 0.42 0.09 0.18 – – –
FeO 0.22 1.52 2.67 0.45 0.92 0.71
CaO 0.12 2.69 1.73 0.32 9.11 16.32
F – – – 9.66 7.60 6.27
TRE-Y-Oxides
80.0
71.9
70.0 63.1
59.4
60.0
52.6
50.1
TRE-Y-Oxides
50.0
40.0
30.0
20.0
11.5
10.0
0.0
Xenotime Y-Th-Silicate Hingganite Bastnaesite Parisite Synchysite
TRE-Y-O 63.1 11.5 52.6 71.9 59.4 50.1
Fig. 4. Total REE + Y oxides (wt%) of each mineral analysed with the EPMA.
Ce, La, Nd, Sm, Pr, Dy, Yb, and Y Deportment - XRT SP
100
90
80
70
Mass (% Element)
60
50
40
30
20
10
0
Ce La Nd Sm Pr Dy Yb Y
Other REE 0.9 0.7 1.3 0.3 1.2 0.0 0.0 0.0
Columbite/Pyrochlore 0.0 0.0 0.2 0.6 0.1 0.1 0.0 0.0
Monazite 6.0 4.7 6.1 1.7 6.7 0.0 0.0 0.0
Synchysite/Parisite 69.7 68.8 69.0 29.4 69.5 0.8 0.0 0.4
Bastnaesite 22.2 25.8 21.0 9.7 22.5 0.2 0.1 0.1
Th-Y-Silicate 1.1 0.0 2.4 9.1 0.0 4.2 2.7 2.7
Xenotime 0.0 0.0 0.0 49.2 0.0 94.7 97.2 96.8
Fig. 5. La, Ce, Pr, Nd, Sm, and Y deportment (mass%) calculated for the head samples.
• Parisite averages Ce2O3 27.08%, La2O3 15.08%, Nd2O3 9.94%, Pr2O3 2.68%, Sm2O3
1.75%, Gd2O3 1.37%, Y2O3 1.196%, and ThO2 0.59%.
• Synchysite averages Ce2O3 21.96%, La2O3 13.44%, Nd2O3 6.67%, Pr2O3 2.1%,
Sm2O3 1.24%, Gd2O3 1.51%, Y2O3 2.47%, and ThO2 0.88%. Other REEs are present
in traces or below the detection limit of the instrument in all of the fluorcarbonates.
Xenotime hosts a total of REE+Y 63.1%, Y-Th-silicates (11.5%), hingganite
(52.6%), bastnaesite (71.9%), parisite (59.4%), and synchysite (50.1%) (Fig. 4).
The deportment of selected REE and yttrium is shown in Fig. 5. The deportment is
based on the mineral mass as determined by the QEMSCAN and average values of the
REE from the EPM analyses, SEM, and theoretical compositions for minerals that were
not analysed.
The LREE (Ce, La, Nd, and most of the Pr) are hosted by F-C-REE (bastnaesite and
synchysite/parisite), whereas significant samarium, and most of dysprosium, ytterbium,
and yttrium are hosted mainly by xenotime.
706 J. Liu et al.
4 Mineral Processing Testwork
Flotation testing on the Lofdal Heavy Rare Earth Project was investigated by other
laboratories (Mintek, Nagrom, and KYSPYmet) in the past. The most promising flow-
sheet was developed by KYSPYmet, using a magnetic concentrate, which achieved an
REO concentrate grading 5% Y2O3 (~8% TREO) from a feed grade of 0.7% Y2O3, with
a REO upgrade of ~7 times and a REO stage recovery of ~50%. The flowsheet was
complex, including one sulphide circuit, carbonate flotation circuit (two cleanings), and
finally xenotime flotation (two cleanings).
The main objective of the current flotation testing programme was to simplify and
improve the flotation flowsheet developed by KYSPYmet in 2016, with similar or bet-
ter flotation performance. Interests have primarily focused on further reducing the cal-
cium and iron contents of the flotation concentrate advancing to subsequent
hydrometallurgical processing, as high calcium or iron gangues would contribute to
complex downstream flowsheet and costly processing.
After systematic investigation of different reagent schemes (collector, depressant)
and operation parameters (pH, pulp density, pulp temperature, grind size, etc.), the
flotation flowsheet was successfully simplified from the previous three flotation stages
to consist of only REE flotation with reasonably good yttrium grade (5.5% Y2O3, or
Table 3. Summary of collector type and dosages in T1 to T5.

Test ID Collector Type Dosage, g/t
T1 SM15 Phosphate ester 700
T2 3434 Modified hydroxamate 2300
AZ FA-1 Fatty acid 30
3000 Modified hydroxamate 100
3000 Modified hydroxamate 100
T5 Sylfat FA-2 Fatty acid 700
Fig. 6. Yttrium grade versus recovery and yttrium versus calcium recoveries of tests T1 to T5.
~9% TREO) and recovery (65%) at a 4% mass yield on an XRT sorting product (test
T11). Selected testing results and findings are summarized in this paper.
4.1 Batch Flotation Testwork
Batch flotation tests were carried out on 2 kg charges of the XRT SP sample to develop
a preliminary flowsheet for REE beneficiation.
Each test was carried out in a Denver D12 laboratory flotation machine, and three
cell sizes (8 L, 4 L, 2 L) were used in this programme. Tests T1 to T5 were performed
on rougher stages only, and tests T6 to T11 included two to three cleaner stages.
Collector Screening The initial collector screening tests were investigated in tests T1
to T5. Collector types and dosages are summarized in Table 3 and the following condi-
tions were used:
• Pulp density: ~20%
• Pulp temperature: 50 °C.
Table 4. Summary of depressant type and dosages in T6 to T9.

Test ID Depressant Dosage, g/t
T6 Na2SiO3 1000
Na2CO3 3000
F220 2400
Na2SiF6 500
T7 Na2SiO3 300
CO2 3748
T8 Na2SiO3 300
Calgon 500
T9 Na2SiO3 300
Calgon 250
Fig. 7. Yttrium grade versus recovery and yttrium versus calcium recoveries of tests T6 to T9.
708 J. Liu et al.
Figure 6 depicts the key results. T1 and T2 were baseline tests based on the most
promising collectors, and testing conditions were revealed from the previous testwork
completed on Lofdal ore (Nagrom, Mintek, KYSPYmet). T1 tested 700 g/t SM15, T2
tested 2300 g/t hydroxamate collector (Florrea 3434), and 30 g/t of fatty acid blends
(AZ FA-1). Test T3 evaluated collector 3434 at a higher dosage (4000 g/t) combined
with collector 3000 (100 g/t). Test T4 investigated a modified hydroxamate collector
(3900) and collector 3000. Test T5 evaluated another fatty acid collector Sylfat FA-2.
The main findings from the collector screening tests can be summarized as follows:
• Both collectors SM15 and 3434/FA-1 (T1, T2) achieved ~91% Y recovery at a 1.7
times upgrade and a mass yield of 52–55%, which is reasonably good for a rougher
test. The main impurity in the rougher concentrates was calcium oxide, which is
likely attributable to calcite based on the head mineralogy. Iron oxide was later
revealed to be another main impurity in the flotation concentrate; however, it was not
the focus at the early stage of testing due to the abundant amount of calcium.
• The hydroxamate collectors (3434, 3900, 3000) demonstrated better overall yttrium
flotation performance and selectivity against calcium compared to fatty acid collec-
tor (FA-2) or phosphoric acid ester collector (SM15). Collector 3900/3000 (T4)
showed the best of yttrium grade versus recovery and best selectivity against calcium.
Thus, collectors 3900 and 3000 have been used as the collector scheme for the sub-
sequent tests.
Table 5. Flotation test results of T10 and T11: iron post-removal versus iron pre-removal.
Assay, % Distribution, %
Weight TREO,
Test No. Products % Y2O3 est CaO SiO2 Fe2O3 Y2O3 CaO SiO2 Fe2O3
T10 2nd Cl Con 5.6 3.82 6.42 11.1 26.7 29.3 61.6 3.1 3.8 18.8
NonMags
2nd Cl Con 2.5 0.55 0.92 0.4 3.2 93.6 4.0 0.0 0.2 26.8
Mags @2A
2nd Cl Con 1-2 8.1 2.81 4.72 7.8 19.5 49.1 65.6 3.2 4.0 45.5
Iron 1st Cl Con 1-3 14.8 1.97 3.30 12.5 24.8 35.7 83.6 9.4 9.3 60.3
Post- 1st Cl Con 1-3 16.2 1.84 3.09 13.3 25.6 33.7 85.2 10.8 10.5 62.2
removal & Scav Con
Ro Conc 1-5 34.0 0.94 1.58 21.5 27.3 19.2 91.9 36.7 23.6 74.6
Ro Tails 63.2 0.04 0.07 19.0 45.8 3.3 7.3 60.5 73.5 23.6
Head (Calc.) 100.0 0.35 0.59 19.9 39.4 8.8 100 100 100 100
T11 2nd Cl Con 1-2 4.2 5.50 9.25 9.1 19.5 32.9 65.1 1.9 2.1 15.8
1st Cl Con 1-3 6.2 3.99 6.71 14.0 20.4 28.2 69.8 4.4 3.2 20.0
Iron Ro Conc 1-5 15.5 1.74 2.92 21.0 25.8 16.5 76.1 16.7 10.1 29.3
Pre- Ro Conc 1-6 16.6 1.66 2.79 21.1 26.0 16.3 77.7 17.9 10.9 30.9
removal Ro Tails 80.3 0.09 0.15 20.0 43.7 4.2 20.4 82.0 88.5 38.3
WHIMS 3.0 0.23 0.39 0.8 7.7 88.7 2.0 0.1 0.6 30.8
Mags @2A
Head (Calc.) 100.0 0.36 0.60 19.6 39.7 8.7 100 100 100 100
Depressant Screening Several depressant types and varied dosages of promising

depressants were evaluated in tests T6 to T12, including sodium silicate, soda ash,
F-220, sodium fluorosilicate, carbon dioxide, and Calgon. The tested depressant types
and dosages are summarized in Table 4.
As shown in Fig. 7, Calgon produced superior results on improving the yttrium
selectivity against calcium. Some of the reagents considered to be good calcite depres-
sant (such as soda ash, F220, CO2) did not demonstrate effective calcium depression.
Dosages of Calgon at 250 g/t and 500 g/t were evaluated in tests T8 and T9, and 250 g/t
in T9 produced the best yttrium recovery and grade.
4.2 Iron Removal by Magnetic Separation
Although a few iron depressants were evaluated in the flotation testwork, such as
HX-800, Cyquest 3223, Dextrin, none of them exhibited attractive selectivity of yttrium
against iron. Magnetic separation to remove iron oxide was evaluated on the flotation
feed (pre-removal) and flotation concentrate (post-removal). The samples were pro-
cessed over Eriez L-20 unit at 2 Amps (~2000 Gauss) on the flotation feed or the 2nd
cleaner concentrate of tests T11, T10, respectively, and the WHIMS magnetic product
was re-passed over WHIMS for cleaning. The non-magnetic products from both passes
were combined and used as the flotation feed or the final flotation concentrate.
As shown in Table 5, WHIMS rejected ~31% of the iron from the flotation feed and
27% from the 2nd cleaner concentrate. WHIMS on the 2nd cleaner concentrate showed
a higher yttrium loss of ~4%, compared to the loss of 2% when WHIMS was performed
on the flotation feed. Additionally, the iron pre-removal can potentially minimize the
reagent consumption by iron oxide minerals. Thus, the iron pre-removal was deter-
mined to be applied for the subsequent testing.
Concentrate NaOH H2O HCl
HCl Gangue Leach Acid Leach REO Product to

Caustic Crack
(GL) (AL) Further Refining
H2SO4 dissolved gangue Ca(OH)2 Ca(OH)2

CaCl2
NaOH Treatment / Final

Acid Regeneration Th Precipitation REE Precipitation
Regen Neutralization (FN)
Ca(OH)2
Gypsum Ca silicates Th Precipitate Metal hydroxides
Fig. 8. Simplified conceptual block flow diagram – caustic crack-based flowsheet.

710 J. Liu et al.
Concentrate H2SO4 H2O HCl Reagent (MgCO3)
Acid Gangue Leach Water Leach Impurity Removal Leach residue to

Acid Bake
(GL) (WL) (IR) Tails Neut
dissolved gangue
CaCl2
Na2CO3 H2SO4 Reagent (soda ash or oxalic acid)
Crude REE U and Th removal

Re-Leach REE Precipitation
Precipitation (IX/SX)
Ca(OH)2
Final
Internal Recycle REO Product to
Neutralization Metal hydroxides
or Neutralization Further Refining
(FN)
Fig. 9. Simplified conceptual block flow diagram – acid bake-based flowsheet.
5 Hydrometallurgical Processing of Mineral Concentrate
A test programme was undertaken evaluating two hydrometallurgical flowsheet options

for the treatment of the mineral concentrate produced in the mineral processing test-
work. The objective of the testwork was to determine the potential for producing a
marketable rare earth product with minimal impurities suitable for further downstream
REE separation. The following flowsheet options were considered:
1. Caustic Crack Flowsheet. Much of the historic testwork provided for review by SGS
focused on the “Caustic Crack” route, which was based on gangue leaching of the
mineral concentrate, and subsequent caustic crack and acid leaching to remove
impurities from the heavy REE. Any REE dissolved in the caustic crack/acid leach
circuit are recovered via precipitation and recombined with the primary acid leach
residue. A simplified BFD of this flowsheet is shown in Fig. 8.
2. Acid Bake Flowsheet. The acid bake flowsheet is considered by Northern Minerals
on its Browns Range xenotime deposit and also applied by Lynas. A simplified BFD
of this flowsheet is shown in Fig. 9.
A detailed composition of the composite flotation concentrate is shown in Table 6.
Preliminary testwork was conducted using Conc 1 (a preliminary low-grade mineral
concentrate), but once flotation performance improved, a higher-grade Conc 2 (6.4%
TREE) sample was used.
5.1 Testwork Supporting Caustic Crack Test-Based Flowsheet
Low-grade (1.2% TREE) mineral concentrate sample was used for gangue leaching and
caustic cracking testwork. Due to the high calcium (as calcite) contents of the mineral
concentrate, gangue acid leach tests were carried out to investigate the removal of
gangue minerals through the addition of hydrochloric acid. Figure 10 shows the disso-
lution of gangue and pay elements as function of pulp pH. At pH 3, around 91% of the
Table 6. Flotation concentrate composition.
Sample ID La, g/t Ce, g/t Pr, g/t Nd, g/t Sm, g/t Eu, g/t Gd, g/t Tb, g/t Dy, g/t Ho, g/t Y, g/t Er, g/t Tm, g/t Yb, g/t Lu, g/t
Conc 1 (LG) 426 683 <256 257 <431 – – 145 1160 – 9056 – – – –
Conc 2 (HG) 1950 3980 338 1330 725 435 2380 729 4860 1150 38900 3560 494 3100 455
TREE, % Sc, g/t Th, g/t U, g/t Si, % Al, % Fe, % Mg, % Ca, % Na, % K, % Ti, % P, % Mn, % Cr, %
Conc 1 (LG) 1.17 – 1494 <85 12.0 2.99 18.7 1.00 12.4 1.12 0.81 0.94 0.45 0.24 0.01
Conc 2 (HG) 6.44 38 8240 154 10.0 1.66 22.8 0.25 6.58 0.979 0.15 3.84 2.03 0.39 0.03
V, % Ag, g/t As, g/t Ba, g/t Be, g/t Bi, g/t Cd, g/t Co, g/t Cu, g/t Li, g/t Mo, g/t Ni, g/t Pb, g/t Sb, g/t Se, g/t
Conc 1 (LG) 0.04 <2 <40 405 7.35 <20 <2 118 42 27 <40 <70 <30 <50 <30
Conc 2 (HG) 0.10 – – – – – – – – – – – – – –
Sn, g/t Sr, g/t Tl, g/t Zn, g/t S, % Cl, g/t
Conc 1 (LG) <20 179 <30 38 0.03 –
Conc 2 (HG) – – – – 0.11 62
The Beneficiation and Hydrometallurgical Process Development for the Lofdal HREE…
711
712 J. Liu et al.
100000 120
90000
100
80000
Fe/Ca Liquor As s ay (mg/L)
70000
Y/Nd Liquor Assay (mg/L)

80
60000
50000 60
40000
40
30000
20000
20
10000
0 0
-1 0 1 2 3 4 5 6
Pulp pH
Fe Ca Y Nd
Fig. 10. Effect of pulp pH on gangue acid leaching.
calcium and 22% of the neodymium was leached, leaving almost all iron and yttrium
behind.
While the tests supported the use of hydrochloric acid for the removal (up to 91%)
of calcium-bearing gangue minerals in advance of other process steps, the added com-
plexity of the overall process likely outweighs the apparent benefits.
Gangue acid leach residue (1.5% Y) was caustic cracked at 140 °C for 2 hours using
50% NaOH (2 t/t NaOH) and the washed residue was subsequently acid leached (HCl)
for 3 hours at 50 °C and controlled acidity of 50 g/L HCl. Under these conditions,
approximately 77% of the thorium dissolved as well as 14% of the yttrium. The acid
leach residue was processed through a second caustic crack step (similar conditions)
and its residue acid leached again (similar conditions), which led to a further thorium
dissolution of 64% (relative to first acid leach residue) and 18% yttrium. It was clear
that despite the aggressive conditions applied, no significant separation between yttrium
and thorium was accomplished and no further testwork investigating this process route
was undertaken.
5.2 Testwork Supporting Acid Bake-Based Flowsheet
Acid Baking and Water Leaching SGS recommended to evaluate the acid bake pro-
cessing route, which is also applied at the other xenotime project (Browns Range) [1]
and also at the Lynas processing plant in Malaysia. At the onset of the program, some
Table 7. Acid bake performance – preliminary performance.
AB1 AB2 AB3 AB4 AB5
Acid baking
Feed sample Con 1 GAL res GAL res Con 1 Con 1
% Y in feed 0.91 1.24 1.24 0.91 0.91
Effective reagent dose (kg/t) 1380 1502 1003 1242 890
Target temperature (°C) 280 200 280 200 250
Test time (h) 3 3 3 3 3
Water leaching
Initial pulp density (% 10% 10% 10% 10% 10%
solids)
Residence time (h) 4 4 4 4 4
Temperature (°C) 25 25 25 25 25
Final filtrate acidity (g/L) 19 121 15 45 9
Acid consumption (kg/t) 1340 517 877 873 924
Sulphur accountability 57% 79% 51% 77% 70%
Metal recovery, %
Nd 55.7 67.7 68.3 53.3 52.0
Y 91.1 66.2 95.7 45.3 62.0
Th 75.5 60.5 56.9 79.9 75.9
Al 48.4 38.2 47.7 40.2 39.5
Fe 70.3 38.8 39.6 82.5 78.4
Mg 96.7 90.7 94.3 93.5 91.9
Table 8. Acid bake performance – optimization (Concentrate 2).

AB12
AB6 AB7 AB8a AB9-1 AB9-2 AB9-3 AB10 AB11 (Bulk)
Acid baking
Feed sample Con 2 Con 2 Con 2 Con 2 Con 2 Con 2 Con 2
% Y in feed 3.89 3.89 3.89 3.89 3.89 3.89 3.89
Effective reagent dose 1440 1446 1175 1052 948 681 1194
(kg/t)
Target temperature (°C) 280 300 300 300 300 300 300
Test time (h) 3 3 3 3 3 3 3
Water leaching
Initial pulp density (% 10% 10% 10% 10% 10% 20% 10% 10% 20%
solids)
Residence time (h) 4 4 4 4 4 4 4 4 2
Temperature (°C) 25 25 25 25 80 25 25 25 25
Final filtrate acidity 31 10 12 16 21 44 10 7 52
(g/L)
Acid consumption (kg/t) 1242 1418 1400 1112 1076 1097 916 686 1067
Sulphur accountability 57% 45% 48% 59% 58% 44% 62% 73% –
Metal recovery, %
Nd 95.6 95.6 96.2 94.7 94.6 94.9 90.2 86.3 95.2
Y 97.8 97.9 98.2 97.6 97.9 97.1 92.4 58.9 97.4
Th 69.9 73.5 81.1 67.9 65.3 62.6 57.1 45.0 75.0
Al 20.1 21.5 30.2 23.9 28.7 19.3 23.9 23.3 27.7
Fe 57.9 60.2 81.9 63.3 63.6 63.2 54.5 53.7 59.4
Mg 70.9 71.3 86.8 74.6 87.8 70.8 77.3 77.3 82.7
a
Acid and feed mixed in Pyrex reactor with overhead mixing
714 J. Liu et al.
concerns existed over the potential negative effect of high levels of calcium in the flota-
tion concentrate. Gypsum formation in the presence of soluble REE is known [1] to
negatively impact REE extraction due to co-precipitation of REE. Therefore, scoping
level testwork was conducted using gangue leach residue (low calcium) and a high
calcium mineral concentrate. A summary of the scoping level testwork in shown in
Table 7.
It was found that high extractions (>90%) of yttrium were feasible with both samples,
provided that sufficient sulphuric acid was added and the bake temperature was at least
280 °C. While higher yttrium extractions were accomplished using a gangue leach resi-
due, it was nonetheless decided to take out the gangue leach circuit and focus on direct
acid baking of flotation concentrate directly. Gangue leaching with hydrochloric acid
requires a hydrochloric acid regeneration step elsewhere in the flowsheet and is expected
to lead to at least two (gangue leach, acid regeneration) additional solid/liquid separa-
tion steps. The additional complexity was not considered worthwhile and it was also
assumed that with further adjustments of the leach process and feed stream (via
improved flotation performance) a simpler flowsheet could be developed capable of
equally high yttrium extraction levels, with capability to produce a high-grade
mixed heavy rare earth oxide or carbonate final product.
Additional acid bake testwork was conducted using an improved flotation concen-
trate sample (“Con 2”, 3.89% Y) and focused on optimizing acid bake (acid addition,
temperature) and water leach (pulp density) test conditions while targeting high (>95%)
yttrium extraction. A summary of the testwork conditions and results are shown in
Table 8.
100
90
80
70
Metal Extraction, %
60
50
40
30
Yrium Neodymium Thorium Iron
20
10
0
600 800 1000 1200 1400 1600
Acid Addion, kg/t
Fig. 11. Effect of acid addition on leach extraction (Concentrate 2).
100
90
80
70
Metal Extraction, %
60
50
AB6
AB7
40
AB8*
AB9-1
30
AB9-2
20 AB9-3
AB10
10 AB11
AB12 (Bulk)
0
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Y Er Tm Yb Lu Th U
Fig. 12. REE extraction in acid bake optimization testwork (Concentrate 2).
Figure 11 shows that in order to achieve Y extraction of >95%, a minimum acid

addition of 1250 kg/t H2SO4 is required. The same data set also shows that thorium
extraction increases with increasing Y extraction. Neodymium and iron extraction are
less affected by increasing acid additions. Individual REE extraction test data are pre-
sented in Fig. 12 showing that except for tests AB10 and AB11, all tests led to yttrium
extraction of greater than 95%. The decrease in extraction in these tests can be attrib-
uted to the lower acid addition rates (1000 and 750 kg/t, respectively).
The calcine of test AB9 (1250 kg/t, 3 h at 300 °C) was split into three cuts and water
leached separately under varying conditions (WL1: 10% solids, 25 °C, WL2: 10% sol-
ids, 80 °C, WL3: 20% solids, 25 °C). The data showed that increasing the leach tem-
perature did not improve REE extraction. The data also showed that increasing the pulp
density from 10% to 20% solids led to a small decrease in REE extraction (yttrium
down by 0.8% and dysprosium down by 2.2%, respectively). Bulk acid bake testing to
produce feed for downstream testing was conducted at 1250 kg/t, 3 hours, 300 °C and
water leached at 20% solids to produce a concentrated pregnant leach solution (PLS) of
14.2 g/L TREE (8.2 g/L Y, 1.2 g/L Dy), 1.3 g/L Th, 1.2 g/L Al, 37.1 g/L Fe, 0.4 g/L Mg,
1.2 g/L Ca, 5.8 g/L Ti, 2.3 g/L P, and 180 g/L S.
Acid Requirement It is worthwhile examining how sulphuric acid is consumed in
each test and how the consumption compares to the theoretical acid demand. Sulphuric
acid (or sulphur) is consumed by rare earth elements (1.5 moles sulphuric acid per mole
rare earth), gangue elements (1 mole acid per mole divalent cation, 1.5 moles acid per
mole trivalent cation, 2 moles acid per mole tetravalent cation), free acidity in the water
leach liquor, and by acid blown off in the furnace. A sulphur balance was set up for each
716 J. Liu et al.
Table 9. Sulphuric acid requirement versus actual acid addition.

Feed sample: Con 1 GAL res Con 2 Con 2
Yttrium grade: 0.9 1.2 3.9 3.9 %Y
Pulp density in WL: 10 10 10 20 % solids
Acid demand (REE): 20 26 84 84 kg/t AB feed
Acid demand (REE + Gangue): 598 490 737 737 kg/t AB feed
Acid demand (as free acid): 160 160 160 60 kg/t AB feed
Unaccounted S (blown off in kiln): 505 433 598 531 kg/t AB feed
Total acid demand: 1263 1083 1495 1328 kg/t AB feed
Actual: 1500 1000 1250 1250 kg/t AB feed
Overall mass pull (to AB feed): 12.0% 8.4% 3.0% 3.0% w/w
Required acid dose: 152 91 45 40 kg/t ore
Actual acid dose: 180 84 37 37 kg/t ore
test and any “unaccounted” sulphur is assumed to have been blown off. It is common to
see higher sulphur accountabilities (i.e. lower sulphur losses) for tests carried out at
lower (200–250 °C) bake temperatures. It is also common to see lower sulphur losses
in larger tests (such as in test AB12), since larger tests use a thicker bed height and
provide less opportunity for acid losses. For the current work, sulphur accountabilities
are provided in Table 7 and in Table 8. Finally for each element, a percent expected
dissolution can be estimated to predict acid requirement.
For this exercise, the following assumptions were made:
• Iron leaching is 90% as ferric and 10% ferrous.
• Aluminium and iron dissolution from “Con 1” samples was set at 30% and 40%,
respectively. For “Con 2” samples this was adjusted to 25% and 60%.
• A target free acidity was set at 20 g/L H2SO4 in the final PLS.
• At 300 °C, the sulphur accountability was set at 60%. This means that it is assumed
that 40% of sulphuric acid is blown off at the bake temperature. By operating at a
lower temperature sulphur losses can be decreased.
Table 9 presents the outcome of the acid requirement investigation. Four different
scenarios are evaluated. The first two scenarios use the low-grade concentrate (gangue
acid leach residue in second scenario). The third and fourth scenarios use the higher-
grade flotation concentrate sample leached at either 10% or 20% solids. Finally, using
mass pull information from the beneficiation testwork, an acid demand and acid addi-
tion relative to plant ore feed can be made. A few interesting observations and conclu-
sions are made:
• The acid consumption by rare earth element only is a very small fraction of total acid
requirement (as low as 1% using the low-grade concentrate sample and up to 6.3%
using the high-grade sample leached at 20% solids).
Table 10. Composition of REE purification process streams.

Acid
bake Impurity REE REE
WL removal REE Re-leach UIX ThSX Oxalate Oxalate REE
PLS filtrate precipitate PLS Barren Raffinate precipitate calcine carbonate
mg/L mg/L % mg/L mg/L mg/L % % %
La 288 208 1.26 842 631 637 1.10 2.38 1.31
Ce 513 382 2.30 1540 1160 1160 1.99 4.31 2.37
Pr 60.4 40.3 0.24 165 123 124 0.21 0.45 0.25
Nd 215 145 0.88 577 433 434 0.75 1.62 0.89
Sm 103 68.7 0.41 270 201 198 0.35 0.75 0.40
Eu 65.6 41.7 0.25 181 120 120 0.21 0.46 0.25
Gd 367 278 1.65 1090 785 800 1.40 3.02 1.67
Tb 99.6 71.4 0.41 281 197 200 0.36 0.78 0.43
Dy 724 553 3.43 2210 1650 1630 2.90 6.27 3.44
Ho 181 126 0.75 501 366 369 0.67 1.45 0.80
Y 5760 4220 26.7 17100 13100 13200 22.6 48.9 28.0
Er 522 399 2.43 1580 1230 1210 2.14 4.63 2.54
Tm 79.7 54.8 0.32 216 163 164 0.29 0.63 0.35
Yb 480 333 1.86 1270 1010 993 1.72 3.73 2.09
Lu 64.8 44.2 0.26 177 134 135 0.23 0.50 0.28
Sc 4.33 0.83 <40 g/t 1.66 0.27 0.14 <12 g/t <25 g/t <15 g/t
Th 840 7.46 438 g/t 1.98 0.11 0.04 0.5 g/t 1.1 g/t 1.1 g/t
U 22.1 7.27 334 g/t 23.3 <0.02 <0.02 <0.2 g/t <0.5 g/t <0.3 g/t
Si 0.15 22.2 0.088 0.19 0.99
Al 733 326 1.86 210 174 171 <0.002 <0.005 0.34
Fe 21900 33.3 0.21 1.2 <0.6 <0.6 0.003 0.007 0.01
Mg 290 13700 0.55 324 274 268 0.14 0.3 0.21
Ca 1170 1120 0.69 519 447 439 0.074 0.16 0.43
Na 75 290 0.16 32 <30 <30 0.14 0.31 0.34
K 144 108 <0.008 29 <30 <30 <0.004 <0.008 <0.005
Ti 3640 2.74 0.01 <5 <4 <4 <0.003 <0.006 0.006
P 1550 <5 <0.004 <5 <5 <5 0.002 0.004 0.01
Mn 466 347 0.27 219 181 178 <0.004 <0.008 0.14
Cr 24.8 <0.3 <0.007 <0.3 <0.2 <0.2
V 94.0 <0.2 <0.006 <0.2 <0.2 <0.2
Cu <40 11.0 45.0 <40 <40
Pb <8 <4 <20 <9 <9
Zn <5 6 20.3 15 15 <0.002 0.004
S 0.01
718 J. Liu et al.
• To minimize overall reagent addition on a plant ore feed basis, it is important to feed
the hydromet plant with a REE feed grade as high as practically possible in order to
avoid excessive reagent consumption. In the examples evaluated below, the overall
acid requirement reduced from 375 kg/t ore to ~60 kg/t ore by using a higher-grade
feed sample.
• Predicted acid requirements are not far from actual acid additions determined by
testwork.
Liquor Purification and Production of Purified REE Precipitate The bulk acid
bake leach solution was processed through a series of fairly conventional downstream
unit operations. Due to schedule requirements, little or no optimization was performed.
The process steps consisted of the following: (a) impurity removal (IR) with MgCO3;
(b) bulk rare earth precipitation with Na2CO3; (c) REE precipitate re-leaching with
H2SO4; (d) uranium removal via ion exchange (IX); (e) thorium removal via solvent
extraction (SX); (f) final recovery of a purified REE precipitate using oxalic acid or
sodium carbonate. Table 10 shows the composition of the various REE loaded interme-
diate and final product streams. Liquor dilution (due to combining wash liquors with
leach or precipitation filtrate liquors) accounted for some of the reduction in liquor
trends. Each of the unit operations separated the REE from one or more of the impuri-
ties, summarized as follows:
(a) In IR most of the Sc, Fe, P, and some of the Al were precipitated and separated from
the REE, which were left in solution.
(b) In Bulk REE Precipitation (RP) the REE (together with the U, Th, residual Al, and
Mn) were precipitated and concentrated into a crude REE carbonate precipitate of
26.7% Y.
(c) The RP precipitate was re-leached into a sulphuric acid leach solution, rejecting
Th, Al, and Fe. In the testwork a 28 g/L TREE liquor was produced. Commercially,
a more concentrated liquor is expected to be produced.
(d) Uranium was removed in the UIX circuit. Due to time constraints this was carried
out as two sequential batch contacts and the excess resin supplied to the contact led
to some thorium co-extraction, which at higher uranium loadings is expected to be
crowded off.
(e) Residual thorium was removed in a single thorium SX contact using a primary
amine extractant.
(f) Separation of residual aluminium, manganese, and other divalent metals was
accomplished in the rare earth oxalate precipitation and calcination step. When
sodium carbonate was used to make the final product, this separation did not take
occur and the final REE product contained elevated levels of silicon, aluminium,
and manganese.
The flowsheet (including the rare earth oxalate precipitation and calcination step)
was able to produce a high grade TREO sample of 49% yttrium, 6.3% dysprosium, and
very low levels of radioactive impurities (1.1 g/t Th, 0.5 g/t U).
6 Conclusions
The flowsheet developed in this programme simplified the flowsheet developed previ-
ously and improved yttrium recovery at a similar grade. This represented a ten times
upgrade of the feed material with good recovery. A higher grade of 2nd cleaner concen-
trate at 5.5% Y2O3 (~9% TREO), ~15 times upgrade, was achievable in test T11 at 65%
yttrium recovery, in a 4% mass yield.
The selectivity of yttrium against the main gangue minerals (calcite and hematite)
was significantly improved with the application of a hydroxamate type of collector,
depressants sodium silicate and Calgon, and iron pre-removal by WHIMS. The amount
of calcium in the flotation concentrate, a high acid consumption impurity, was largely
reduced with ~2% CaO reported to the concentrate.
The developed flotation flowsheet also made advances to reduce the OPEX cost in
the commercial plant by showing promising results with a high pulp density slurry
(35% vs. ~20% in previous studies) and coarser flotation feed (100% passing
75 μm/53 μm vs. 100% passing 38 μm in previous studies).
The testwork has shown that high (>95%) extraction of REEs from a xenotime min-
eral concentrate is possible by applying an acid bake process. The originally proposed
caustic crack process route did not accomplish a good separation between REE and
impurities, such as Th and U. The acid bake testwork showed the impact of the grade of
the mineral concentrate on the overall reagent consumption in the hydromet processing
plant and emphasized the importance of producing a high-grade concentrate.
The REE-loaded PLS was easily processed through a fairly conventional REE liquor
treatment flowsheet consisting of impurity removal, crude rare earth precipitation, re-
leaching, uranium removal via IX, thorium removal via SX, and production of a
high-purity rare earth oxide via rare earth oxalate precipitation and calcination. The
final REE oxide product contained 98.2% REO (of which 48.9% Y and 6.3% Dy and
only 1.1 g/t Th, 0.5 g/t U).
Acknowledgements The authors thank the management of NCM and SGS Canada for permission to
publish this paper.
Reference
Verbaan N, Bradley K, Brown J, Mackie S. A review of hydrometallurgial flowsheets considered in

current REE projects. In: Simandl GJ, Neetz M, editors. Symposium on strategic and critical mate-
rials proceedings, November 13–14, 2015. Victoria, British Columbia: British Columbia Ministry
of Energy and Mines, British Columbia Geological Survey Paper 2015-3; 2015
Hydrometallurgical Flowsheet Options for Treatment
of Wicheeda Lake Flotation Concentrate: Acid Bake
Versus Caustic Crack
Niels Verbaan1 (*), Tassos Grammatikopoulos1, Mike Johnson1, Jing Liu1,

John Goode2, Kris Raffle3, and Craig Taylor3
1
niels.verbaan@sgs.com
2
J.R. Goode & Associates, Toronto, ON, Canada
3
Defense Metals Corporation, Vancouver, BC, Canada
Abstract. Defense Metals is currently developing the Wicheeda rare earth element
(REE) carbonatite deposit in British Columbia, Canada. The deposit is dominated by
gangue carbonates, including ankerite and dolomite, while the REE-bearing minerals
include bastnaesite/parisite/synchysite and monazite, hosting primarily light rare earths,
including praseodymium and neodymium. This paper presents the results from a metal-
lurgical testwork programme conducted on flotation concentrate samples produced dur-
ing a flotation pilot plant at SGS Lakefield. The concentrate samples assayed 40–50%
TREO and contained approximately 21% monazite and 64% REE fluorocarbonates.
From 2019 to 2021, a hydrometallurgical flowsheet was developed employing caustic
conversion of REE phosphates and fluorocarbonates followed by hydrochloric acid leach-
ing capable of extracting 90–92% REE into the leach solution. The leach solution was
treated in a series of impurity removal processes to remove elements, such as iron, alu-
minium, phosphate, and thorium. REEs were recovered from the purified solution via
precipitation. In late 2021, testwork was initiated investigating alternative treatment
options. Preliminary acid bake testwork yielded improved metallurgical extraction per-
formance (> 95% PrNd) and showed the potential to reject thorium into the leach
residue.
The paper and presentation will provide updated results of the caustic crack and acid
bake processes and comparisons between the two routes in terms of metallurgical perfor-
mance (extraction, reagent consumption) and process simplicity. Mineralogical charac-
terization of leach residues will be used to explain the metallurgical performance of the
two process routes.
Keywords: Rare earth elements · Monazite · Bastnaesite · Parisite · Synchysite ·

Caustic crack · Acid bake · Leaching
https://doi.org/10.1007/978-3-031-17425-4_84
722 N. Verbaan et al.
1 Introduction
Defense Metals is advancing the Wicheeda rare earth element (REE) carbonatite deposit
in BC, Canada. The Wicheeda deposit comprises an indicated mineral resource of 5 Mt
averaging 2.95% total rare earth oxides (TREO) and inferred mineral resources of 29.5
Mt averaging 1.83% TREO, reported at a cut-off grade of 0.5% TREO within a concep-
tual Lerchs–Grossman (LG) pit shell [1].
The Wicheeda deposit is modelled as a composite layered syenite–carbonatite sill
complex having dimensions of approximately 400 m north–south by 100–250 m east–
west. The sill complex comprises low REE grade syenite at its base, overlain by transi-
tional intermediate REE grade xenolithic carbonatite, and finally relatively higher REE
grade dolomite–carbonatite rocks, which form the main mineralization of the Wicheeda
deposit outcropping at surface.
The Wicheeda project was the subject of a positive PEA [1] (preliminary economic
assessment) study that contemplated a 1.8 Mt/a mill throughput open-pit mining opera-
tion over a 16-year mine life at an average head grade of 2.33% REO and yielding
average annual REO production of 25,423 t. The PEA describes a well-developed base
case flotation concentration and hydrometallurgical pre-leach, caustic crack. leach
flowsheet capable of achieving high REE recoveries into a mixed REE precipitate prod-
uct. Alternative treatment options, including acid bake, have the potential to simplify
the flowsheet and may contribute to decreased capital and/or operating costs.
Previous metallurgical studies by SGS [2] for Defense Metals have reported on the
metallurgical development testwork leading to a processing flowsheet consisting of
mineral processing and hydrometallurgical techniques.
2 Production of Mineral Concentrate
The mineral concentrate used for this work was produced during pilot plant operations
conducted at SGS in 2020. The flotation pilot plant was designed, built, and operated
based on batch flotation and locked cycle test results. Approximately 26 tonnes of bulk
material was processed at a throughput of 140 kg/h, for an operating duration of 178 h,
spread out over 14 single shifts, and 2 extended shifts. Ultimately the pilot plant pro-
duced around 1.2 tonnes of REO concentrate, including ~950 kg high-grade material
(45–50% TREO). The confirmatory campaigns showed an average REO recovery of
~77%, grading ~51% TREO, and a mass pull of 7%.
The composition of the sample used for the bulk of the hydrometallurgical testwork
is presented in Table 1. The concentrate (37.0% TREE or 44.4% TREO) consists of
synchysite/parisite (61%), monazite (21%), and carbonates (14%, i.e. dolomite and
ankerite).
Hydrometallurgical Flowsheet Options for Treatment of Wicheeda Lake Flotation… 723
Table 1. Head assays of the flotation concentrate.

La % 13.5 TREE % 37.0
Ce % 17.5 LREE % 36.2
Pr % 1.36 HREE % 0.74
Nd % 3.88 Th g/t 3810
Sm % 0.37 U g/t 7
Eu g/t 775 Si % 0.12
Gd g/t 1690 Al % 0.02
Tb g/t 128 Fe % 2.02
Dy g/t 339 Mg % 2.44
Ho g/t 25.4 Ca % 9.49
Y g/t 709 Na % 0.12
Er g/t 26.5 K % 0.01
Tm g/t 1.5 P % 2.81
Yb g/t 3.9 Mn % 0.35
Lu g/t 0.5 S % 0.03
3 Hydrometallurgical Process Development Testwork
3.1 Process Development (2020–2021)
Due to the high presence of dolomite (Ca contents in flotation concentrate was ~10%),
the focus of the hydrometallurgical testwork was on non-sulphate-based leaching
processes:
• Roasting followed by Acid Leaching (H2SO4, HCl and HNO3 were evaluated)
• Caustic Crack followed by Acid Leaching (HCl based)
• Gangue Leaching followed by Caustic Crack and Acid Leaching (HCl based).
Roast-Acid Leaching The Roast-Acid Leaching process consisted of roasting the
feed sample at 625 °C for 2 h under oxidizing conditions followed by acid leaching at
20–40 g/L acid. The preliminary test results (Table 2) indicated that leaching was
incomplete regardless of which mineral acid was used. The low phosphorous extraction
indicated that monazite was not cracked using the Roast-Acid Leach process route and
this route was therefore not pursued further.
Gangue Leaching and Caustic Cracking The gangue acid leaching and caustic
cracking testwork was described in a previous paper [2] and showed that reasonably
high extractions (90%) of REE were possible under the following conditions:
• Feed is flotation concentrate or flotation concentrate after a Gangue Acid Leach
process in which dolomite and ankerite are selectively leached using dilute (~ pH 3)
hydrochloric acid)
• Caustic addition of 2000 kg/t
• Caustic strength of 40–60% NaOH (w/w)
• Crack time of 2–4 h
• Crack temperature is 140–150 °C
Table 2. Test results – Roast-Acid Leaching.

Test: L9 L10 L11
Feed Calcined at 625 °C, 2 h
Leach Reagent H2SO4 HCl HNO3
Reagent Strength (%) 96% 36% 70%
Initial Pulp Density (%) 15 15 15
Retention Time (h) 2 2 2
Final Acidity (g/L) 22.3 21.9 41.6
Temperature (°C) 70 Ambient 70
Weight Loss (%) 25 59 48
Extraction (%)
La 44 56 49
Ce 52 39 9
Pr 47 54 52
Nd 54 55 56
Sm 58 57 57
Fe 78 72 91
Mg 97 96 100
Ca 11 88 98
P 0.6 0.2 0.1
F 41 48 13
• Water dilution (before filtration) to ~13% NaOH (w/w)

• Acid leaching of caustic cracked / water washed residue at 15–25% solids (w/w) and
50 °C for 4 h
• Final acidity of 30–50 g/L HCl.
The testwork results showed that even under very aggressive conditions, complete
REE extraction was not achieved, reaching a plateau around 90%. The final acid leach
residue still contained 10–12% P on average, which suggested that the monazite con-
version within the caustic crack step was incomplete. Reprocessing the acid leach resi-
due (representing around 1% of flotation concentrate weight) through another round of
caustic cracking and acid leaching after pulverization would result in nearly 99% Nd
extraction, whereas the original crack/leach was limited to 90% Nd extraction.
Leach Liquor Purification and Recovery of Purified REE Product The acid leach
solutions were used in a test programme focusing on the recovery of a clean REO prod-
uct. Results of this work have been previously presented [2] and showed that high
purity mixed REE products could be produced from leach liquors.
The leach liquor purification process by hydrolysis focused on the use of calcium-
based reagents (limestone) in order to produce a final calcium chloride liquor suitable
for regeneration of hydrochloric acid. Effective removal of thorium (100%), iron (94%),
aluminium (76%), and phosphorous (99%) was achieved at pH 2.5 with low co-
precipitation of LREE (around 3%). Further reduction of some impurities is feasible at
higher pH levels, but increased REE losses will be encountered.
A multi-stage impurity removal circuit was therefore envisioned with the first stage
operating at pH 2.5 (or slightly lower), removing most of the Fe/Th/Al/P with minimum
REE co-precipitation. Sodium sulphate was added to the filtrate to remove barium and
any radium as sulphates. The filtrate was further treated with NaHS (IR2) to remove
residual lead (and associated isotopes). Residual iron levels can be removed in a sec-
ondary neutralization (SN) step, where the pH is further adjusted to pH 3.5 to precipi-
tate the remainder of the impurities (which are recycled back to the acid leach circuit to
recover any co-precipitated REE) and produce a clean feed solution to final rare earth
precipitation.
REE contained in the purified leach liquor was recovered via one of two methods: by
direct precipitation of hydroxides with hydrated lime to pH 8 or by selective precipita-
tion through addition of oxalic acid. Both worked equally well in terms of REE
Table 3. Test results – liquor purification and production of purified REE hydroxide or oxide.
Element Units Conc IR IR2 SN Hydroxide Calcined
PLS Filtrate Filtrate Filtrate Precip. Oxalate
La mg/L, % 38,700 34,700 30,200 37,600 26.6 27.4
Ce mg/L, % 37,900 33,400 30,300 36,400 25.3 35.5
Pr mg/L, % 4010 3560 3280 3970 2.6 4.47
Nd mg/L, % 11,200 9850 9360 10,100 7.5 13.4
Sm mg/L, % 1060 907 935 983 0.7 1.37
Eu mg/L, g/t 233 193 180 205 1430 2870
Gd mg/L, g/t 484 435 443 442 3450 6270
Tb mg/L, g/t 38.7 32.3 37.1 36.9 238 434
Dy mg/L, g/t 98.2 84.3 87.4 93.9 673 1080
Ho mg/L, g/t 7.93 6.71 7.78 7.04 50.4 81.4
Y mg/L, g/t 180 158 137 172 1110 1180
Sc mg/L, g/t 0.12 <0.07 <0.07 <0.07 <200 <300
Th mg/L, g/t 15.3 0.03 <0.03 0.03 <0.009 1.8
U mg/L, g/t 0.24 0.07 0.08 0.02 <600 <0.5
Si g/t – – – – 1402 600
Al mg/L, g/t <20 <20 <2 1.4 370 300
Fe mg/L, g/t 1590 6.5 6 0.8 560 500
Mg mg/L, g/t 1150 1440 1320 1300 302 100
Ca mg/L, g/t 14,000 27,800 26,000 26,900 2430 4400
Na mg/L, g/t 1120 1970 1860 1850 445 <1000
P mg/L, g/t 157 <9 <5 <5 44 40
Mn mg/L, g/t 127 117 114 114 697 <100
Zn mg/L, g/t 4.2 5.5 2.7 2.8 39 20
Ba mg/L, g/t 603 11.4 12.3 13.9 – 42
F mg/L, % – – – – – < 0.005
Ni mg/L, g/t 24 357 19.5 21 – 124
Pb mg/L, g/t <50 13.9 3.11 <60 – <200
Sr mg/L, g/t 705 746 748 732 – <80
S % – – – – – 0.01
p recipitation efficiency, generating a high-grade REE precipitate suitable for further

refining. Table 3 lists key assays of the main process streams, showing successful
removal of Fe, Th, Sc, and Ba in IR, removal of Pb in IR2 and polishing of Fe in
SN. The use of hydrated lime resulted in some silicon contamination, as well as minor
quantities of iron and aluminium. Carefully controlling the pH is expected to help
reduce the level of calcium in the product. Total REE grade at pH 8 was 63.3%.
The REE oxalate precipitate was also prone to calcium contamination, though this is
directly related to the amount of oxalic acid added. By limiting the oxalic acid addition
to a near- or sub-stoichiometric amount, the calcium grade in the precipitate can be
controlled. The final TREE grade was 83.3% (98.9% TREO) when the oxalate precipi-
tate was calcined to convert the material from REE oxalates to oxides.
Overall Flowsheet and Potential Issues Based on the testwork results a conceptual
flowsheet was developed and is presented in Fig. 1. While this flowsheet is capable of
high REE extractions and producing a high-grade REE product, several uncertainties
(and related risks) were identified within the flowsheet, specifically:
• The overall flowsheet is complex, requiring the regeneration of two separate reagent
regeneration processes (NaOH and HCl). Each of these operations will require addi-
tional testing as the project advances.
• To limit chloride consumption within the chloride flowsheet, wash efficiencies of the
acid leach residue and various impurity removals S/L separation steps need to be
high. There will be a trade-off between wash efficiency (i.e. recovery of chloride)
and energy consumption (i.e. evaporation of excess wash water during the HCl
regeneration process steps).
• To limit sodium consumption within the caustic flowsheet, wash efficiency of the
caustic water wash residue, and causticization residue S/L separation steps needs to
be high. There will be a trade-off between wash efficiency (i.e. recovery of sodium)
and energy consumption (i.e. evaporation of excess wash water within the caustic
regeneration steps). Related to this concern, any sodium from the caustic crack cir-
cuit carried over into the chloride stream will likely buildup to high levels before
bleeding out via moisture content of the precipitates and leach residue.
• The monazite contained in the concentrate appears to be passivated requiring at least
two passes of caustic crack processing to obtain >95% extraction of Nd. Based on
this concern, a separate mineralogy study was initiated late 2021 using TIMA-X to
examine the feed concentrate and various intermediate process residues from the
caustic crack process flowsheet.
Mineralogical Analyses of Feed and Leach Residues A mineralogical examination

was conducted using TIMA-X (TESCAN Integrated Mineral Analyzer) on four sam-
ples (Table 4), including a Flotation Concentrate (identified as HG), Gangue Leach
(BGL1) Residue, Caustic Crack/Water Wash Residue (BCC1), and Acid Leach Residue
(AL5-BCC1).
Table 4 shows that the concentrate consists primarily of synchysite/ parisite (61%),
monazite (ca. 21%), and carbonates (14%, i.e. dolomite and ankerite). It can be seen
that most (ca. 10%) of the carbonates are removed in the gangue leach operation leav-
ing around 4% of the carbonates in the gangue leach residue. Based on the caustic crack
HCl
Gangue Leach (GL)
Concentrate
NaOH
Ca(OH)2
Caustic Causticization
Caustic Crack 1 Causticization
Regeneration residue
HCl HCl
Acid Leach 1 Acid Leach 2

Caustic Crack 2 Leach Residue
(AL1) (AL2)
Ca(OH) 2 Na2SO4 / NaHS
Impurity Hydroxide / Rare Earth REO Product to

Impurity Removal
Sulphate / Sulphide Precipitation Further Refiing
H2 SO4
HCl to GL, AL1

and AL2
Acid Regeneration Gypsum
Fig. 1. Conceptual BFD – caustic crack-based flowsheet.
Table 4. Modal mineralogy of the concentrate and products.

HG BGL-1 Final BCC1 Washed AL5-BCC1 Final
Mineral / Sample Units Concentrate Residue Residue Residue
Mass Units 100 79 64 15
Monazite % 20.86 21.93 6.53 8.12
Monazite K80 μm 81 81 39 88
Hydrated Monazite % 0.86 0.94 0.00 91.65
Hydrated Monazite K80 μm 8 8 10 173
Synchysite/Parisite % 61.13 68.42 65.72 0.03
Altered Synchysite/Parisite % 2.49 3.83 12.36 0.00
Bastnaesite % 0.22 0.29 1.39 0.00
Qaqarssukite % 0.07 0.03 0.00 0.00
REE-CaP Mixtures % 0.00 0.00 6.44 0.00
Lanthanite % 0.07 0.07 0.21 0.02
Thorite % 0.07 0.08 0.00 0.00
Carbonates % 13.86 4.05 0.28 0.01
Quartz/Feldspars % 0.13 0.11 0.11 0.05
Apatite % 0.02 0.02 6.60 0.00
Other % 0.23 0.21 0.34 0.10
Total % 100 100 100 100
residue composition it appears that considerable monazite has been cracked, and REE
are dominated by fluorcarbonate REE minerals. The acid leach process seems to have
removed all F-C-REE, but a hydrated monazite has been formed.
A closer examination reveals that the monazite in the caustic crack residue is par-
tially dissolved. It is rimmed by depleted in phosphorous REE compounds which
resemble bastnaesite/synchysite/ parisite (but possibly without F and C) (Fig. 2), and by
hydrated monazite (Fig. 3). The latter is depleted in P but enriched in Ce and La
(Table 5). While the monazite (Fig. 4) is well exposed in the concentrate, it is poorly
exposed (due to the rims) in the caustic cracked residue. This may therefore provide an
explanation to the partial (<90%) leaching of the primary monazite in a single pass of
cracking and acid leaching. The coarser grain size of hydrated monazite in the final resi-
due might also indicate the need for re-grinding in between the first and second pass of
caustic crack/acid leaching.
Most of the monazite in the acid leach residue (AL5-BCC1) appears to be partially
and texturally degraded. It hosts significant Th and F, and it is enriched in Ce more than
La and Nd. The fluorocarbonates in the caustic cracked residue (BCC1) appear to be
also texturally and chemically degraded (deficient in F and C) and are absent in the final
acid leach residue (AL5-BCC1).
Fig. 2. Clockwise, back-scattered image of monazite and synchysite/parisite, pseudocolour mineral

image, P and Ce X-ray maps.
Fig. 3. Back-scattered image of monazite (point 1) rimmed by hydrated monazite (point 3).
Table 5. Semi-quantitative chemical analysis of monazite and hydrated monazite.

Spectrum O F Na Mg P Cl Ca Fe Sr La Ce Nd Total Mineral ID
1 29.4 13.6 1.3 22.9 27.4 5.3 100 Monazite
2 34.1 0.9 2.0 5.1 0.4 11.0 8.4 14.4 20.2 3.6 100 Hydrated
monazite
3 27.7 1.3 0.4 28.8 35.6 6.1 100 Hydrated
monazite
4 30.0 2.0 4.7 1.5 9.3 1.9 20.2 25.3 5.2 100 Hydrated
monazite
3.2 Process Development (2021–2022)
The mineralogical analyses provided a plausible explanation (and remedy) for the
recovery losses of the caustic crack/acid leach process and provided rationale for evalu-
ating alternative flowsheets. Acid bake processing is considered on many REE projects
[3–5] and is also applied on Lynas’ monazite/apatite hosted Mt Weld deposit and the
Bayan Obo-sourced bastnaesite/monazite concentrates. A short series of scoping acid
bake tests was performed in late 2021 and showed that high extractions (>95%) of Pr/
Nd were possible at 1500–2000 kg/t H2SO4 and temperatures of 250–300 °C. The pre-
liminary tests also showed that some thorium rejection can be realized by operating at
the higher temperature. Based on the positive outcome of these preliminary tests, a
more detailed test programme was initiated at the start of 2022, focused on optimizing
REE recoveries and maximizing thorium rejection to the leach residue at minimum
reagent addition.
Acid Requirement Calculation A theoretical sulphuric acid requirement by the con-
centrate can be estimated taking into account:
Monazite Exposure
100.0
90.0
80.0
70.0
Mass % Monazite
60.0
50.0
40.0
30.0
20.0
10.0
0.0
BGL-1 Final BCC1 Washed AL5-BCC1
HG Concentrate
Residue Residue Final Residue
Locked 0.1 0.1 6.4 0.2
>0-<10% Exposed 1.1 1.2 26.2 9.3
>10-<20% Exposed 1.3 1.3 47.8 9.5
>20-<30% Exposed 1.2 1.1 9.3 6.7
>30-<40% Exposed 1.3 1.0 3.3 7.4
>40-<50% Exposed 1.4 1.1 1.1 6.1
>50-<60% Exposed 2.5 1.7 0.8 6.8
>60-<70% Exposed 3.1 2.4 0.6 8.5
>70-<80% Exposed 5.0 3.6 0.6 4.1
>80-<90% Exposed 12.2 10.0 0.5 3.7
>90-<100% Exposed 26.2 27.4 0.1 2.0
100% Exposed 44.5 49.1 3.2 35.9
Fig. 4. Exposure of monazite in all samples.
• The composition of the concentrate, assuming 1.5 moles sulphuric acid per mole
lanthanide), gangue elements (1 mole acid per mole divalent cation, 1.5 moles acid
per mole trivalent cation, 2 moles acid per mole tetravalent cation, etc.).
• Target free acidity at the water leach stage. At low pulp densities this can be a signifi-
cant fraction of overall acid demand.
• Sulphuric acid losses in the calcination step. The boiling point of concentrated sul-
phuric acid is 327 °C, so evaporative losses are expected to increase at higher tem-
peratures. For the purpose of this exercise an acid loss percentage of 20% has been
assumed. In the testwork, it is seen that actual acid losses are dependent on test
temperature, feed charge size (larger tests lead to lower acid losses), and exposure of
sulphated mixture to furnace atmosphere (i.e. tests equipped with a lid display lower
acid losses).
• Decomposition of sulphated metals (REE, Th, Fe, etc.) into their metal oxides and
SO2. At temperatures at or below 300 °C this is assumed to be negligible, but this
will increase at more elevated temperatures where salts, such as ferric sulphate, are
expected to decompose.
Assuming a target free acidity of 15 g/L H2SO4 at a water leach pulp density of 5%
solids and 20% sulphur losses across the calcination step, the acid requirement can be
calculated not only for the concentrate sample as used (37% TREE) in the test pro-
gramme but also for lower- and higher-grade samples as shown in Table 6. The data
show that under the assumptions applied an acid dosage of 1230–1365 kg/t (relative to
concentrate) should be expected. These values can be reduced if:
• The water leach can operate at higher pulp densities (and avoid solubility limitations
or REE losses due to gypsum formation).
• The level of free acid in the water leach can be reduced.
• More sulphuric acid be retained in the sulphated mixture. The acid demand calcula-
tion shows that if sulphur losses could be limited to 5%, the required acid dosage
could be reduced to ~1100 kg/t for the 37% TREE concentrate sample.
Acid Bake Test Results The testwork carried out in 2022 examined the impact of
concentrate grade, acid dosage, test temperature, test procedure, water leach pulp den-
sity, and temperature. This paper includes information available at time of
preparation.
The impacts of acid bake temperature and acid dosage are shown in Fig. 5 and in
Fig. 6. The temperature data show that while neodymium extraction remains high
(>90%) across the temperature range investigated, the extraction of dysprosium drops
at temperatures above 300 °C, and thorium extraction decreases with increasing tem-
perature. Based on these results a bake temperature of 300 °C seems optimum. The
effect of acid dosage is clear and an optimum performance seems to be reached at or
around 1375 kg/t, close to the expected acid demand for that concentrate.
It is of interest to note the two data points for thorium at 300 °C (Fig. 5) and 1500 kg/t
(Fig. 6). The only difference between these two tests is the size of the tests. High tho-
rium extraction was accomplished in the test with higher (3×) feed weight. This effect
was investigated in a second series of tests (conducted at the low end of the acid addi-
tion range, 1000 kg/t), with one test using 100 g of feed and the other 400 g. A lid was
placed on both crucibles. These two tests were subsequently compared to a test carried
out without a lid, but otherwise identical conditions. A sulphur balance around the cal-
cination step revealed that the acid losses were low in the covered tests (15% in the
small test and 6% in the larger test), but significantly higher in the uncovered test. Metal
extraction and sulphur accountability (plotted as “extraction”) are presented in Fig. 7
and show very significant impact of reducing acid losses on REE and Th extraction with
Nd extraction increasing from 71% (uncovered) to 96% (covered, larger batch size).
The effect on thorium is even more pronounced.
While acid retainment control measures (such as covering a crucible) exist in lab-
scale experimentation, control measures during continuous processing in a rotary kiln
may be more limited or of a different nature. A relationship between sulphur levels in
the calcine and metal extraction was established (Fig. 8) which can be used during
continuous experimentation to gain quick feedback on predicted metal extraction and
acid losses.
Table 6. Concentrate composition and acid requirement.
TREE % 33.0 37.0 43.2 46.5 Mole Acid/Mole Metal
La % 11.8 13.5 15.5 16.7 1.5
Ce % 15.8 17.5 20.5 22.1 1.5
Pr % 1.28 1.36 1.67 1.80 1.5
Nd % 3.47 3.88 4.58 4.91 1.5
Sm % 0.34 0.37 0.45 0.49 1.5
Eu % 0.070 0.078 0.095 0.102 1.5
Gd % 0.153 0.169 0.207 0.221 1.5
Tb % 0.011 0.013 0.016 0.017 1.5
Dy % 0.031 0.034 0.042 0.046 1.5
Ho % 0.002 0.003 0.003 0.003 1.5
Y % 0.055 0.071 0.094 0.071 1.5
Er % 0.0028 0.0027 0.0048 0.0050 1.5
Tm % 0.0002 0.0002 0.0002 0.0002 1.5
Yb % 0.0005 0.0004 0.0004 0.0005 1.5
Lu % 0.0004 0.0001 0.0006 0.0006 1.5
Th % 0.35 0.38 0.45 0.48 2.0
U % 0.0002 0.0007 0.0002 0.0002 1.0
Si % 0.11 0.12 0.07 0.10
Al % 0.02 0.02 0.02 0.03 1.5
Fe % 2.55 2.02 0.99 1.29 1.45 (assume 90% Fe3)
Mg % 3.13 2.44 1.17 1.53 1.0
Ca % 10.8 9.49 7.13 7.79 1.0
P % 2.59 2.81 3.19 3.09
Mn % 0.46 0.35 0.17 0.22 1.0
Acid for Metals kg/t 817 783 711 785
Free Acid kg/t 276 276 276 276
Acid Lost kg/t 273 265 247 265
Total Acid Required kg/t 1365 1324 1233 1326 Relative to concentrate
Total Acid Required kg/t 139 116 91 90 Relative to ore
100
90
Calc/Est. Extraction (%)
80
70
60
50
40
30
20
10
0
0 100 200 300 400 500 600
Bake Temperature (°C)
Nd Th Dy Filtrate FA (g/L)
Fig. 5. Impact of acid bake temperature at 1500 kg/t.

100
90

80
70
60
50
40
30
20
10
0
500 750 1000 1250 1500 1750
Acid Addition (kg/t)
Nd Th Dy Filtrate FA (g/L)
Fig. 6. Impact of acid addition at 300 °C.
100
90
80
70
Metal Extraction, %
60
50
40
30 AB5 (100 g, no lid, 39% acid retained)
20 AB13 (100 g, with lid, 85% acid retained)
10 AB14 (400 g, with lid, 94% acid retained)
0
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Y Er Tm Yb Lu Th
Fig. 7. Effect of acid retainment on extraction (1000 kg/t H2SO4, 300 °C, 5% solids in water leach).
Tests investigating the effect of pulp density showed that increasing from 5% to 10%
solids did not have any negative impact on REE extraction, providing an opportunity to
reduce acid addition. At 10% solids, a liquor containing 48 g/L TREE was produced.
Conceptual Flowsheet While the testwork is still ongoing at time of preparation, it is
clear that high extractions can be accomplished in a single leach operation. Leach liquor
treatment testwork is ongoing and is expected to lead to a conceptual flowsheet configu-
ration as shown in Fig. 9.
100
90
80
70
60
50
40 Nd Dy Th
30
20
10
0
0 5 10 15 20 25
S in calcine, %
Fig. 8. Correlation between sulphur levels in the calcine and metal extraction.
H2SO4 H2O Fe2(SO4) 3 MgCO3 MgCO3
Acid / Feed Primary Secondary

Water Leach L
Acid Bake Neutralization S
Neutralization
Mixing (WL)
(PN) (SN)
H2SO4
Flotation
Leach L S
Concentrate Residue S Re-Leach L
(37% TREE)
CaO Na2CO3 Na2CO3
Bleed REE
L Water L
S S
Precipiattion
Treatment (WT)
(RP)
Liquid process stream

Solid/pulp process stream
Reagent
Purified /
Gypsum Mixed TREO
Precipitate
Fig. 9. Conceptual acid bake-based flowsheet.
4 Caustic Crack Versus Acid Bake
Leach extractions for acid bake and caustic scenarios are shown in Fig. 10. For both pro-
cess routes single pass extraction data are used. Applying a second pass of caustic crack
and acid leaching might increase PrNd extraction to around 97–98% extraction from
100
90
80
70
Metal Extraction, %
60
50
Acid Bake: AB8 (1375 kg/t H2SO4,
40 300C, WL: 5% solids)
30
Caustic Crack: CC6 (2000 kg/t NaOH,
50% NaOH, AL: 50 g/L HCl)
20
10
0
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Y Er Tm Yb Lu Th
Fig. 10. Leach extraction. Acid bake versus caustic crack.
concentrate. Increasing the acid addition to the acid bake (or reducing acid losses) can
improve Nd extraction to 98–99% and Dy from 83% to 85% (or even to 93% at 2000 kg/t).
However, both adjustments come at a considerable cost (increased reagent addition) and/
or circuit complexity. The comparison (as presented) shows the following:
• Acid baking leads to higher PrNd leach extraction than caustic cracking.
• Caustic cracking leads to higher TbDy leach extraction than acid baking.
• The acid bake flowsheet does not incorporate reagent regeneration, has limited recir-
culation of process streams, and appears less complex, and with the long-established
Bayan Obo acid bake operations, and more recently, the Lynas operation is a well-
established process.
• The caustic cracking – acid leaching process is more complex with two major recir-
culating streams (NaOH and HCl) and has only been applied on an industrial scale
to monazite concentrates – never to fluorocarbonate/monazite concentrates.
The authors believe the acid bake route holds significant promise and are planning
to test the acid bake flowsheet in an integrated pilot plant at SGS Lakefield in H2 2022.
5 Conclusions
Defense Metals has developed a caustic crack-based flowsheet capable of high (90%)
extraction of neodymium from Wicheeda Lake flotation concentrate samples. REEs in
the Wicheeda deposit are hosted by synchysite/parisite and monazite. Mineralogical
analyses of the leach residue revealed that the monazite is only partially leached, which
is explained by the formation of a rim around the original monazite particle of hydrated
monazite and a phosphorous-depleted REE compounds resembling

bastnaesite/synchysite/parisite.
Further investigations into alternative process flowsheets showed that high recover-
ies (>97% Nd) can be achieved using an acid bake process. The acid bake process
flowsheet appears to be less complex than the caustic crack process route. Avoiding the
loss of acid by decomposition and volatilization during the bake process is important to
maintain high extraction and low acid addition.
References
1. SRK Consulting (Canada) Inc. Independent preliminary economic assessment for the wicheeda
rare earth element project. British Columbia; 2022. Retrieved from https://www.sedar.com/
DisplayCompanyDocuments.do?lang=EN&issuerNo=00042886. 1 Mar 2022.
2. Verbaan N, Liu J, Johnson M, Gunning C, Aghamirian M, MacNeil T, Wallster D, Taylor
C. Beneficiation and extraction of REE from defense metals’ Wicheeda lake deposit in BC. In:
Proceedings of the 59th Conference of Metallurgists, COM 2020, published by The Canadian
Institute of Mining, Metallurgy and Petroleum, Processing of Critical Materials Symposium. 2020.
3. Verbaan N, Bradley K, Brown J, Mackie S. A review of hydrometallurgial flowsheets considered in
current REE projects. In: Simandl GJ, Neetz M, editors. Symposium on strategic and critical mate-
rials proceedings, November 13–14, 2015. Victoria, British Columbia: British Columbia Ministry
of Energy and Mines, British Columbia Geological Survey Paper 2015–3; 2015.
4. Goode J. Where did that earbud come from? Current rare earth production facilities. In: IMPC
2016: XXVIII International Mineral Processing Congress Proceedings published by The Canadian
Institute of Mining, Metallurgy and Petroleum, REE Symposium. Quebec City, 2016.
5. Goode J. A comparison of acid baking and caustic cracking for REE recovery. In: Proceedings of
the 59th Conference of Metallurgists, COM 2020: The Canadian Institute of Mining, Metallurgy
and Petroleum, Processing of Critical Materials. 2020.
Toward Carbon-Negative Cement and Battery
Materials
Jeremy Ley1 (*), David Dreisinger1 (*), Mike Johnson2, Niels Verbaan2,
and Sridevi Thomas2
1
Negative Emissions Materials Inc., Claymont, DE, USA
Jeremy.ley@nemomaterials.com; David.dreisinger@nemomaterials.com
2
SGS Canada Inc., Lakefield, ON, Canada
Abstract. Decarbonization is the challenge of the century. Carbon-negative materi-

als production and supply chains can be part of the solution. Negative Emissions
Materials, Inc (NEM) has developed a process for production of carbon-negative
nickel and cobalt battery material intermediates using a chloride based hydrometal-
lurgical process route applied to saprolite ores. Amorphous silica, a supplementary
cementitious material that can reduce the clinker content of cement mixes by 30%, is
a coproduct. Magnesium hydroxide, a specialty chemical with several environmental
applications, is the other major coproduct of the NEM process. The NEM process
uses HCl leaching of saprolite to produce the silica residue, and the leachate is neu-
tralized to produce an iron/aluminum/chromium residue as a potential coproduct.
The iron-free solution is further neutralized to produce a mixed hydroxide precipitate
(MHP) of nickel and cobalt followed by a high-purity magnesium hydroxide precipi-
tate. The final solution is a concentrated sodium chloride solution which is directed
to chlor-alkali processing to produce HCl and NaOH for acid leaching and neutral-
ization. The process has advanced quickly through bench and pilot testing and is
moving toward the development of the first large-scale demonstration plant. The pro-
cess and the supporting testing will be presented.
Keywords: Carbon negative · Nickel · Cobalt · Hydrochloric acid · Sodium

hydroxide · Chlor-alkali · Mixed hydroxide precipitation · Silica · Magnesium
hydroxide
1 Introduction
1.1 Negative Emissions Materials
The Grantham Foundation for the Protection of the Environment leads their website
with the statement “Climate change is the greatest challenge humanity has ever faced.
It is the race of our lives.” All over the world we are seeing individuals, corporations,
governments, and investors joining this race. The decarbonization of the transport sec-
tor is in full swing with accelerating uptake of electric and hydrogen powered vehicles.
Power generation from wind and solar sources is increasing, and storage of electricity
to stabilize grid power dependent on renewables is advancing. All of these efforts are
making a difference as the race continues.
https://doi.org/10.1007/978-3-031-17425-4_85
738 J. Ley et al.
The demand for raw materials to enable many of the emerging transformative tech-
nologies is accelerating. Specifically, lithium, nickel, cobalt, and manganese are
required for lithium ion battery manufacturing, and rare-earth elements are needed for
electric vehicle traction motors and wind turbines. Further, when recovering critical
raw materials for future use, the extraction industry must focus on zero waste produc-
tion. All by-products/coproducts should to be considered for use rather than storing in
liquid or solid waste impoundments. Lastly, and perhaps most importantly, recovery of
critical materials should be performed to produce a “carbon-negative” product, i.e., the
best outcome would be to provide for future supply while creating pathways to remove
carbon from the atmosphere.
Negative Emissions Materials (NEM) was founded to develop innovative technolo-
gies for critical materials supply, with by-products to enable an overall “carbon-
negative” outcome. Technology development has focused on treatment of nickel
saprolite ores, olivine, and asbestos tailings materials. These materials all contain mag-
nesium silicate or magnesium hydroxy silicate minerals in addition to nickel, cobalt,
iron, and other minor elements. The vision of NEM is to process these materials for
nickel and cobalt battery material supply, silica product for material substitution in the
cement industry, iron-containing precipitate for iron making or alternate use, and mag-
nesium hydroxide. Magnesium hydroxide from the NEM process is carbon negative
and may be used in the future to sequester atmospheric carbon dioxide using a variety
of processes.
This paper is the first in a series of papers to report progress toward carbon-negative
cement and battery materials.
1.2 Process Flowsheet
The NEM general process flowsheet is shown in Fig. 1. The process involves the
following:
Crushing and Grinding Crushing and grinding in recycle brine solution containing a
variety of chloride salts, including magnesium chloride and sodium chloride. The pur-
pose of recycle of a brine solution is to avoid the addition of water which can only be
managed by evaporation which is expensive in terms of energy.
HCl Leaching This process uses HCl at high strength (typically 30–36% HCl by
weight in water; a typical product from an HCl production facility attached to a chlor-
alkali plant). The raw materials contain a variety of silicate minerals including magne-
sium, iron, nickel, cobalt, and minor impurity elements.
The chemistry therefore comprises the following major reactions:
Mg 2 SiO 4 4HCl 2 MgCl2 SiO2 2H 2 O

Ni 2 SiO 4 4HCl 2 NiCl 2 SiO2 2H 2 O
Fe 2 SiO 4 4HCl 2 FeCl 2 SiO2 2H 2 O
Toward Carbon-Negative Cement and Battery Materials 739
Olivine/Mafic Rock/Saprolite/Asbestos Tailing
Crushing and Grinding
HCl Solution Brine

HCl Leaching
Recycle
Was h Amorphous Silica Residue

Washing of Residue
Water f or Cement Manufacture
NaOH Soln Iron and Aluminum

Precipitation
Was h F e/Al Oxide/Hydroxide

Washing of Residue
Water Precipitate
NaOH S oln Ni/Co Mixed Hydroxide Ni/ Co Products

Production (to battery manufacture)
NaOH S oln Mn Oxide/Hydroxide Manganese Precipitate

NaOCl Soln Precipitation
NaOH Soln Mg Hydroxide

Precipitation
Was h Mg Hydroxide
Washing of Residue
Water Precipitate
Electricity
Chlor-Alkali Plant to
make HCl and NaOH for
Recycle
Fig. 1. The NEM simplified process flowsheet.
Note other minerals present such as iron oxides or aluminum oxides, may also react
with HCl to form additional salts in solution.
FeO OH 3HCl FeCl3 2H 2 O

AlO OH 3HCl AlCl3 2H 2 O
Note that natural minerals are not pure compounds. The minerals may contain a
variety of elements (e.g., Mg, Ni, Co, Fe in one silicate mineral) and may be hydrated
or weathered. Suitable feed materials include the following:
A. Nickel saprolite ores
B. Olivine ores
C. Asbestos ores and tailings.
The conditions for leaching are typically a temperature between 85 and 95 °C to the
boiling point (probably 115 or even higher). Acid addition ranges from 500 to 1000 kg
740 J. Ley et al.
HCl per dry tonne of solid feed and will vary with the chemical composition of the feed.
The brine recycle solution in the flowsheet below ensures that acid leaching is per-
formed with a high total salt content as, for example, NaCl or MgCl2 or both.
The leaching time can vary from 1 to 8 hours. The leaching can be done in a single-
stage or two-stage countercurrent. Single stage means that the acid and ore are added
together and allowed to react at temperature to completion, while two stage means that
fresh ore is contacted with partly reacted solution so as to maximize the consumption
of acid (low terminal acidity) and in the second stage, the partly leached ore (from the
first stage) is contacted with high acid to maximize extraction of Mg/Ni/Co/Fe, etc. The
two-stage process requires an additional solid/liquid separation step to ensure counter-
current movement of solids and liquids.
The products of HCl leaching are a weakly acidic solution containing various chlo-
ride salts and a silica-rich residue recovered as a solid product. This residue is washed
to remove salts and excess acid with fresh water and then directed to cement manufac-
ture where the silica is used as a replacement for other materials (thus lowering the
carbon intensity of cement manufacture) and a strengthener to improve the yield
strength of concrete (high-performance concrete).
Iron and Aluminum Removal The iron and aluminum content in the solution is pre-
cipitated as a mix of oxide and hydroxide solids by raising the pH with NaOH solution.
The NaOH solution is added as a 50% solution and may be diluted with recycle brine
solution for process convenience and enhanced pH control (it may be difficult to control
pH by adding such a strong base). The NaOH neutralizes the excess acid and precipi-
tates Fe/Al and other trivalent cations if present, according to the following reactions:
HCl NaOH NaCl H 2 O

FeCl3 3NaOH FeO OH 3NaCl H 2 O
2 FeCl3 6 NaOH Fe 2 O3 6 NaCl 3H 2 O
AlCl3 3NaOH AlO OH 3NaCl H 2 O
2 AlCl3 6 NaOH Al 2 O3 6 NaCl 3H 2 O
CrCl3 3NaOH CrO OH 3NaCl H 2 O
2CrCl3 6 NaOH Cr2 O3 6 NaCl 3H 2 O
The pH adjustment is conducted with stoichiometric amounts of NaOH. Over-

addition will result in undesirable precipitation of Ni/Co, so careful control of addition
must be maintained. The temperature will be 75 °C to the boiling point. Seed (precipi-
tate) may be recycled to ensure growth of suitable sized particles for enhanced solid/
liquid separation. Precipitation time can be 1–8 hours. NaOH is added progressively
through the precipitation tanks (continuous) so as to enhance precipitation of coarser/
separable precipitates. The product undergoes S/L separation and washing.
The iron/aluminum residue may be treated to form commercial products. This is a
continuing program by NEM.
The iron and aluminum removal process can be performed in a two-stage arrange-
ment to allow recycle of the second-stage precipitate to the leaching section to mini-
mize any nickel and cobalt loss and to maximize the removal of iron, aluminum, and
chromium. Further, if the iron is partly in the ferrous state, a small amount of sodium
hypochlorite may be added to oxidize residual ferrous to the ferric state.
Nickel and Cobalt Recovery Nickel and cobalt are present in solution as NiCl2 and
CoCl2 salts. The recovery of Ni/Co can be done in many ways including the direct pre-
cipitation of mixed hydroxide precipitate (MHP). This can be done directly from the
solution coming from the iron precipitation. Alternately ion exchange loading and elu-
tion with a bis-picolylamine resin can be used to recover an ion exchange eluant con-
taining nickel and cobalt chloride at higher concentration.
In all cases, a solution of sodium hydroxide is added to form the precipitates.
NiCl 2 2 NaOH Ni OH 2 2 NaCl

CoCl2 2 NaOH Co OH 2 2 NaCl
Other metals will also precipitate with the Ni/Co in minor amounts, such as Mn and
Fe (remaining iron in solution). If an excess of sodium hydroxide is added, then mag-
nesium will co-precipitate as magnesium hydroxide. The selectivity of MHP precipita-
tion can be enhanced by using two-stage MHP precipitation. The second-stage
precipitate is recovered and recycled to the first stage or to the discharge from the main
leaching step (where acid is present to redissolve the Ni/Co and other metals from the
second-stage leach).
The mixed hydroxide precipitate is recovered by solid–liquid separation and wash-
ing. A pressure filter is often used with a “squeeze” cycle to minimize the entrained
moisture in the washed MHP cake prior to shipping.
The precipitation is carried out between 25 and 90 °C and terminal pH is in the range
of 5–8. Note that pH measurement is difficult in a strong salt solution, and sodium
hydroxide addition can also be controlled by stoichiometry. The precipitation time is
1–8 hours. Seed recycling is used to maximize particle size and minimize contamina-
tion. The process (as in all steps) is conducted continuously.
Manganese Removal Manganese is an undesirable impurity in magnesium hydrox-
ide, and cannot be selectively precipitated as a hydroxide in the presence of magnesium
hydroxide. Accordingly, oxidation and precipitation are used to remove manganese
from solution.
MnCl2 NaOCl 2 NaOH MnO2 H 2 O 3NaCl
The manganese dioxide precipitate is filtered and washed.

Magnesium hydroxide precipitation The remaining magnesium in solution is pre-
cipitated by addition of NaOH to form Mg(OH)2.
MgCl2 2 NaOH Mg OH 2 2 NaCl
This can be done by adding NaOH to MgCl2 solution or by reversing the order of
addition. The goal is to provide a near complete removal of Mg as Mg(OH)2 from solu-
tion, requiring a near stoichiometric addition of NaOH.
742 J. Ley et al.
Chlor-Alkali Plant The final solution is NaCl and H2O with some minor contami-
nants in solution. This solution is directed to a chlor-alkali plant for manufacture of
NaOH, Cl2, and H2. This involves many steps. The Cl2 and H2 can be burned and water
scrubbed to form strong HCl solution for recycle to leaching.
The excess heat from combustion is recovered as steam and used to evaporate excess
water from solution.
2 Experimental Results
2.1 Raw Materials for Leaching
The three types of raw materials used in this study were obtained from various sources.
Table 1 shows the elemental composition of the raw materials tested. The asbestos tail-
ing was obtained from a tailing deposit in Quebec. Olivine (used as a foundry sand
material) was provided by Essix Resources under the trade name Incast75. The nickel
saprolite sample was provided by SGS Canada from an existing inventory of material.
The nickel and cobalt content of the material increases from asbestos tailing to oliv-
ine to nickel saprolite sample. Iron levels are variable from ~5% for the olivine sample
up to 18% for the saprolite. The magnesium content was highest for olivine at 29.79%
and only 12.2% for nickel saprolite. The silicon content varied over a narrow range of
15.6–19.49% Si. The samples were either used as received or ground to finer size as
required.
2.2 HCl Leaching Test Results
A series of HCl leaching tests were performed to assess the extraction of the key ele-
ments (Ni/Co/Mg/Fe/Al) and the quality of silica residue produced as a product.
Table 1. Composition of raw materials for extraction.

Analysis (%) Asbestos tailing Olivine sample Nickel saprolite sample
Ni 0.239 0.34 1.81
Co 0.011 0.011 0.052
Cu 0.0019 0.002 0.005
Zn 0.0019 0.004 0.02
Fe 7.51 5.13 18.0
Mg 22.2 29.79 12.2
Al 0.5 0.19 2.19
Cr 0.25 0.28 0.78
Mn 0.07 0.08 0.65
Ca 0.26 0.08 0.33
Si 16.5 19.49 15.6
Na 0.06 0.04 0.02
S NA NA 0.02
The leach extractions were highly sensitive to the acid addition, reported as kg HCl/
tonne of material. The basis is kg HCl on a 100% basis and per tonne of dry ore (Fig. 2).
The nickel extraction for the saprolite sample (Fig. 2) approached 100% at ~750 kg
HCl/t of ore. Iron and magnesium extractions were slightly lower. It was technically
feasible to produce ≥75% SiO2 content in the residue at the 750 kg HCl/t addition rate.
The olivine extraction results (Fig. 3) showed very similar extractions of Ni/Mg/Fe,
consistent with the uniform mineralogy of the olivine sample. The acid required to
reach maximum extraction was nearly 1200 kg HCl/t ore due to the more basic charac-
ter of the olivine. The SiO2 grade of the leach residue exceeded 80%.
The results of the leaching of asbestos tailings (Fig. 4) showed very high Ni/Mg/Fe
extractions at +800 kg HCl/t. The SiO2 grade again exceeded 80% in concert with high
base metal extractions.
The batch leaching results showed that a variety of raw materials could be treated
with suitable levels of HCl to maximal extraction of the key metals and production of a
silica-rich residue.
100 100
Residue Silica Grade (%)
95 95
90 90
Extraction (%)
85 85
80 80
75 75
70 70
65 65
60 60
600 700 800 900 1000 600 700 800 900 1000
Acid Addition (kg HCl/t) Acid Addition (kg HCl/t)
Mg Fe Ni SiO2
Fig. 2. Batch leaching results for saprolite. Typical conditions: Temperature 100 °C, 4 h, 87 g/L Mg
(added as MgCl2).
100 100
95 95
90 90
Extraction (%)
85 85
80 80
75 75
70 70
65 65
60 60
900 1100 1300 1500 900 1100 1300 1500
Mg Fe Ni SiO2
Fig. 3. Batch leaching results for olivine. Typical conditions: Temperature 100 °C, 4 h, 87 g/L Mg
(added as MgCl2).
744 J. Ley et al.
100 100

95 95
90 90
Extraction (%)
85 85
80 80
75 75
70 70
65 65
60 60
700 800 900 1000 1100 700 800 900 1000 1100
Mg Fe Ni SiO2
Fig. 4. Batch leaching results for asbestos tailings. Typical conditions: Temperature 100 °C, 4 h,
87 g/L Mg (added as MgCl2).
2.3 Cementitious Properties of the Leach Residues
An extensive study of the cementitious properties of the leach residues was conducted.
The results of this study will be reported in further publications. The leach residues
containing high levels of residual silica were well suited to addition to cement.
2.4 Batch Testing of the Downstream Process Operations
In addition to an extensive series of batch leach tests, the downstream operations were
also tested step by step in a series of further experiments. Table 2 shows the changing
composition of solutions within the sequential unit operations in the NEM flowsheet
applied to a nickel saprolite sample.
The leach solution contains over 3 g/L of Ni and high levels of Mg, Fe, and Al. The
IR solution shows very low terminal Fe and Al concentrations and somewhat reduced
Ni concentration, a consequence of some dilution due to base addition and some co-
precipitation of Ni. The sodium level is increased due to NaOH addition, while
manganese is diluted to 640 mg/L Mn in solution. The MHP (primary) solution shows
very low Fe/Al and significant reduction in Ni. This experiment produced a high-grade
MHP product (+40% Ni on a dry basis). The MnR solution shows an excellent rejection
of Mn from solution (to 0.05 mg/L of Mn) by oxidation and pH adjustment. Further, the
MP precipitation results show very low residual content of Mg, Ni, Fe, Al, and Mn. The
final concentration of Na was 125 g/L Na, corresponding to 318 g/L NaCl. This brine
would undergo further treatment in a conventional chlor-alkali circuit to polish minor
contamination before electrolysis.
Table 2. Solution compositions (mg/L) through downstream unit operations.

Operation Leach IR MHP MnR MP
Mg 69,600 63,000 68,800 62,300 3.1
Ni 3370 2210 300 1.7 0.6
Fe 36,900 3.6 0.3 0.2 0.5
Al 3470 10 0.2 0.2 0.2
Mn 750 640 630 0.05 0.05
Na 24 25,500 27,400 25,200 125,000
Test ID SL2 IR5 MHP1 MnR3 MP5
Leach HCl Leach, IR iron removal, MHP mixed hydroxide precipitation, MnR manganese removal,
MP magnesium hydroxide precipitation
2.5 Continuous Pilot Plant Results
A continuous pilot plant was established at SGS Canada to integrate the key elements
of the process, from ground ore feed to the production of silica residue, iron/aluminum
precipitate, mixed hydroxide of nickel and cobalt, manganese precipitate, and finally
magnesium hydroxide precipitate. The barren solution after magnesium removal is also
a product but in this case for recycle through a chlor-alkali facility for manufacture of
sodium hydroxide and hydrochloric acid supply.
The pilot plant was run on a prepared sample of ground saprolite (Table 1) with
grinding performed in recycle brine solution (Fig. 1). Leaching was performed with
~750–800 kg HCl/t feed material and 95 °C for 10 days total (two periods of 5 days).
The leach slurry was collected and filtered in a pressure filter to recover the silica resi-
due and the leach solution was directed to primary neutralization. A hematite-rich seed
slurry was added to the feed solution as it entered four stages of neutralization with
NaOH solution. The slurry product was thickened and the thickener UF was divided
into seed recycle and final product. The primary neutralization thickener overflow was
sent through secondary neutralization (four stages followed by thickening), where addi-
tional NaOH solution was added. The secondary neutralization thickener UF was recy-
cled to leaching and the OF was directed to MHP production. MHP production was
performed in two stages (primary and secondary), and the second-stage MHP thickener
UF was recycled to leaching. The MHP thickener OF was sent to manganese removal.
This was performed by oxidation with NaOCl and pH adjustment (with caustic addi-
tion) to form a manganese oxide product. Finally, the manganese-free filtrate advanced
to magnesium hydroxide precipitation with sodium hydroxide addition (Fig. 5).
Saprolite Leaching The pilot plant operation was divided into 18 periods (each period
was ~12 hours) over which data were collected and mass balances calculated. Figure 6
shows metal extraction version time over these periods. The extraction of Ni in the
saprolite leaching process was generally in the range of 96–99%. The Fe extraction was
slightly lower and the Mg lower again. The acid addition rate was varied during the run
and the changing extraction results are reflected by an increase or decrease in acid addi-
tion. The reported extractions are calculated by the Si-tie method where Si is assumed
to be insoluble.
746 J. Ley et al.
Recycle
Brine HCl NaOH
Saprolite Feed Grind L Primary Neut S Fe/Al

Leach (L)
(batch) S (PN) L
Precip.
Leach Residue
NaOH NaOH
NaOH Secondary L L Secondary S

Primary MHP
Neut (SN) S S MHP L
MHP Product
NaOH
NaOH
Mn Removal L Magnesium L to chlor-alkali
NaOCl (MnR) S Precip (MP) S pre-treatment
Mn Precip. Mg Precip.
Fig. 5. Overall flowsheet of the continuous pilot plant.
100 100
Si-tie Extraction(%)
95 95
90 90
85 85
80 80
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Fe Ni Mg
Fig. 6 Saprolite leaching results over the 18 time periods of the pilot plant operation.
Primary Neutralization The primary neutralization results are shown in Figs. 7 and
8. The iron removal results are consistent at nearly 100% efficiency. The aluminum
removal results showed an increase toward the end of the pilot plant with a commensu-
rate rise in nickel co-precipitation. The nickel in the primary neutralization residue is
lost from the circuit, so better control is required to avoid a nickel loss while still remov-
ing sufficient aluminum in this step. Aluminum precipitation is an indicator of potential
nickel loss. It is important to not over-add NaOH at this stage, otherwise nickel loss will
increase. The solid assay from the primary neutralization circuit showed a plateau of
approximately 80% Fe2O3 with 5.5% Al2O3 and 4% Cl. The iron and aluminum precipi-
tates are hydrated and the chloride in the residue is likely due to formation of hydroxy
chloride precipitates of iron and aluminum.
Secondary Neutralization The secondary neutralization results (Figs. 9 and 10) show
excellent removal of iron and aluminum. However, if aluminum removal is too effi-
Calculated Precipitation (%), Fe, Al 100 5.0
Calculated Precipitation (%), Ni

80 4.0
60 3.0
40 2.0
20 1.0
0 0.0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Fe Al Ni
Fig. 7. Primary neutralization calculated precipitation.
100 7
Solids Assay (%), Fe2O3
Solids Assay (%), Al2O3, Cl

6
80
5
60 4
40 3
2
20
1
0 0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Fe2O3 Al2O3 Cl
Fig. 8. Primary neutralization solids assay.
100 100
Calculated Precipitation (%), Fe, Al
Calculated Precipitation (%), Ni
95 90
90 80
85 70
80 60
75 50
70 40
65 30
60 20
55 10
50 0
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Fe Al Ni
Fig. 9. Secondary neutralization precipitation efficiency.

748 J. Ley et al.
25 1.4
Solids Assay (%), Fe, Al, Ni
Solids Assay (%), Mg, Mn

1.2
20
1.0
15 0.8
10 0.6
0.4
5
0.2
0 0.0
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Fe Al Ni Mg Mn
Fig. 10. Secondary neutralization solids assay.
cient, the precipitation of nickel increases. This is not a problem in the sense that the
secondary neutralization residue is recycled and nickel is releached. However, the
nickel needs to move downstream to MHP precipitation, and therefore, nickel co-
precipitation and aluminum removal have to be kept in balance to avoid excessive
nickel buildup in the Leach Secondary Neutralization part of the circuit. This was an
important learning from the pilot plant.
Primary and Secondary Mixed Hydroxide Precipitation The primary mixed
hydroxide precipitation results (Figs. 11 and 12) showed that nickel could be precipi-
tated to form high-grade MHP at as high as ~40% Ni on a dry basis (periods 5–8). The
results also show the need for effective control of NaOH addition. During periods 9–18,
excess NaOH was added, leading to increased precipitation of Mg and some Mn. The
stability of this circuit is impacted by the upstream process steps and especially by the
recycling of nickel and cobalt back to leach. Further, the measurement of pH in the
strong brine solution as a measure of control is difficult and impacts the control of the
MHP circuit. Future pilot plant operations will have improved pH electrode sensors and
control.
The secondary mixed hydroxide precipitation results (Figs. 13 and 14) generally
show the effective capture of residual nickel arriving from the primary mixed hydroxide
circuit. Again the correct dosage and control of NaOH addition is a critical issue mov-
ing forward with the process scaleup. Under-addition of NaOH will result in loss of
soluble Ni and Co to the manganese removal circuit, while over-addition will cause the
precipitation of Mg and Mn.
Manganese Precipitation The manganese removal circuit (Figs. 15 and 16) was sta-
ble and yielded very high levels of manganese removal by effective oxidation and pH
adjustment. Small levels of magnesium precipitation occurred through the operation,
unavoidable due to the elevated pH used for manganese removal. The operation between
period 5 and 10 showed some nickel in the manganese precipitate, due to incomplete
nickel removal in the secondary mixed hydroxide precipitation circuit. Beyond this
point (periods 11 and beyond), the nickel in solution in the feed to manganese precipita-
tion was very low and, hence, the content of the manganese precipitate was very low.
Calculated Precipitation (%), Ni, Co 100 50
Calculated Precipitation (%), Mg, Mn

90 45
80 40
70 35
60 30
50 25
40 20
30 15
20 10
10 5
0 0
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Ni Co Mg Mn
Fig. 11. Primary mixed hydroxide precipitation efficiency.
50 14
Solids Assay (%), Co, Mg, Mn

Solids Assay (%), Ni
12
40
10
30 8
20 6
4
10
2
0 0
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Ni Co Mg Mn
Fig. 12. Primary mixed hydroxide precipitation solid assay.
100 30
Calculated Precipitation (%), Mg, Mn
Calculated Precipitation (%), Ni, Co
90
80 25
70 20
60
50 15
40
30 10
20 5
10
0 0
5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Ni Co Mg Mn
Fig. 13. Secondary mixed hydroxide precipitation efficiency.
Magnesium Precipitation The magnesium precipitation results were excellent

(Table 3). The magnesium precipitation circuit operated toward the end of the 10-day
pilot plant run. The key impurity elements were generally very low with the exception
of chloride, which likely formed magnesium hydroxy chloride precipitates under
startup conditions. Any of the other di- or trivalent metals present in the feed to magne-
sium precipitation will co-precipitate with the magnesium. Magnesium precipitation
750 J. Ley et al.
50 2.5
Solids Assay (%), Ni, Mg, Mn
Solids Assay (%), Co

40 2.0
30 1.5
20 1.0
10 0.5
0 0.0
5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Ni Mg Mn Co
Fig. 14. Secondary mixed hydroxide precipitation solids assay.
100 2.0
Calculated Precipitation (%), Mn
Calculated Precipitation (%), Mg

95 1.5
90 1.0
85 0.5
80 0.0
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Mn Mg
Fig. 15. Manganese precipitation efficiency.
40 20
Solids Assay (%), Mg, Mn
Solids Assay (%), Ni, Co
30 15
20 10
10 5
0 0
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Time period
Mg Mn Ni Co
Fig. 16. Manganese precipitation solids assay.
Table 3. Magnesium precipitate analysis.

Chemical Analysis (%)
Period Fe Al Cl Ni Co Mn MgO
16 <0.01 0.02 4.4 <0.01 <0.01 <0.01 61.3
17 0.01 0.04 0.2 0.11 <0.01 0.01 63.1
efficiency was ~100%. The brine formed as a product from this process step is ideal as
a feed to brine softening ahead of the chlor-alkali plant operation.
3 Conclusions
Following an extensive bench test program, the Negative Emissions Materials process
was piloted on a sample of nickel saprolite ore for a total of 10 days. All the key metrics
were achieved in the pilot plant operation:
• The leach extractions of nickel and cobalt were in the range of 96–99% in the pri-
mary HCl leach.
• The primary neutralization circuit removed iron and aluminum effectively with min-
imal co-precipitation of nickel and cobalt.
• The secondary neutralization circuit was effective at polishing residual iron and alu-
minum content from the solution prior to mixed hydroxide precipitation.
• The primary mixed hydroxide precipitation produced product grading up to 40% Ni
on a dry basis. Under controlled conditions, co-precipitation of magnesium and
manganese could be avoided. The testing highlighted the need to develop improved
measurement of pH in the strong brine solutions used in this process so as to enhance
control and selectivity of the key process steps.
• The secondary mixed hydroxide precipitation was effective at precipitating residual
value metals.
• The manganese removal circuit utilizing oxidation and pH adjustment for precipita-
tion was outstanding in performance with virtually 100% removal of manganese
from solution.
• The magnesium precipitation process product was high grade and low in metallic
impurities. The magnesium precipitation process is dependent on all of the upstream
processes to produce a suitable precipitate product. The brine from the magnesium
precipitation process was virtually free of any impurities and suitable as a source of
NaCl brine to proceed to brine softening and chlor-alkali processing to regenerate
HCl and NaOH for the process.
The silica leach residue from batch leaching of saprolite, olivine, and asbestos tail-
ing was evaluated as an additive to cement. The results confirmed that the leach resi-
dues were reactive and suitable for cement making. The testing of the cementitious
properties of the residue will be reported in a further publication.
The overall recovery of nickel and cobalt from the process is expected to be in the
range of +95%. The MHP product is suitable for further post-processing to produce
battery material precursor materials to support the rapidly increasing demand in the
electric vehicle space. The magnesium hydroxide product from the process is an ideal
material to support decarbonization. Future papers will provide further comment and
data on this part of the work and process development.
Acknowledgments The authors thank the management of Negative Emissions Materials and SGS
Canada for permission to publish this paper.
The Demonstration of the Cuprion Process
for Recovery of Nickel, Cobalt, Copper,
and Manganese from Seabed Nodules
David Dreisinger1 (*), Niels Verbaan2, and Marlon Canizares2

1
Dreisinger Consulting Inc., Delta, BC, Canada
2
Abstract. Dreisinger Consulting Inc. (DCI) has evaluated the application of the Cuprion
process for processing of seabed nodules from the Clarion Clipperton Zone in the Pacific
Ocean. A program of bench and pilot plant testing was conducted to demonstrate the
reductive leaching of ground nodules in a seawater-based ammonia–ammonium carbon-
ate leach solution. The batch leach testing confirmed that extraction of copper, nickel, and
cobalt exceeded 90%, leaving a manganese carbonate residue for manganese recovery.
The Cuprion leach process was piloted over a 5-day period in a series of continuous
overflow reactors with closed loop solution recycle. Pilot metal extractions confirmed or
exceeded the lab leach testwork results and led to increased understanding of the Cuprion
process. The pilot plant solutions containing extracted metals were treated by oxidation
and solvent extraction to demonstrate co-extraction of nickel and copper. The pilot plant
residue contained manganese carbonate formed by the reductive leaching chemistry. This
residue may be treated to extract and recovery manganese as EMM along with residual
amounts of copper, nickel, and cobalt. The results of the batch and pilot testing are
presented.
Keywords: Seabed nodules · Nickel · Cobalt · Copper · Manganese · Cuprion
1 Introduction
1.1 Processing of Nodules
The Clarion-Clipperton Zone (CCZ) seabed nodule deposits occur in the deep ocean
between Mexico and Hawaii and contain significant quantities of nickel, cobalt, copper,
manganese, and various critical metals. Table 1 shows the general composition of CCZ
manganese nodules [4, 5].
The recovery of metals from seabed deposits requires mineral extraction from the
sea floor to the ocean surface and transport to a land-based processing facility. DCI
have considered a number of processing technologies for the seabed nodules. Dames
and Moore [3] summarized a number of historical processes including the following:
https://doi.org/10.1007/978-3-031-17425-4_86
754 D. Dreisinger et al.
Table 1. Composition of CCZ Manganese Nodule [4, 5].

Element Composition Element Composition
Al 2.9% Mn 25.4%
As 159 g/t Mo 520 g/t
Ca 1.7% Ni 1.28%
Co 0.24% S (SO4) 1.84%
Cu 1.02% Si 7.6%
Fe 6.9% Zn 0.14%
Ammoniacal systems
Gas reduction and ammoniacal leaching
Cuprion atmospheric leaching
Chloride systems
Reduction and HCl leach
HCl reduction roast and acid leach
Segregation roast
Molten salt chlorination
Sulfate Systems
High-temperature and high-pressure acid (H2SO4) leach
Smelting and H2SO4 leach
H2SO4 reduction leach
Reduction roast and H2SO4 leach
Sulfate roast
After an extensive process review, the Cuprion process was selected by DCI for
further investigation.
1.2 Cuprion Process
The Cuprion process is a development of the Kennecott Copper Corporation (KCC) and
is often referred to as the KCC – Cuprion Process [1–6]. The process involves a reduc-
tive ammonia–ammonium carbonate leach of the manganese nodules. The reducing gas
is a producer gas containing carbon monoxide, hydrogen, and small amounts of carbon
dioxide, oxygen, and nitrogen. The carbon monoxide (and hydrogen) reduces cupric
ion (hence the name Cuprion for the process) to the cuprous state which is stabilized by
the ammonia–ammonium carbonate solution composition. The cuprous ion then reacts
with manganese dioxide to from the manganous ion (Mn2+) which promptly precipi-
tates as manganese carbonate. The nickel, copper, and cobalt contained in the manga-
nese nodules is leached into solution as the manganese dioxide is converted to
manganese carbonate.
Agarwal et al. [1] has summarized the chemistry of the leach as follows:
The Demonstration of the Cuprion Process for Recovery of Nickel, Cobalt, Copper… 755
Manganese dioxide reduction:
MnO2 2Cu NH 3 2 2 NH 4 2 CO3 MnCO3 2Cu NH 3 4

2
CO32 2H 2 O
Carbon monoxide reduction:
2Cu NH 3 4 CO CO32 2H 2 O 2Cu NH 3 2 2 NH 4 2 CO3

2
Net overall reaction:
MnO2 CO MnCO3
The leaching of the base metals results in the dissolution of the ammines of copper,
nickel and cobalt according to the following equations:
CuO NH 4 2 CO3 2 NH 3 Cu NH 3 4 CO3 H 2 O

CoO NH 4 2 CO3 3NH 3 Co NH 3 5 CO3 H 2 O
NiO NH 4 2 CO3 4 NH 3 Ni NH 3 6 CO3 H 2 O
The mixed ammine leach solution is then treated by solvent extraction with a
hydroxyoxime to recover copper and nickel metal (by electrowinning). Cobalt (oxi-
dized to +3 state after leaching) and molybdenum are not extracted during the solvent
extraction process. Cobalt and molybdenum are recovered from the SX raffinate as a
mixed byproduct from the steam – stripping and recovery of ammonia and carbon
dioxide.
The considerations behind development of the Cuprion process were highlighted by
Agarwal [1].
The unique characteristics of manganese nodule ore must be taken into account in
considering economic approaches to its metallurgical processing. These include the
following:
• The metal values are finely disseminated in the oxide matrix of the ore. Consequently,
the ore is not amenable to physical beneficiation.
• The ore is very porous with a porosity of 50–60%. When drained, it still contains
30–40% free moisture with another 10–15% chemically combined water.
As a result of the above, pyrometallurgical treatment of nodules requires smelting
the entire ore to recover only three high-value metals, principally nickel, copper and
cobalt. In addition, substantial energy must be expended to remove large quantities of
water. Pyrometallurgical approaches are, therefore, not well suited to nodule processing.
The situation improves somewhat if the business objective is altered to produce manga-
nese as an additional product.
It was concluded that an economically superior metallurgical process for nodules
would have the following characteristics:
Ambient temperature,
Low pressure,
Operable in the presence of sea water,
Inexpensive or recyclable reagents,
Low energy consumption,
High and selective recovery of nickel, copper, cobalt and molybdenum,
Non-corrosive to materials of construction,
Low toxicity reagents, and
Acceptable environmental impact.
The process is very mild with 50 °C leaching of ground nodules with atmospheric
pressure carbon monoxide reduction (Fig. 1).
Manganese may be recovered from the Cuprion residue as electrolytic manganese
metal (EMM). The Cuprion residue, containing manganese as manganese carbonate,
may be processed by the EMM process to produce high-purity EMM (selenium free).
Importantly, the acid leaching of manganese carbonate provides supplemental recovery
of unleached nickel, cobalt, copper, and other metals. The Cuprion-EMM process com-
bination therefore will provide high overall recovery of the key metals in the seabed
nodules.
The following equations describe the EMM process steps using Cuprion leach resi-
due as feed (Fig. 2):
Manganese leaching
Fig. 1. The Cuprion process for manganese nodule processing.

Fig. 2. Cuprion residue (MnCO3) processing to manganese metal.
MnCO3 H 2 SO 4 MnSO 4 H 2 O CO2 g
Solution purification
MeSO4 NH 4 2 S MS NH 4 2 SO 4 , where Me Cu, Ni, Co, Zn, Cd
Manganese electrowinning
MnSO 4 H 2 O Mn ½ O2 g H 2 SO 4
2 Cuprion Process Testing
2.1 Polymetallic Nodule Samples
A total of 22.6 kg of crushed and dried polymetallic nodules (PMN) was available for
testing. A sample was cut and assayed (Table 2). The as-received nodules had a K80 of
250 μm. In some tests, the nodules were ground to smaller size.
Table 2. CCZ manganese nodule analysis.

Element Composition Element Composition
Al 2.62% Mn 27%
As 114 g/t Mo 490 g/t
Ca 1.57% Ni 1.20%
Co 0.14% S (SO4) 0.09%
Cu 1.06% Zn 0.151%
Fe 6.3%
2.2 Batch Cuprion Leach Tests
The Cuprion process was studied in batch to look separately at cuprous reduction of
manganese dioxide and carbon monoxide reduction of cupric copper. Because of the
reductive nature of the leach and how rapidly the reduced Cu(I) ions can be oxidized by
the oxygen in the air, some of the testwork was focused on developing reliable and
appropriate testwork procedures. The leach reactor used for all the tests was a well-
sealed reactor where the introduction of air was minimized either by the constant addi-
tion of CO gas and/or by keeping the ports in the reactor closed at all times. During
nodules addition, as required by each test, one port was opened to add the nodules and
it took approximately 5–10 seconds to add the nodules, which was the only time the
reactor was opened.
A picture of the setup (test L15) is shown in Fig. 3. The setup experienced small
variations as it progressed throughout the program; the changes were for the most part
in appearance not in function, and the location of the pH and ORP probes were changed
a couple of times as the authors learned how to best optimize the available space and to
avoid false reading due to the buildup of CO gas in the electrode fittings. Liquor sam-
ples were collected from the recirculating line; this was extremely important to mini-
mize the time the reactor was opened to the atmosphere as well as to make sure the
sample collected for Cu(I) was not exposed to air; failure to do this earlier in the pro-
gram caused the Cu1+ assays to be unreliable and/or meaningless as the oxidation of
Cu(I) is a much faster process than the reduction of Cu(II) to Cu(I). Several procedural
changes (related to the test control, nodule addition, filter cake washing, etc.) were also
made throughout the test program. These are discussed in the sections below.
Tests Without CO(g) Addition Four scoping tests were carried out early in the pro-
gram to study the effect of the lixiviant composition on the nodules leaching without
Cu(I) regeneration (L1–L4). Conditions for these tests are presented in Table 3.
The underlying objective for these tests was to obtain preliminary data on the recov-
ery of Ni, Co, and Cu, varying the amount of CO2 in solution as well as the composition
of the lixiviant. These tests were not controlled in any way and the ORP was left to
fluctuate as a result of the solution composition and nodules addition.
The lixiviant for these four tests and all the subsequent tests was prepared by dis-
solving ammonium carbonate and ammonium hydroxide in synthetic sea water. The
reagent addition order was usually: sea water, ammonium carbonate, ammonium
hydroxide, and Ni/Co/Zn/Cu carbonates. The test procedure was adjusted from test to
Fig. 3. Cuprion reactor setup.
Table 3. Lixiviant and test conditions for leach tests L1 to L4.

CO
Stoich NH3 CO2 Ni mL/
Cu + g/L Addition g/L g/L g/L Nodule Addition CuCl addition min
L1 20 100% 100 20 0 Gradual, 5 min At ambient T 0
intervals
L2 20 100% 100 50 0 Gradual, 5 min At 50 C, just prior to 0
intervals nodules
intervals nodules
intervals nodules
test to improve the performance of each test. The lixiviant prepared using sea water and
containing the indicated amount of NH3 and CO2 was prepared ahead of time and stored
in a sealed container. This lixiviant was added to the reactor and depending on the test
the indicated amount of CuCl was added to the lixiviant before nodules addition. For L1
the CuCl was added at room temperature, and the lixiviant was then heated to the target
temperature of 50 °C. For L2, L3, and L4 the lixiviant was heated to 50 °C and then the
CuCl was added. This change was incorporated since it was observed that during L1,
the oxidation of the Cu(I) occurred during the heating period. The oxidation of Cu(I) to
Cu(II) effectively removes the ability to leach MnO2 and hence the ability to leach Cu/
Ni/Co.
ORP was used as a proxy to get a sense of the amount of Cu(I) in solution; although
a clear correlation could not be obtained between ORP and Cu1+/Cu2+, it was observed
early in these tests that ORP values were higher than around −100 to 0 mV, indicated
that most of the copper was in the Cu(II) form and that reduction of manganese dioxide
will not occur. Nodules containing approximately 66% solids by weight were added at
a rate of 7.97 gwet every 5 min for tests L1 to L3 and at a rate of 10.63 gwet every 20 min
for test L4. The amount of copper added as Cu(I) was calculated based on the total
nodule addition, i.e., 95.63 gwet. For L1 to L3 and 31.88 gwet for L4.
Metal recoveries were calculated based on residue assays and solids feed assays
either by direct mass balance or by manganese or (later) iron tie. Due to the high copper
content of the synthetic lixiviant and relatively small quantity of additional (leached)
copper units, it was difficult to base metal extraction calculations on liquor assays. This
effect was not present for nickel, cobalt, or molybdenum, which were not normally
added to the synthetic lixiviant. Therefore, liquor assays could be used to gauge extrac-
tion of those elements, but not of copper.
Nickel and cobalt recoveries for tests L1 and L2 were low. In L1 the nickel leach
efficiency was 59.1%, while the cobalt recovery was only 32.8%. Meanwhile, in L2 the
nickel leach efficiency was slightly higher at 68.2% as well as for cobalt at 50.2%.
Copper leach efficiency in both tests was negative as the residue contained more copper
than the feed. It was postulated that this increase in copper was a product of insufficient
washing of soluble copper units or of copper precipitation during the washing steps.
The copper concentration in the lixiviant was 20 g/L and it would not take much liquor
or copper precipitation in the filtered residue to contaminate the leach residue to
increase copper levels in the residue.
In test L3, metal recoveries were negative as well. The lixiviant for this test was
prepared using carbonates of nickel, cobalt, and copper, and it was suspected at the time
that either these did not completely dissolve during the lixiviant preparation or had
precipitated during the test and therefore led to inflated metal levels in the residue and
consequently led to negative calculated extraction values.
In test L4, nickel and cobalt recoveries were 96.6% and 90.4%, respectively, much
higher than in tests L1 and L2. In this test the amount of Cu(I) available was 3 times the
amount for either L1 or L2. The ORP trend showed that the ORP remained low for the
duration of the test. This was the first test that showed that lower ORP values (indicat-
ing relatively high concentration of Cu(I) ions) are needed if high recoveries are desired
(Table 4).
Overall, these tests could not be considered true Cuprion leach tests as Cu(I) ions
were not regenerated during the tests, and the amount of copper in solution was much
higher than what is practically feasible. However, a better understanding was gained of
the parameters that needed optimization before a “real” Cuprion leach could be carried
out.
Cu(I) oxidation occurs very rapidly, and any Cu(I) addition should be reserved to the
last minute before the nodules are added to the reactor. It was discovered that the air
contained in the reactor overhead space was sufficient to oxidize the Cu(I) and therefore
lead to poor extraction of metals. Proper washing of the leach residue was also extremely
important especially since residue analysis was the only practical way of quantifying
the leach efficiencies.
Table 4. Leach residue assays and metal extraction for L1–L4.

Test Residue Assay (g/t) Extraction (%)
Cu Ni Co Mo Cu Ni Co Mo
Feed 10,600 12,000 1400 490
L1 12,200 4700 900 60 −20.3 59.1 32.8 87.2
L2 9700 3300 600 60 −5.7 68.2 50.5 85.9
L3 19,900 27,000 22,200 80 −124.7 −169.4 −1798 80.5
L4 8000 300 100 10 −1.8 96.6 90.4 97.2

CO
Cu+ Stoich NH3 CO2 Ni mL/
g/L Addition g/L g/L g/L Nodule Addition CuCl addition min
L5 20 300% 100 20 0 All at once At 50 C, just prior to ~140
nodules
L6 5 200% 100 20 0 All at once At 50 C, just prior to ~140
nodules
L7 5 200%* 100 20 0 Gradual, 20 min At 50 C, just prior to ~140
intervals nodules
Tests with CO(g) Addition L5–L7 test conditions are presented in Table 5. L5 showed
high nickel and cobalt extraction (Table 6) and a decreasing ORP over time. By 60 min,
the lixiviant was nearly colorless indicating that the cupric ammine species formed by
oxidation of cuprous was nearly completely reduced by the CO(g) addition. It was
found that the solution color and ORP measurements were very useful for controlling
the Cuprion process. Residue washing was believed to again impact copper recovery.
For L6 it was decided to reduce the amount of Cu (I) in solution to 5 g/L to facilitate
the washing of the leach residue. In this test, the stoichiometric amount of copper to
manganese in the nodules was reduced to 200% excess and the nodules as in L5 were
all added at once at the beginning of the test.
The nickel and cobalt recoveries were lower than in test L5 (92.3% and 76.4%,
respectively) and it was not immediately clear why this was the case since despite the
differences in test procedure nothing seemed out of the ordinary. Copper recovery was
76.6% calculated using the mass and chemical analysis of the residue. Despite that, for
the first time in the program positive copper recoveries were recorded. However, it was
suspected that still some copper was left behind after washing leading to slightly sup-
pressed copper extraction values.
Test L7 was completed similarly to test L6 with a few additional modifications. A
new washing procedure was implemented where the leach residue was repulp washed
with 1 L of ammonia/ammonium carbonate solution (100 g/L NH3 and 20 g/L CO2) and
then displacement washed with 1 L of DI water. The second modification in the proce-
dure was that nodules were added at 20 min intervals (similarly to L4) for the duration
of the test and the CO gas flow was stopped if the ORP reached below −270 mV and
re-started if it went above −200 mV.
Table 6. Leach residue assays and metal extractions for L5–L7.

Feed 10,600 12,000 1400 490
L5 8700 600 100 10 −1 93.9 91.2 97.5
L6 2100 780 280 10 76.6 92.3 76.4 97.6
L7 6000 3900 430 70 44.7 68.3 70.0 86.1
The addition of extra nodules was intended to increase the net percent solids in the
leach slurry. It was expected that the constant addition of CO would maintain a certain
level of Cu(I) in solution allowing the continued leaching of more nodules. ORP, pH,
and temperature were constantly (every second) monitored for the duration of the test.
The regular addition of nodules caused the ORP to climb throughout the test, indicating
that the Cu(I) was being consumed and the CO addition was not able to reduce enough
Cu(II) to keep up with the leaching. It is possible that once the Cu(I) concentration
decreased beyond a certain value, the reduction slowed down too much to keep up.
There is a note in one of the original Cuprion patents [2] to this effect.
…Thus, it would be expected that as the cupric ion became high in value, its reduction with
carbon monoxide would proceed at a relatively rapid rate; however, this is not the case. Indeed,
it has been discovered that under the conditions proposed in the present process, an increase in
the amount of cupric ions has little effect on this reaction rate. Furthermore, it has been discov-
ered that in order to maintain a high cuprous ion concentration, the cuprous ion level can never
be allowed to go to an extremely low level.
Later, the author(s) continue:

Thus, a species that should be present in order to form cuprous ions from cupric ions is the
cuprous ion itself as well as the cupric ion. Because of this autocatalytic nature of the cuprous
ion regeneration, the amount of cuprous ions should never be allowed to go to zero. In short, no
matter what the chemistry of the regeneration of cuprous ion might be, applicants have discov-
ered that in order to maintain enough cuprous ions at all stages of the Cuprion process by
regeneration from cupric ions, the cuprous ion concentration must be maintained at a fairly high
level…
Unfortunately, this was not the case in the L7, where the regular addition of nodules
consumed the Cu(I) ions earlier in the test and the ORP continued climbing until the
very end of the test.
Tests with ORP Control and Optimization Tests Tests L8–L23 were intended to
demonstrate effective Cuprion control by careful monitoring of ORP. When the ORP
returned to < −280 mV, sufficient cuprous was present to allow the next addition nod-
ules without depleting the cuprous. The ORP always increased upon addition of nod-
ules due to conversion of cuprous to cupric but then recovered as the CO reacted with
the cupric and restored a high cuprous/cupric ratio.
Table 7 shows the test conditions for L8 to L23. Table 8 summarizes the recoveries
obtained in each test.
The results of this series of tests provided the following insights to the Cuprion pro-
cess chemistry:
1. Tests L8 and L9 were repeat tests. L8 suffered mechanical issues. L9 provided clear
success in achieving +90% extraction of Cu, Ni, Co, and Mo. L9 demonstrated the
importance of controlling the ORP within a range to ensure that cuprous was
always available to reduce the manganese dioxide in the nodule feed and very
importantly, to catalyze the further reduction of cupric to cuprous. Some unreacted
black grit was also identified in the filter cake from L8/L9 that may indicate the
need to reduce particle size to ensure high extraction.
2. The feed to L10 was screened to remove coarse particles. The Cu, Ni, Co, and Mo
recoveries were in line with those of L9 while the leach residue was free of black,
unreacted particles.

CO
Cu+ Stoich NH3 CO2 Ni mL/
g/L Addition g/L g/L g/L Nodule Addition min Objectives
L8 10 400%* 100 20 0 11 g, when ~90 Effect of gradual nodule
ORP < −280 mV addition, controlled by
ORP
L9 10 400%* 100 20 0 11 g, when ~90 Duplicate (reproduce L8)
ORP < −280 mV
L10 5 457%* 100 20 0 6 g, when ~90 Effect of screening off o/s
ORP < −280 mV and reduced cuprous level
L11 5 457%* 100 20 0 5 g, when ~ 90 Feed 100% passing 75
ORP < −280 mV micron
L12 5 457%* 100 20 0 5 g, when ~90 Effect of fine grind
ORP < −280 mV
L13 5 457%* 100 20 0 5 g, when 150 Effect of high CO flow
ORP < −280 mV
L14 10 913%* 100 20 0 5 g, when ~90 Effect of high cuprous
ORP < −280 mV levels in lixiviant
L15 5 457% 100 50 0 5 g, when ~90 Effect of high CO2 levels
ORP < −280 mV in lixiviant
L16 5 457% 100 20 0 5 g, when ~50 Effect of low CO flow
ORP < −280 mV
L17 5 457% 100 20 0 5 g, when ~90 Effect of using coarse feed
ORP < −280 mV
L18 5 457% 100 20 6 5 g, when ~90 Effect of simulated
ORP < −280 mV lixiviant (NiCoMo)
L19 5 457% 100 20 7 5 g, when ~90 Repeat of L18
ORP < −280 mV
L20 5 457% 100 20 0 5 g, when ~90 Continuous Nodules
ORP < −280 mV addition (slurry)
L21 5 457% 100 15 6 5 g, when ~ 90 Effect of Low CO2 levels
ORP < −280 mV in lixiviant with Synthetic
sol.
ORP < −280 mV lixiviant (NiCoMo)
ORP < −280 mV lixiviant (Low NiCoMo)
Table 8. Leach residue assays and metal extractions L8 to L23.

Feed 10,600 12,000 1400 490
L8 4800 590 <100 10 67.6 96.5 >94.9 98.5
L9 1200 280 <100 10 90.8 98.1 >94.2 98.3
L10 770 100 <100 10 90.3 98.9 >90.5 97.3
L11 760 130 <100 10 91.3 98.7 >91.4 97.5
L12 620 180 <100 10 92.8 98.1 >91.2 97.5
L13 780 130 <100 10 90.9 98.7 >91.2 97.5
L14 1300 110 <100 10 85.2 98.9 >91.4 97.5
L15 420 <100 <100 10 96.0 >99.1 >92.8 97.9
L16 900 <100 <100 10 89.6 >99.0 >91.3 97.5
L17 1500 540 140 10 77.0 92.7 83.8 96.7
L18 410 510 1900 30 94.9 94.4 NA 9.19
L19 330 490 920 20 95.9 94.6 12.7 94.6
L20 1600 200 100 10 80.7 97.9 90.9 97.4
L21 730 56,600 7200 140 NA NA NA NA
L23 1000 720 710 10 90.7 94.1 49.9 98.0
3. In test L11, the nodules were ground to 100% passing 75 microns and in test L12
to 100% passing 37 microns. The extraction of Cu, Ni, and Co was in line with that
of earlier tests showing little material impact of finer grinding.
4. Test L13 used a higher CO(g) flow (150 mL/min versus 90 mL/min in prior tests).
Metal extractions remained very similar to previous tests but frequency of nodule
addition increased, indicating a more rapid cupric reduction process. This may
indicate a mass transfer limitation for CO(g) under the conditions studied.
5. Test L14 was conducted using 10 g/L of total copper in the lixiviant. The Ni and Co
extractions were similar to the previous test, but copper extraction was lower
(98.9% Ni, 91.4% Co, and 85.2% Cu), possibly due to incomplete washing of the
final residue. Nodule addition frequency increased, confirming that higher total
copper/cuprous levels in solution are advantageous.
6. In test L15, the concentration of CO2 in solution was increased to 50 g/L and the
total copper concentration was decreased back to 5 g/L. The extraction of copper,
nickel, and cobalt were among the highest recorded in the test program at 96.0%,
99.1%, and 92.8%, respectively. However, the nodule addition rate was much
slower, indicating the reduction of cupric was slower under the test conditions.
7. Test L16 used a slower CO(g) addition rate and the extractions were in line with
L9. The measured pH was slightly higher in L16 (c/w L9) and this resulted in a
faster nodule addition rate, in line with faster cuprous regeneration at higher pH.
8. Test L17 was a duplicate of L11 but using the original feed with a coarser particle
size. All other conditions were maintained constant. The recoveries of copper,
nickel, and cobalt were lower than in L11 at 77.0%, 95.7%, and 83.8%,
respectively.
9. Test L18 used a synthetic lixiviant containing 6.69 g/L Ni, 0.513 g/L Co, 0.67 g/L
Zn, and 0.28 g/L Mo in addition to 5 g/L total Cu. The extraction of Cu and Ni were
93.8 and 93.2% but the Co extraction was low. This confirmed that there was a limit
to cobalt extraction at high steady-state concentration of cobalt in solution. This
appears to be due to the relatively weaker complexation of cobalt by ammonia
compared with nickel. Test L18 had a higher pH than earlier tests. This again
showed a more rapid nodule addition rate confirming the relationship between pH
and rate of cuprous regeneration. Test L19 was a repeat of L18 and confirmed low
cobalt extraction.
10. Test L20 was a test conducted in preparation for piloting the Cuprion leach process.
Continuous addition of a slurry containing 25% nodules in seawater was used. The
speed of addition of nodules was controlled to maintain low ORP. The Cu, Ni, and
Co extractions were 80.7%, 97.9%, and 90.9%, respectively. Copper extraction
may again have been impacted by poor washing of the residue. It was demonstrated
during this test that while maintaining all the other parameters constant, i.e., pH,
solution composition, CO(g) flow, etc., the nodule addition rate could be used to
modulate the reduction reaction in the reactor. This was an important finding and
provided important lessons and confidence in advance of the Cuprion process mini-
pilot plant for leaching.
11. Tests L21 and L22 were intended to revisit the use of a synthetic lixiviant contain-
ing nickel, cobalt, zinc, and molybdenum. L21 also had a lower CO2 content
(15 g/L), while L22 had 20 g/L CO2. In both cases the synthetic solution was not
stable, and nickel and cobalt precipitated from the solution during the test. Copper
recoveries were ~90% in both cases.
12. Test L23 was similar to L22, but the concentration of elements in the synthetic
solution was half of the previous tests (L18, L19, L21, and L22). In this case the
recoveries for copper, nickel, and cobalt were 90.7%, 94.1%, and 49.9% respec-
tively. Table 7 shows the relationship between initial cobalt concentration and
cobalt recovery. Although only three data points were available, it was clear that
high steady-state cobalt concentrations led to lower ultimate cobalt extraction
(Table 9).
2.3 Mini-Pilot Plant Cuprion Leach Testing
Circuit Description and Operating Targets The circuit consisted of seven reactors in
cascade named R1 and L1 to L6, and two surge reactors collecting the overflow from
the thickener. Leaching of the nodules occurred in reactors L1 to L6. The lixiviant from
R1 was pumped at a predetermined rate to L1 where it was mixed with the slurry com-
ing from the nodule feed slurry tank. Carbon monoxide gas (CO) was sparged in this
tank. The leach slurry overflowed to L2 where more nodules were added from the nod-
ule’s slurry tank. Nodules were added into L1, L2, L3, and L4. Reactors L5 and L6
provided extra residence time to allow for the continued leaching of the nodules as well
as the reduction of any cupric remaining in solution. CO gas was sparged to all reactors
Table 9. Initial cobalt concentration vs. final extraction.
Co Concentration (mg/L) Cuprion Extraction

513 ~0.0
461 12.7
255 49.9
28 +90%
in the cascade and it was manually adjusted based on the reactor’s ORP (mV) reading.
The off-gas from all reactors was directed to a gas scrubber consisting of dilute NaOH.
The slurry from L6 was pumped to a thickener that was also sealed to avoid air
intake. The solids settled in the thickener and overflow was directed (pumped) to one of
two surge tanks at the top of the circuit. Once one of these surge tanks was full, the line
coming from the thickener was directed to the other surge tank and the content of the
full reactor was quickly pumped to the R1 reactor closing the loop. The surge tanks
were where ammonia or ammonium carbonate solution was added based on titrations
performed at the bench; also based on these titrations, ammonium hydroxide was added
to L1, L3, and/or L5. A photograph of the circuit during operations is shown below. The
thickener, CO cylinders, and nodules slurry tank are outside the picture frame. The pilot
plant was operated for 5 days.
The operating targets were set as follows:
• Temperature: 50 °C using heating mantles
• ORP: −280 mV and −400 mV
• pH: not controlled but expected range between 9.5 and 10.0
• Nodule addition rate: Starting rate of 2.4 g of dry nodules per minute with increase
during the campaign
• Residence time: the solution retention time in each reactor was ~1.5 hours
• CO gas flow: The initial flow rate was set at 360 mL/min but adjusted as required to
control the ORP within range (Fig. 4).
Cuprion Leach Mini-Pilot Plant Results The temperature in the leach tanks was
held steady to ~50 °C for the duration of the pilot plant. The pH generally varied
between 9.25 and 10.0. The ORP values were generally well controlled with some
upsets due to equipment failure or excess addition of nodules for a short period of time.
However, the Cuprion chemistry was resilient and recovered from any upset. It was
necessary to add ammonia solution to the leach reactors continuously to counteract the
volatilization of ammonia by the venting of excess CO(g). This is a feature of small-
scale testing and not indicative of need for additional ammonia addition on the full-
scale application.
The chemical analysis of samples taken regularly from the L6 reactor is presented in
Figs. 5 and 6. Because the solution in the circuit was recycled, the concentration of met-
als was expected to constantly increase until it reached a limit. The concentration of
nickel, cobalt, molybdenum, and zinc increased steadily throughout the 5 days of oper-
ations (Fig. 5) showing no signs of reaching a limit. The copper trend in Fig. 5 is not
Fig. 4. Pilot plant setup.
truly indicative of the copper recoveries since there was a large background of copper
in solution and small changes due to additional copper leaching do not materially
change the concentration. The assays of the leach residue from L6 are presented in
Fig. 7.
Copper extractions were initially lower than expected (Fig. 8). It was suspected that
this was caused by inadequate washing of the residues. A revised washing procedure
was implemented (greater amounts of washing water), and the results showed a clear
improvement in copper recoveries. Samples submitted prior to this date were later
recalled from the assay lab and re-washed and submitted for assay again, confirming
that incomplete washing contributed to suppressed extraction values. Nickel extraction
was around 99% throughout the run.
At the time of shutdown, samples were collected from reactors L1 to L6, the solids
washed, and submitted for analyses, from which metal extractions were calculated
(Fig. 9). While, nickel and cobalt extraction showed a flat profile, copper showed an
increase toward the end of the circuit.
Cobalt and zinc recoveries decreased steadily in the second half of the campaign. In
Fig. 10, the recoveries of cobalt and zinc are plotted against the equilibrium concentra-
tion of these metals in solution. The recovery of cobalt started to decrease after the
cobalt in solution reached between 30 and 40 mg/L. Insufficient samples were analyzed
for zinc at the beginning of the plant, and it is difficult to determine if recoveries were
Fig. 5. L6 liquor assay trends for all metals.
Fig. 6. L6 liquor assay trends for Ni, Co, Zn, and Mo.
Fig. 7. L6 solids assays.
Fig. 8. Metal extraction trends.

Fig. 9. Metal recoveries in reactors L1 to L6.
higher when zinc levels in solution were near zero. At concentrations higher than
30 mg/L, the zinc recovery suffered significantly, even more so than cobalt. While zinc
is not an extremely valuable metal, any zinc (and any other metals) left in the Cuprion
leach residue will need to be dealt with in the manganese circuit prior to manganese
electrowinning.
2.4 Metal Recovery from Solution
The recovery of nickel and copper using solvent extraction was demonstrated using
LIX 84I extractant. Cobalt was pre-oxidized to the cobaltic state prior to solvent extrac-
tion testing. Cobalt is not extracted by LIX 84I after pre-oxidation. Limited bench-scale
testing confirmed that ammonia could be scrubbed from the loaded solvent and that
nickel could be selectively stripped for separate recovery.
Fig. 10. Zinc and cobalt extraction vs. metal concentration in solution.
3 Conclusions
Following an extensive bench test program, The Cuprion process was operated success-
fully in a continuous pilot plant for 5 days leading to high metallurgical recoveries. The
pilot plant produced sufficient leach solution for copper/cobalt/nickel recovery testing
and a leach residue for manganese recovery testing.
The Cuprion bench testing showed the importance of good ORP control and proved
that if sufficient cuprous copper is present, leaching proceeds quickly. The bench test-
ing also showed that while the leaching reaction may be quick, the cupric to cuprous
regeneration reaction is slow and is the rate limiting step within the Cuprion process.
The bench testing also shows that if the system is depleted of cuprous copper, leaching
of nodules stops, and at the same time it has been shown that it is difficult to regenerate
cuprous from cupric copper without any initial cuprous copper present.
The fundamental control parameters of this process (pH, ORP, temperature, and
nodule addition) were demonstrated and tested successfully in a mini-pilot plant. The
ORP (used as a proxy to assess the reduction of copper) was controlled successfully by
the addition of CO gas and careful controlled addition of nodules. It was demonstrated
that good reductive conditions could be obtained in the reactors if these two parameters
are monitored and controlled carefully. Despite some shortcomings in the reactor
design, the pilot plant ran trouble free for 5 days. There were no shutdowns or major
stoppages. Any problems that arose during the operations were dealt with without com-
promising the stability of the operation. Metal extractions were in the expected range,
especially after a more stringent washing regime of the solids was implemented.
Metal recoveries (specifically copper and nickel) measured during the pilot plant
were in the expected range, especially after a more stringent washing regime of the
solids was implemented. The bench Cuprion bench tests provided some preliminary
indications that cobalt recoveries would suffer with increasing cobalt levels in the PLS.
This is why the pilot plant started up without cobalt present in the lixiviant. The behav-
ior observed in the bench tests was confirmed in the pilot plant, which demonstrated
that cobalt recoveries suffer when cobalt levels in solution increased.
Acknowledgments The authors thank the management of DCI and SGS Canada for permission to
publish this paper.
References
1. Agarwal JC, Barner HE, Beecher N, Davies DS, Kust RN. Kennecott process for recovery of cop-
per, nickel, cobalt and molybdenum from ocean nodules. Min Eng (NY). 1979;31:1704–9.
2. Barner HE, Kust RN, Cox RP. Elevated pressure operation in the Cuprion process. US. Patent
3,988,416, 26 Oct 1976.
3. Dames, Moore; E.I.C. Corporation. Description of manganese nodule processing activities for
environmental studies. Vol. III. Processing systems technical analysis. Rockville: U.S. Dep.
Commerce – NOAA, Office of Marine Minerals; 1977. p. 540. NTIS PB 274912(set).
4. Haynes BW, Law SL, Barron DC, Kramer GW, Maeda R, Magyar MJ. Pacific manganese nodules:
characterization and processing. USBM Bulletin 679. 1985.
5. Haynes BW, Law SL, Barron DC, Kramer GW, Maeda R, Magyar MJ. Laboratory processing
and characterization of waste materials from manganese nodules. USBM Report of Investigation
8938. 1985.
6. Szabo LJ. Recovery of metal values from manganese deep sea nodules using ammoniacal cuprous
leach solutions. U.S. patent 3,983,017, 28 Sept 1976.
Magnet Rare Earth Production Must Double in Ten
Years. Where Will Production Come From?
J. R. Goode1 (*)
1
J.R. Goode and Associates, Ontario, Toronto, ON, Canada
jrgoode@sympatico.ca
Abstract. Decarbonization to limit climate change requires replacing fossil fuels with
renewable energy resources, electrification of vehicles, and several other strategies. The
technologies that must be deployed consume critical materials, particularly the rare earth
elements (REEs) praseodymium, neodymium, terbium, and dysprosium. It is generally
estimated that over the next 10 years, the consumption of these elements will double and
that a deficit of about 70,000 t/a of PrNd oxide will occur unless new production is
brought on line. To put this into context, a deposit similar to Mountain Pass or Mt. Weld
and associated processing facilities must be commissioned almost every year over the
next decade. This presentation reviews the characteristics of REE sources, especially ore
deposits, and attempts to identify those sources with the potential to deliver the required
REE production.
Decarbonization to limit climate change requires replacing fossil fuels with renewable
energy resources, electrification of vehicles, and several other strategies. The
technologies that must be deployed consume critical materials, particularly the rare
earth elements (REEs) praseodymium, neodymium, terbium, and dysprosium. It is
generally estimated that over the next 10 years, the consumption of these elements will
double and that a deficit of about 70,000 t/a of PrNd oxide will occur unless new
production is brought on line. To put this into context, a deposit similar to Mountain
Pass or Mt. Weld and associated processing facilities must be commissioned almost
every year over the next decade. This presentation reviews the characteristics of REE
sources, especially ore deposits, and attempts to identify those sources with the potential
to deliver the required REE production.
Present sources of REE include:
Ionic Clay Deposits. These low-grade (<0.1% rare earth oxides [REOs]), free-digging,
shallow (usually <20 m deep) deposits, typically containing perhaps 1 Mt of miner-
alized material, were formed by intensive long-term weathering of REE-bearing
rock. The REEs are sorbed on clay minerals formed during weathering and can be
readily extracted using cheap reagents, sometimes using in situ leaching. Ionic clay
deposits in Southern China and Myanmar account for most of the world’s heavy
REE (HREE) production. An ionic clay project in Brazil is under construction and
others in Southern Africa, Chile, and Australia are under development.
https://doi.org/10.1007/978-3-031-17425-4_87
774 J. R. Goode
Hard Rock Deposits. Carbonatite deposits are the main source of light REEs (LREEs)
at this time. Examples of this deposit type are Mountain Pass and Bayan Obo with
grades of approximately 8% REO and resources in the tens to hundreds of million
tonnes. Weathered carbonatites, such as Mt. Weld in Australia, Tomtor in Russia,
and Maboumine in Gabon can have enhanced REE grades. Many of the deposits
now under investigation are of this type.
Alkaline/peralkaline deposits are not presently exploited at a significant scale but are of
some interest because they are often high in HREEs (e.g., Strange Lake and the
Nechalacho Basal zone in Canada and Dubbo in Australia). Alkaline/peralkaline
deposits often pose challenges due to complex REE mineralogy and the presence of
Zr, Nb, and other elements. Grades are generally <2% REO and resources are in the
tens to hundreds of millions of tonnes.
Hydrothermal vein deposits are in small-scale production at Gakara in Burundi, pilot
production at Browns Range in Australia and under development at Steenkampskraal
in South Africa. Mineralization is often dominated by monazite, although Browns
Range is dominantly xenotime.
Mineral Sands. Monazite recovered from heavy mineral sand deposits was the main
source of rare earths for many years. Today, heavy mineral sands are mined for
ilmenite, rutile, and zircon and until recently, the contained REE minerals were sent
to waste or stockpiled. By-product REE are now being exploited by IREL (India),
Iluka Resources (Eneabba project), and Chemours/Energy Fuels – with other devel-
opments expected. Mineral sand deposits can contain billions of tonnes of material.
When processed, the sands might yield a heavy mineral concentrate (HMC) amount-
ing to 3% mass. In favourable circumstances, monazite, containing 60% REO, might
constitute 4% of the HMC leading to an REO content in the resource of 0.07% REO.
Potential REE sources include:
Deep Ocean Muds. Muds in the Eastern South Pacific can contain 0.1–0.23% REEs, of
which about 20% can be HREEs. Central North Pacific muds can contain 0.04–0.1%
REEs, again with approximately 20% HREEs. Muds off Marcus Island contain as
much as 0.6% REEs, including 0032% Dy. The depth of these REE deposits is typi-
cally 5000 m, which makes for mining challenges, but reported REE extractions can
be 80% using a mild sulphuric acid leach. None of these deposits is exploited at the
present time, but they are often posited as possible solutions to the REE shortage.
Environmental questions surround these opportunities.
REE Recovery from Waste. Many present-day waste streams carry REEs and there are
large quantities of such waste in existing tailings deposits. Examples include wastes
from coal extraction and combustion, red mud from bauxite processing, phospho-
gypsum wastes (e.g., Phalaborwa, RSA), tailings from iron ore recovery operations
(Pea Ridge and Mineville, USA, Bayan Obo, China), tailings from uranium opera-
tions (Elliot Lake, Canada, Mary Kathleen, Australia), and tailings from REE opera-
tions (Lynas and Bayan Obo). These potential sources of REEs are not currently
exploited, although in the USA there is much interest in REE extraction from wastes,
especially coal-sourced wastes.
REE Recycling. All the REEs now in consumer or industrial products will eventually
reach their end-of-life (EOL) and be discarded or enter a recycling system. Examples
Magnet Rare Earth Production Must Double in Ten Years. Where Will Production… 775
include fluorescent lamps, hard disc drives, wind turbine generators, and automobile
traction motors. At this time, <10% of EOL material is being recycled because recy-
cling processes are complex, and recent economics are not very favourable. This
situation is changing as magnet-REE prices are increased. Besides EOL wastes,
magnet-making processes generate as much as 30% of the REE input as production
waste including machining swarf. China practices extensive recycling of such waste,
other countries less so.
It is not sufficient to look at the in situ grade of an REE deposit expressed as the
percentage of total rare earth oxide (TREO) or total rare earth element (TREE). It is
necessary to also consider the distribution of the individual REEs and the value of co-
or by-products, as well as the presence of undesirable elements. In some cases,
by-products appear to confer a considerable increase to the in situ value of the REO. For
example, Tomtor ore increases from $48/kg to $66/kg of in situ REO; Dubbo value
increases even more significantly, from $54/kg to $123/kg. However, the cost of recov-
ering by- or co-products can be very high and the net value of co- or by-products could
be very low or even negative.
The percentage recovery of REO and the co- or by-products can have a profound
effect on the credit due to co- or by-products. Exact data are sparse, but the recovery of
Fe2O3 into iron concentrate from Bayan Obo ore is about 70% while that of REO is
somewhere between 10% and 20%. Thus, the in situ ore might have REO with a value
of $30/kg contained REO which gets increased to $32/kg when co-product Fe content
is added, but the value of the REO in the recovered REO is $30/kg plus $10/kg credit
for Fe for a total of $40/kg REO in the product.
Something else to be aware of when considering the in situ REE value of projects is
the recovery from ore to a marketable product. Some REE producers accept quite low
recoveries to allow for the production of high-grade concentrates that permit lower-cost
hydrometallurgical operations. For example, only 60–70% of the REEs in the ore mined
at Mt. Weld and Mountain Pass reports to the final products.
REE recovery from the ionic clays varies considerably. Testwork on Serra Verde ore
indicated 40–60% recovery of PrNd from ore to product, Koppamurra averaged 45%,
Makuutu about 75%, and Tantalus approximately 80%.
Non-technical issues are as important as technical issues. We refer to such things as
a social license to operate, remote or poor access, climate, country risk, the savoir faire
of the company directors and management, and the level of project development (scop-
ing study, preliminary feasibility study, feasibility study, basic engineering, construc-
tion). All of these factors, in conjunction with the properties of the REE deposit, will
impact the odds of success.
This presentation (available from jrgoode@sympatico.ca) will discuss the capability
of the present producers to maintain or expand their production of magnet REEs. It will
also review REE projects that are at an advanced stage of development or in construc-
tion and those projects that are less advanced but might contribute to REE before the
end of this decade. The results are worrisome and prompt a brief discussion of
solutions.
Having sufficient mining and primary treatment facilities to produce a plentiful sup-
ply of mixed REE products does not satisfy the demand for magnet REE. There must
also be sufficient REE separation, metal and alloy production, and magnet manufactur-
ing capacity. These factors will also be discussed.
Extraction of Nickel, Cobalt, and Scandium
from Low-Grade Philippine Laterite Ores Using
Atmospheric Leaching with Reduction Pre-Treatment
Edith Andrea D. Lucy1 (*) , Geleni Shalaine P. Bello1 , Chris G. Arnejo1 ,

Christine P. Naira1, Ven Jordan Diputado1 , and Juancho Pablo S. Calvez1
1
Mines and Geoscience Bureau, Diliman, Quezon City, Philippines
edithandrea.lucy@mgb.gov.ph
Abstract. Methods of processing low-grade laterite ores resulting in high purity leaching
products are currently being explored due to the rapid depletion of high-grade ores. In this
study, reduction pre-treatment is conducted prior to the atmospheric leaching of limonitic
laterite ore to produce a nickel-cobalt-scandium-rich leach solution and iron-rich mag-
netic residues. The leaching of pre-treated ore resulted in higher recovery of Ni, Co, and
Sc when compared to the yield from the leaching of untreated ore. Effects of reduction
temperature and time, lixiviant type, pH, leaching temperature and time, and solid-to-
liquid ratio were also investigated using 2k–p Fractional Factorial Design of Resolution
IV. Analysis of Variance showed that all the factors have significant effects on the leaching
behavior of the following elements, with reduction temperature, type of lixiviant and pH
level being the most significant in the recovery of Ni, Co, Sc, and Fe while s/l ratio and
leaching temperature for the recovery of SiO2. Recovery values from the screening exper-
iments ranged from 54.2–94.6% Ni, 32.4–90.1% Co, 0.82–79.5% Sc, 1.6–86.7% Fe, and
0.00–41.7% SiO2.
Keywords: Nickel laterite · Reduction pre-treatment · Atmospheric leaching ·

Selective reduction · Fractional factorial design
1 Introduction
Nickel laterites ore is commonly found and mined in tropical countries like Indonesia,
the Philippines, and Cuba. These ore deposits account for 73% of the world’s nickel
resources and 40% of nickel production. Such ore deposits occur in three layers: limo-
nite, smectite, and saprolite. Limonite and smectite ores have high Fe content, 40%
total Fe or greater. Saprolites, on the other hand, have high magnesium (Mg) and low
Fe contents, 10–20% Mg and 10–25% Fe.
As the world continues to propel itself to modern age technology and strive to reduce
greenhouse gas emissions, the demand for a handful of key metals such as Ni, Co, and
REEs, which within nickel laterite ores are mostly found in limonite zones, has been
https://doi.org/10.1007/978-3-031-17425-4_88
778 E. A. D. Lucy et al.
consistently rising. Limonite ore which contains low Ni, high Fe, and Sc (>50 ppm)
concentrations is suitable for high-pressure acid leaching (HPAL) process. However, in
the Philippines, Coral Bay Nickel Corporation and Taganito HPAL Nickel Corporation
are utilizing HPAL processing technology. Due to high capital investment, maintenance
costs, and energy requirements, 29 of the 31 mining companies in the country are
involved only in the direct shipment of ores. Furthermore, the two processing compa-
nies only regard mixed Ni-Co sulfide precipitate as their main product and generally
treat Fe as a waste, despite high concentrations in the ore feed.
Due to these considerations, there is an increasing interest in developing alternative
processes such as atmospheric leaching (AL) and mixed hydroxide precipitate (MHP)
production, since no autoclave, hydrogen sulfide handling, and injector are required in
these methods, thus, reducing the associated process, capital, and maintenance costs.
However, the main issues that hinder AL to be put into practice are its high acid con-
sumption and the difficulty of the subsequent processing of the resulting leach pulp
such as solid-liquid separation due to the formation of silica gel [1]. This leach liquor
also contains significant concentrations of soluble Fe. Thus, in this study, reduction pre-
treatment is conducted prior to leaching which transforms initially soluble Fe in the
feed into an insoluble, magnetic form of Fe and selectively leaches Ni and Co from ore
using acid [2].
2.1 Material
Limonitic nickel laterite ore samples used in the study were collected from LNL
Archipelago Inc. (LAI), in Sta Cruz, Zambales, Philippines. The samples were air-dried
for seven (7) days to remove excess hydration, and were then crushed, ground, and
screened to obtain an 80% passing size of 149 microns (100 mesh).
2.2 Experimental Procedures
Reduction Roasting The pellets were prepared using the mixture of ore, Ipil-ipil char-
coal, and bentonite clay at a 1: 0.15: 0.015 ratio and screened to achieve a pellet diam-
eter range of 10–12.5 mm. For each run of reduction roasting, 400 grams of the pellets
were loaded to the rotary kiln furnace using different setting combinations to produce
reduced ores with different reduction degrees.
Acidic Leaching The leaching experiments were performed at atmospheric pressure
in 1-liter two-neck Pyrex boiling flasks. The mixture was agitated using a magnetic stir-
rer at 400 rpm and was heated on a hot plate. For each run of acidic leaching, 100 grams
of reduced samples was fed in the setup while maintaining the corresponding parameter
combinations. After each experiment, the slurry was filtered and the obtained residue
was rinsed with distilled water, placed in the drying oven, and then pulverized up to
Extraction of Nickel, Cobalt, and Scandium from Low-Grade Philippine Laterite Ores… 779
Table 1. Factor levels for the reduction roasting-leaching study (2k–p Fractional Factorial – Resolution
IV).
Level
Factor −1 1
A = Reduction time, min 30 60
B = Reduction temperature, °C 700 800
C = Type of lixiviant HNO3 H2SO4
D = pH 2.00–2.25 1.00–1.25
E = Leaching time, min 240 360
F = Leaching temperature, °C 50 90
G = Percent solids 20 40
−200 mesh. These samples were prepared for chemical analysis to determine the %
extraction of nickel, cobalt, scandium, iron, and silica.
2.3 Analytical Method
The chemical analysis of the sample was carried out using Atomic Absorption
Spectroscopy (AAS) for the nickel and cobalt, Titration for the iron, Gravimetry for the
silica in the Chemical Laboratory Services Section, Mines and Geosciences Bureau
(CLSS-MGB), and Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-
OES) for the scandium.
2.4 Experimental Design and Statistical Analysis
The effects of varying reduction temperature and time, type of lixiviant, pH level,
leaching temperature and time, and s/l ratio were investigated using a Two-level (2k–p)
Fractional Factorial Design of Resolution IV at a 95% significance level. Analysis of
Variance is used to assess the main effects of the factors and the interactions among
them. A total of 32 runs corresponding to the seven (7) variables of the 2VII-II fractional
factorial design with low and high levels, were performed, as shown in Table 1.
3.1 Characterization of Untreated and Reduced Laterite Ore
From the results of X-Ray diffraction analysis, it was found that the untreated laterite
ore sample mainly consisted of goethite (FeO(OH)), as shown in Fig. 3. This consti-
tutes its high iron content, ranging up to 45.8%, presented in Table 2, along with the
other abundant elements in the sample. The presence of these elements was also
Table 2. Chemical analysis of the LAI ore.

Sample Ni, % Co, % Fe, % Sc, ppm SiO2,
A 0.69 0.09 45.8 72.0 3.54
(Untreated ore)
B 0.75 0.09 47.5 80.0 6.09
(700 °C, 30 min)
C 0.78 0.10 49.0 88.0 5.89
(700 °C, 60 min)
D 0.74 0.11 51.7 95.0 6.58
(800 °C, 30 min)
E 0.77 0.11 51.1 94.0 6.33
(800 °C, 60 min)
Fig. 1. SEM images of the laterite ore from LAI taken at 5000× magnification.
Fig. 2. (L-R) Quantitative elementary maps showing the occurrences of (a) iron, (b) nickel, (c) cobalt,
(d) silicon, (e) oxygen, (f) chromium, (g) aluminum, and (h) sulfur on the laterite ore from LAI and the
corresponding EDX spectrum of the SEM image.
confirmed in the Energy-dispersive X-ray (EDX) spectroscopy and Field Emission

Scanning Electron Microscope (FE-SEM) analyses of the LAI ore using Dual Beam
Helios Nanolab 600i at the Advanced Device and Materials Testing Laboratory
(ADMATEL), where it was notably seen as inherently porous, as referred to in Figs. 1
and 2, respectively.
The XRD patterns, as seen in Fig. 3, also show the successive transformation of the
untreated ore at different reduction roasting conditions. Upon reduction roasting at
Fig. 3. XRD Analyses of the (a) untreated raw ore, (b) pre-treated ore reduced at 700 °C for 30 min-
utes, (c) pre-treated ore reduced at 700 °C for 60 minutes, (d) pre-treated ore reduced at 800 °C for
30 minutes, and (e) pre-treated ore reduced at 800 °C for 60 minutes.
700 °C, the goethite phases of the untreated laterite ore were entirely diminished due to
its reduction by the carbon monoxide that evolved from the oxidation of the reduction
agent (Ipil-ipil charcoal), while peaks of magnetite (Fe3O4), were intensified. When the
roasting temperature was increased to 800 °C, a portion of goethite transformed into
magnetite and then wustite. Furthermore, prolonging the roasting time from 30 to
60 minutes showed a slight increase in such transformation. This suggested that after a
certain duration of roasting time, the reduction rate decreases due to the excessive
thickness of the product iron layer, resulting in greater resistance to diffusion of carbon
and reducing gas to the surface of unreduced iron oxide [3].
Table 3. %Recoveries of comparative runs of the AL with and without reduction pre-treatment.
Atmospheric leaching with reduction Atmospheric leaching without reduction
Element pre-treatment pre-treatment
Ni 59.2 19.3
Co 75.8 41.4
Sc 47.8 37.1
Fe 26.3 19.1
Fig. 4. (L-R) Pregnant leach solutions of AL with reduction pre-treatment and without reduction
pre-treatment.
3.2 Efficacy of Reduction Pre-treatment on the Extraction of Nickel, Cobalt

and Scandium from Laterite Ore
Efficacy of reduction pre-treatment of laterite ore on the extraction of Ni, Co, and Sc
was determined by comparing the %recovery obtained between leaching untreated ore
and reduced ore (treated at 700 °C for 40 minutes) at 20 wt% solids, 85–90 °C, 360 min-
utes, and experimental acid consumption of 386.4 kg H2SO4 per ton of ore, as shown in
Table 3. AL with reduction pre-treatment obtained higher recoveries of nickel, cobalt,
and scandium than the AL without reduction pre-treatment, consequently validating the
feasibility of the process treatment. Significant improvement in the extraction of key
metals may be observed once optimization experiments are conducted.
Furthermore, as shown in Fig. 4, the pregnant leach solutions of the AL of reduced
reduction pre-treatment exhibited a green clear solution which suggests the presence of
Ni2+ and Fe2+ while AL of untreated ore retained the earthy color of the leached feed.
3.3 Influence of Reduction Pre-treatment and Leaching Factors
This study employed reduction pretreatment process mainly to obtain maximum recov-
eries of nickel, cobalt, and scandium, while minimizing the recovery of iron and silica,
as these both pose handling problems in the leaching and subsequent downstream pro-
cesses. The matrix of the design of experiments and the results obtained in the different
treatment combinations are listed in Table 4. Influences of factors such as reduction
Table 4. Recoveries of nickel, cobalt, scandium, iron, and silica (2k–p Fractional Factorial – Resolution
IV).
%Recovery
Sample mark A B C D E F F Ni Co Sc Fe SiO2
LAI-A01.1 −1 −1 −1 −1 −1 1 1 55.8 54.6 3.51 1.68 4.76
LAI-B02.1 1 −1 −1 −1 −1 −1 −1 58.4 42.4 5.35 5.73 9.67
LAI-C03.1 −1 1 −1 −1 −1 −1 −1 67.4 70.0 28.3 30.5 0.00
LAI-D04.1 1 1 −1 −1 −1 1 1 70.0 64.0 27.7 6.92 2.86
LAI-A05.1 −1 −1 1 −1 −1 −1 1 58.8 57.2 7.15 22.8 2.91
LAI-B06.1 1 −1 1 −1 −1 1 −1 60.2 53.0 11.5 29.5 17.0
LAI-C07.1 −1 1 1 −1 −1 1 −1 82.6 73.9 39.4 41.9 0.00
LAI-D08.1 1 1 1 −1 −1 −1 1 75.7 70.3 28.1 43.7 37.6
LAI-A09.1 −1 −1 −1 1 −1 −1 −1 58.2 58.3 0.82 6.98 3.53
LAI-B10.1 1 −1 −1 1 −1 1 1 68.7 51.1 7.72 30.6 8.16
LAI-C11.1 −1 1 −1 1 −1 1 1 71.7 68.2 16.4 20.9 18.7
LAI-D12.1 1 1 −1 1 −1 −1 −1 81.1 69.9 24.1 28.7 33.0
LAI-A13.1 −1 −1 1 1 −1 1 −1 65.1 64.8 15.1 29.2 17.6
LAI-B14.1 1 −1 1 1 −1 −1 1 61.4 47.7 8.47 19.2 20.3
LAI-C15.1 −1 1 1 1 −1 −1 1 92.5 88.5 63.1 72.4 29.0
LAI-D16.1 1 1 1 1 −1 1 −1 90.0 82.0 49.6 58.5 0.00
LAI-A01.2 −1 −1 −1 −1 1 1 −1 54.2 53.3 2.43 8.22 8.40
LAI-B02.2 1 −1 −1 −1 1 −1 1 65.3 32.5 2.51 15.3 0.78
LAI-C03.2 −1 1 −1 −1 1 −1 1 71.1 74.3 32.7 24.3 0.00
LAI-D04.2 1 1 −1 −1 1 1 −1 59.6 62.6 13.2 8.55 0.00
LAI-A05.2 −1 −1 1 −1 1 −1 −1 56.7 55.3 5.77 20.1 0.49
LAI-B06.2 1 −1 1 −1 1 1 1 69.8 54.8 8.71 28.2 0.00
LAI-C07.2 −1 1 1 −1 1 1 1 82.5 71.4 32.9 39.6 6.10
LAI-D08.2 1 1 1 −1 1 −1 −1 81.7 73.8 33.8 40.2 41.7
LAI-A09.2 −1 −1 −1 1 1 −1 1 60.0 59.0 1.75 12.2 2.00
LAI-B10.2 1 −1 −1 1 1 1 −1 64.1 52.9 12.5 12.6 0.00
LAI-C11.2 −1 1 −1 1 1 1 −1 61.9 68.2 18.3 21.1 0.00
LAI-D12.2 1 1 −1 1 1 −1 1 77.1 68.6 24.7 26.6 30.8
LAI-A13.2 −1 −1 1 1 1 1 1 58.8 64.4 15.0 33.5 20.2
LAI-B14.2 1 −1 1 1 1 −1 −1 62.0 49.8 16.5 26.7 11.6
LAI-C15.2 −1 1 1 1 1 −1 −1 94.6 90.1 79.6 86.7 0.00
LAI-D16.2 1 1 1 1 1 1 1 89.8 88.0 64.7 72.4 0.00
temperature and time, type of lixiviant, pH level, leaching temperature and time, and s/l
ratio were evaluated, through the analysis of variance. Table 5 shows the summary sta-
tistics for the different models that can be applied to the experimental results.
The analysis of variance for the 2k–p Fractional Factorial Design shows that the
responses, namely recoveries of nickel, cobalt, scandium, iron, and silica fit the linear
model (p-value < α = 0.05), which suggests the significance of the main effects of the
reduction pre-treatment and leaching factors. Meanwhile, adding the two-factor inter-
actions (2FI) terms is also found to significantly improve the respective models of the
%recoveries of all the responses, while the addition of the three-factor interaction (3FI)
terms is only significant for the recovery of nickel. In addition, the S values, for all the
Table 5. Summary of the ANALYSIS OF VARIANCE and model of the %recoveries of nickel, cobalt,
scandium, iron, and silica (α = 0.05).
Ni Co Fe Sc SiO2
Model, p-value 0.00 0.0000 0.0000 0.0000 0.0020
Linear, p-value 0.00 0.0000 0.0000 0.0000 0.0000
2FI, p-value 0.001 0.0030 0.0000 0.0000 0.0490
3FI, p-value 0.048 0.1830 0.6790 0.223 0.2950
S 5.18 5.9479 4.0089 5.680 7.5010
R2 94.5% 98.2% 96.2% 96.0% 81.7%
Fig. 5. Pareto charts of the standardized effect on recoveries of nickel, cobalt, scandium, iron, and
silica.
analyses, which are measured in the units of the response variable, %, are low, which
means that the data values are close to the fitted values; the R2 values are also more than
80%. These two indicators imply that the responses have a good fit with the model and
that the generated model equations are representative of the system under the given
experimental domain [4]. All these further confirm the validity of the statistical model
and the influences of the factors on the %recoveries at the significance level of 95%.
The Pareto charts, as seen in Fig. 5, show the standardized effect of main factors and
their interactions in descending order. The reference line in the chart indicates which
effects are statistically significant. The reduction temperature (B), type of lixiviant (C)
and pH level (D) have the most significant effect on metal extraction of nickel, cobalt,
scandium, and iron while the leaching temperature (F) and s/l ratio (G) for the extrac-
tion of silica. Therefore, the experimental run with a treatment combination which set
the factors B, C and D at a high level (LAI-C15.2 and LAI-C15.2) had the highest
extraction of key metals.
Effect of Reduction Temperature and Time The results indicate that reduction tem-
perature has the highest significant effect on metal extraction due to the high probability
of oxygen forming bonds with iron than other key elements such as nickel in the reduc-
tion process, thus the metallization of these oxides is increased. This increase improves
the kinetics of their extractability when subjected to leaching [5].
Effect of pH Level Sufficient acid addition is required to destroy the mineral inclu-
sions in the ore to achieve the complete leaching of minerals. Comparing the runs with
pH levels, 1.00–1.25 and 2.00–2.25, higher extraction values of the metals can be
observed in the former pH range both for the nitric and sulfuric acids. However, optimi-
zation of this factor must be done to prevent or lessen iron and silica dissolution.
Effect of Type of Lixiviant Extraction values of nickel, cobalt, scandium, and iron
show similar dissolution characteristics, as these are mainly entrapped in hematite and
goethite minerals. Higher extraction values of nickel, cobalt, and scandium were
observed when sulfuric acid was used, as compared with nitric acid. Alongside this, a
higher co-extraction rate of the impurity metal, iron, was observed.
Effect of Solid-to-Liquid Ratio It is observed that the increase in the s/l ratio contrib-
uted negatively to the extraction of all the metals since a high amount of acid was
needed to achieve the target pH of the system. To maintain both the s/l ratio and the
target pH, the amount of water was lowered to accommodate this acid addition, affect-
ing the ease of acid dissociation and the release of H+ ions that are needed for metal
extraction.
Effect of Leaching Temperature The results show enhanced Ni/Co/Sc release in the
pregnant leach solutions when subjected to higher leaching temperatures for a longer
time. This is suggestive of the highly temperature-dependent leaching behavior of the
system, which is also chemical reaction-controlled in nature. However, this is also
accompanied by a significant increase in the leaching of associated impurity metal,
iron. From thermodynamics and kinetics, the higher temperature increases the extrac-
tion rate as it mainly overcomes the activation energy for the reaction to proceed. This
also suggests the exothermic nature of metal dissolution as the rate of reaction and mass
balance differences between the products and reactants change [6].
4 Conclusions
The 2k–p Fractional Factorial Design of Resolution IV was successfully applied in

the screening of reduction pre-treatment and leaching factors. Employing the reduction
pre-treatment results in the recovery of nickel, cobalt, and scandium up to 94.64%,
90.09% and 79.58%, respectively. However, an increase in its recovery also results in a
proportional increase in the %recovery of iron as indicated by the increased acid con-
sumption of the setups. The silica, on the other hand, increases proportionally in higher
leaching setups with lower s/l ratio, and leaching temperatures.
Acknowledgments This project is financially supported by the Philippine Council for Industry,
Energy and Emerging Technology Research of the Department of Science and Technology. The authors
acknowledge the assistance of LNL Archipelago Inc. for providing the laterite ore sample.
References
1. Queneau PB, Berthold CE. Silica in hydrometallurgy: an overview. Can J Metall Mater Sci.
2013;25:201–9. https://doi.org/10.1179/cmq.1986.25.3.201.
2. Chang Y, Fan C, Li B, Zhai X, Zhang T. Selective leaching nickel from the pre-reduced limonitic
laterite ore at atmospheric pressure. School of Material and Metallurgy, Northeastern University,
Shenyang P.R. China. Adv Mater Res. 2012;560–561:494–8. Trans Tech Publication, Switzerland.
https://doi.org/10.4028/www.scientific.net/AMR.560-561.494
3. Kumar M, Jena S, Patel S. Characterization of properties and reduction behavior of iron ores for
application in sponge ironmaking. Miner Process Extr Metall Rev. 2007;29(2):118–29. https://doi.
org/10.1080/08827500701421896.
4. Mohammadreza F, Mohammad N, Ziaeddin S. Nickel extraction from low grade laterite by agi-
tation leaching at atmospheric pressure. Int J Min Sci Technol. 2014;24(4):543–8. https://doi.
org/10.1016/j.ijmst.2014.05.019.
5. Purwanto H, Shimada T, Takahashi R, Yagi J. Recovery of nickel from selectively reduced lat-
erite ore by sulphuric acid leaching. ISIJ Int. 2003;43(2):181–6. https://doi.org/10.2355/
isijinternational.43.181.
6. Pickles C, Elliott R. Thermodynamic analysis of selective reduction of nickeliferous limonitic
laterite ore by carbon monoxide. Miner Process Extr Metall. 2015;124(4):208–16. https://doi.org/1
0.1179/1743285515y.0000000009.
Innovative Solvent Extraction to Produce High-Purity
Cobalt and Nickel Sulphates from Sulphide
Concentrate
M. A. Halim1 (*), S. Kumar1, J. Chen1, V. I. Lakshmanan1, S. DasGupta1,

and J. F. NdoutouMve1
1
Process Research Ortech Inc., Mississauga, ON, Canada
halim@processortech.com
Abstract. The use of cobalt and nickel salts is consistently increasing in the production
of strategic alloys while easily recoverable resources are gradually decreasing.
Environmentally sensitive sulphide concentrates have been used widely in commercial
plants to produce cobalt and nickel products. Process Research ORTECH Inc. (PRO) has
developed an innovative atmospheric chloride leaching technology to recover high-purity
battery-grade cobalt and nickel products from sulphide concentrates. Chloride lixiviant is
an effective leachant that permits rapid leaching rates at modest temperatures and atmo-
spheric pressures, with high metal recovery. This process evaluated the solvent extraction
process steps for producing cobalt and nickel sulphate from their respective chloride solu-
tions. In this chapter, detailed solvent extraction reaction mechanisms, potential process
design parameters, key results, and developed flowsheets will be discussed.
Keywords: Solvent extraction · Chloride leaching · Sulphide concentrate ·

Conversion · Cobalt sulphate · Nickel sulphate
1 Introduction
Demand for high purity salts of cobalt and nickel is increasing while the supply of eco-
nomically and environmentally recoverable resources is dwindling. Chloride-based
hydrometallurgical processes are potential alternatives for the production of these value
metals by meeting the more stringent environmental requirements while reducing capi-
tal and operating costs. Recycling of chloride lixiviant and minimization of waste pro-
duction are apparent advantages of chloride-based hydrometallurgical processes.
Mixed-chloride, containing hydrochloric acid (HCl) and magnesium chloride (MgCl2),
is an effective leachant which permits rapid leaching rates at modest temperature and
atmospheric pressure, along with high metal recovery. Considering these advantages,
innovative mixed chloride processes have been developed by Process Research
ORTECH Inc. (PRO) [1–15].
A mixed-chloride leaching process has been developed for the recovery of valuable
metals such as cobalt (Co) and nickel (Ni) from sulphide concentrate (Fig. 1) [1, 4].
https://doi.org/10.1007/978-3-031-17425-4_89
Fig. 1. PRO process flowsheet for the recovery of valuable metals from sulphide concentrate in mixed
chloride medium [1, 4].
Leaching of this concentrate was conducted at atmospheric pressure and 90–95 °C with
a lixiviant containing HCl and MgCl2. After solid–liquid separation, the pregnant leach
solution is advanced to solvent extraction (SX) process steps to separate iron (Fe) fol-
lowed by copper (Cu), cobalt (Co), and nickel (Ni). Co and Ni are present as cobalt
chloride (CoCl2) and nickel chloride (NiCl2), respectively, in their respective pregnant
strip solutions. This paper describes the potential design parameters and flowsheets for
the conversion of these strip solutions bearing the CoCl2 and NiCl2 into cobalt sulphate
(CoSO4) and nickel sulphate (NiSO4), respectively, by additional solvent extraction
process steps.
2 Cobalt and Nickel Solvent Extraction
2.1 Conversion of CoCl2 into CoSO4
A cation exchange extractant such as Cyanex 272 has an excellent affinity to extract Co
at relatively lower pH and Co can be easily stripped. Extraction of Co with Cyanex 272
depends on the equilibrium pH of the aqueous phase. High Co distribution coefficients
can be achieved if the equilibrium pH is maintained at ≥5.0. Extraction of Co with
Cyanex 272 (HR) can be described by the following reaction:
2 HR org. CoCl2 aq. R 2Co org. 2 HCl aq. (1)

Innovative Solvent Extraction to Produce High-Purity Cobalt and Nickel Sulphates… 789
Equation (1) suggests that each mole of Co2+ ion substitutes two moles of hydrogen
ions (H+), which significantly decreases the equilibrium pH of the aqueous phase result-
ing in lower Co extraction. The neutralization of the produced H+ significantly enhances
the extraction of Co, which can be achieved by saponification of Cyanex 272 with an
alkaline solution prior to CoSX.
CoSX tests were performed with an organic phase containing 20% (v/v) Cyanex
272 in ShellSol D80. Saponification of this organic phase was carried out with 0.375 M
(NH4)2CO3 solution at an organic to aqueous (O/A) phase ratio of 1 for 5 minutes at
ambient temperature, conducting the following reaction:

2 HR org. NH 4 2 CO3 aq. 2 R NH 4 org .
CO2 H 2 O aq. (2)
A simulated CoCl2 feed solution was prepared by dissolving the Co salt in hydrochloric
acid (HCl) solution prior to conducting the CoSX tests. A McCabe-Thiele Co extrac-
tion isotherm was constructed with the saponified organic phase by contacting with
CoCl2 feed solution with varying O/A phase ratios. Each of these tests was also per-
formed for 5 minutes at ambient temperature. The resulting extraction isotherm is
shown in Fig. 2, which suggests that about 8.1 g/L Co can be selectively extracted over
chloride ion (Cl─) with this organic phase. Furthermore, it is indicated that ~34.1 g/L
Co can be extracted by a two-stage counter-current extraction at an O/A phase ratio of
4.25:1.
A sufficient amount of Co-loaded organic phase was prepared for developing the Co
stripping isotherm. Loading of Co to the organic phase was performed at an O/A phase
ratio of 2.5:1 for 5 minutes at ambient temperature followed by a three-stage washing
of the loaded organic phase with de-ionized (D.I.) water. Each washing test was per-
formed for 3 minutes at ambient temperature at an O/A phase ratio of 10:1 to remove
entrained Cl─ ions from the loaded organic phase. Extraction and washing test results
are shown in Table 1 along with the stripping results, which suggests that a small
amount (~35.1 mg/L) of Co was scrubbed from the loaded organic phase after water
washing, which could be recovered by recycling the wash water to the extraction stage.
After washing, a total of ~8.06 g/L Co remained in the washed-loaded organic phase.
Stripping of cobalt from the loaded organic phase with H2SO4 can be shown by the
following reaction.
R 2Co org. H 2 SO 4 aq. 2 HR org. CoSO 4 aq. (3)
Each stripping SX test was performed by contacting the washed-loaded organic phase
with 1.5 M H2SO4 solution for 3 minutes at ambient temperature and varying the O/A
phase ratio. The cobalt stripping isotherm is shown in Fig. 3. The cobalt stripping iso-
therm suggests that >105 g/L Co can be found in the pregnant strip liquor by a two-
stage counter-current contacting of the washed-loaded organic phase with 1.5 M H2SO4
solution at the O/A phase ratio of 13:1. Consequently, almost complete stripping of
cobalt from the organic phase can be achieved under these conditions. The pH of the
pregnant strip solution obtained from the O/A ratio of 15:1 was ~1.85. Very high Co to
Cl─ ratios (~10,000:1) were observed in the final pregnant strip solution. The overall SX
flow sheet for the conversion of CoCl2 into CoSO4 is shown in Fig. 4.
Fig. 2. The McCabe–Thiele cobalt extraction isotherm with 20% (v/v) Cyanex 272 in ShellSol D80
saponified with 0.375 M (NH4)2CO3 solution at equilibrium pH of ~5.5–6.0.
Table 1. Results of Co extraction, impurity scrubbing/washing and Co stripping results.

Co Ni Ca K Al Mg Cl─ Efficiency for Co
Sample O/A mg/L %
Cobalt extraction
Cobalt feed 34,260 20 61 261 <5 <5 37,050
Cobalt raffinate 2.5:1 14,030 8 34 251 <5 <5 36,900 59
Scrubbing/washing
Washed water-I 10:1 284 <5 28 206 <5 <5 897 0.4
Washed water-II 10:1 63 <5 14 286 <5 <5 105 0.1
Washed water-III 10:1 5 <5 12 275 <5 <5 <5 0.01
Cobalt stripping
Pregnant strip-I 20:1 111,050 15 163 230 <5 <5 13 69
Pregnant strip-II 15:1 110,000 15 144 182 <5 <5 11 91
Pregnant strip-III 10:1 88,900 14 127 213 <5 <5 6 ~100
Pregnant strip-IV 5:1 42,190 9 82 263 <5 <5 4 ~100
Pregnant strip-V 1:1 8,402 <5 26 245 <5 <5 9 ~100
Pregnant strip-VI 1:5 1,617 <5 33 236 <5 <5 7 ~100
Fig. 3. The McCabe–Thiele cobalt stripping isotherm with 1.5 M H2SO4.
Fig. 4. Flowsheet for the conversion of CoCl2 into CoSO4 by SX with 20% (v/v) for the organic phase
consisting of 20% (v/v) Cyanex 272 in ShellSol D80.
2.2 Conversion of NiCl2 into NiSO4 by Solvent Extraction
Nickel has a tendency to form neutral chloro-complexes i.e., nickel chloride (NiCl2) in
a hydrochloric acid medium. A simulated NiCl2 feed solution was prepared by dissolv-
ing the Ni salt in HCl solution before conducting NiSX tests. Conversion of NiCl2 into
NiSO4 was performed with a synergistic extraction system containing of 15% (v/v) LIX
63, 15% (v/v) Cyanex 272 and 30% (v/v) TBP in ShellSol D80. This combination was
chosen due to the higher Ni loading and better phase disengagement observed com-
pared to the other organic phases of 10% LIX 63, 10% Cyanex 272 and 20% TBP in
D80; 15% LIX 63, 15% Cyanex 272 and 30% TBP in D80; 10% LIX 84-I, 10% Cyanex
272 and 20% TBP in D80; 10% PC 88A, 10% LIX 63 and 20% TBP in D80; and 20%
15
12
9
[Ni]org (g/L)
3 30.8 g/L Ni
0
0 9 18 27 36 45
[Ni]aq (g/L)
Fig. 5. The McCabe–Thiele Ni extraction isotherm at the equilibrium pH of 6.0–6.5 for a synergistic
extraction system consisting of 15% (v/v) LIX 63, 15% (v/v) Cyanex 272 and 30% (v/v) TBP in
ShellSol D80 saponified with 0.375 M (NH4)2CO3.
PC 88A and 20% TBP in D80. Nickel extraction with this synergistic organic system
(HR) can be described by the following reaction number (4). Similar to the CoSX,
saponification of this organic phase was conducted with 0.375 M (NH4)2CO3 at ambient
temperature prior to Ni solvent extraction tests. This was found to provide faster phase
separation times compared to saponification using ammonium hydroxide. Nickel
extraction tests were completed by varying the O/A phase ratio at ambient temperature
for 5 minutes. The McCabe–Thiele Ni extraction isotherm is shown in Fig. 5, which
suggests that ~8.1 g/L Ni can be loaded to the organic phase at an equilibrium pH of
6.0–6.5, while 30.8 g/L Ni can be completely extracted by a two-stage extraction at an
O/A phase ratio of 1:0.26.
2 HR org. NiCl 2 aq. R 2 Ni org. 2 HCl aq. (4)
Washing of Ni-loaded organic is required for the removal of entrained Cl− ions prior
to the stripping of Ni into sulphate media. Firstly, enough Ni-loaded organic phase was
prepared by conducting the saponified organic phase with the synthesized Ni feed solu-
tion at an O/A phase ratio of 3:1 at ambient temperature. Subsequently, the Ni-loaded
organic phase was washed with D.I. water 3 times at O/A 10:1 at ambient temperature
prior to stripping Ni. Assay results of the aqueous phases are presented in Table 2,
which suggests that ~9.1 g/L of Ni is loaded to the organic phase and most of the
entrained Cl− ions are removed.
Stripping of nickel from the loaded organic phase with H2SO4 can be shown by the
following reaction.
R 2 Ni org. H 2 SO 4 aq. 2 HR org. NiSO 4 aq. (5)

Table 2. Results of Ni extraction, organic washing and Ni stripping SX tests.

Ni Cl− Co Ca K Mg Mn Zn Efficiency for Ni
Sample description O/A mg/L %
Ni extraction
Nickel feed solution 32,500 34,050 5 12 421 9 <5 <5
Nickel raffinate 3:1 8035 33,850 2 <5 139 <5 <5 <5 75
Organic washing
Washed water-I 10:1 316 665 1 <5 73 <5 <5 <5 0.004
Washed water-II 10:1 85 212 <1 <5 91 <5 <5 <5 0.001
Washed water-III 10:1 <5 <5 <1 <5 61 <5 <5 <5 0.000
Nickel stripping
Nickel pregnant strip-I 15:1 90,950 17 7 47 230 32 <5 <5 74
Nickel pregnant strip-II 10:1 73,250 20 5 25 214 17 <5 <5 90
Nickel pregnant strip-III 5:1 39,550 33 2 17 201 <5 <5 <5 97
Nickel pregnant strip-IV 1:1 8,205 25 1 9 244 9 <5 <5 ~100
Nickel pregnant strip-V 1:5 1,720 40 <1 <5 174 9 <5 <5 ~100
Fig. 6. The McCabe–Thiele nickel stripping isotherm with 1.5 M H2SO4.
Nickel stripping tests were performed on the loaded and washed organic phase using
1.5 M H2SO4 solution at ~40 °C for 5 minutes and varying the O/A phase ratio. The
McCabe–Thiele Ni stripping isotherm is shown in Fig. 6, and assay results of aqueous
solutions are also shown in Table 2. The stripping isotherm suggests that ~90 g/L Ni can
be found in the pregnant strip solution and loaded Ni can be completely stripped by a
two-stage stripping with 1.5 M H2SO4 at an O/A phase ratio of 11:1. The residual chlo-
ride ion in the ~90 g/L Ni strip liquor is very low (~17 mg/L Cl) (Table 2). The pH of
the pregnant strip solution obtained from the test conducted at the O/A ratio of 15:1 was
~1.78. These results suggest that the synergistic organic system consisting of 15% (v/v)
LIX 63, 15% (v/v) Cyanex 272 and 30% (v/v) TBP in ShellSol D80, which is
Fig. 7. Flowsheet for the conversion of NiCl2 into NiSO4 by solvent extraction process steps with a
synergistic extraction system consisting of 15% (v/v) LIX 63, 15% (v/v) Cyanex 272 and 30% (v/v)
TBP in D80.
saponified with 0.375 M (NH4)2CO3 by following the reaction number (2), could be
used in SX plants for the conversion of NiCl2 into NiSO4. The overall flowsheet is
shown in Fig. 7.
3 Conclusion
Innovative solvent extraction process steps were evaluated for producing high-purity
CoSO4 and NiSO4 from their respective simulated CoCl2 and NiCl2 pregnant strip solu-
tions based on PRO’s mixed chloride technology for base metal sulphide concentrate.
An ammonium carbonate solution was used to saponify organic phases prior to con-
ducting the CoSX and NiSX tests. Cobalt and Nickel McCabe–Thiele extraction and
stripping isotherms were developed with organic phases consisting of 20% (v/v) Cyanex
272 in ShellSol D80; and 15% (v/v) LIX 63, 15% (v/v) Cyanex 272 and 30% (v/v) TBP
in ShellSol D80, respectively. The stripping isotherms suggest that a very high concen-
tration of Co (>105 g/L) and Ni (~90 g/L) in pregnant strip solutions can be produced
by using these organics. The small amount of chloride ions detected in the Co and Ni
pregnant strip solution (~13 mg/L and ~17 mg/L respectively) furthermore suggests
that these organic phases can be used in commercial SX plants for the conversion of
CoCl2 and NiCl2 into CoSO4 and NiSO4.
References
1. Lakshmanan VI, Sridhar R, Chen J, Halim MA. A mixed-chloride atmospheric leaching process for
the recovery of base metals from sulphide materials. Trans Indian Inst Metals. 2017;70(2):463–70.
2. Lakshmanan VI, Sridhar R, Chen J, Halim MA. Development of mixed-chloride hydrometallurgi-
cal processes for the recovery of value metals from various resources. Trans Indian Inst Metals.
2016;69:39–50.
3. Harris GB, Lakshmanan VI, Sridhar R. Canadian Patent 2467288, 14 November 2006.
4. Harris GB, Lakshmanan VI, Sridhar R, Puvvada G. US Patent 7736606B2, 15 June 2010.
5. Lakshmanan VI, Sridhar R, Halim MA. US Patent Application 2013/0283976 A1, 31 October 2013.
6. Lakshmanan VI, Sridhar R, Harris GB, Puvvada G. US Patent 7803336, 28 September 2010.
7. Lakshmanan VI, Sridhar R, Harris, GB, Puvvada G. Canadian Patent 2513309, 15 June 2010.
8. Lakshmanan VI, Sridhar R, Roy R, Ramachandran V. Proceedings of stamp hydrometallurgy 2008,
Phoenix, Arizona. 2008. p. 895.
9. Lakshmanan VI, Sridhar R, Harris GB, Magee J. Proceedings of a symposium of the minerals,
metals and materials society, Charlotte, North Carolina. Imrie WP, Lane DM, editors. 2004. p. 219.
10. Lakshmanan VI, Sridhar R, Rishea M, Delaat R. US Patent 6699446, 2 March 2004.
11. Lakshmanan VI, Sridhar R, Rishea M, DeLaat R. US Patent 6500396, 31 December 2002.
12. Lakshmanan VI, Sridhar R, Rishea M, DeLaat R. Canadian Patent 2289967, 2 July 2002.
13. Lakshmanan VI, Harris GB. Proceedings of a symposium 24th annual hydrometallurgical confer-
ence of CIM, Toronto, Ontario. Harris GB, Krause E, editors. 1994. p. 3.
14. Lakshmanan VI, Lawson GJ, Nyholm PS. Proceedings of international solvent extraction society,
vol 2, Lyon, France. 1974. p. 1169.
15. Lakshmanan VI, Roy R, Halim MA. Innovative process for the production of titanium dioxide. In:
Lakshmanan VI, Roy R, Ramachandran V, editors. Innovative process development in metallurgi-
cal industry. Springer International Publishing; 2016. p. 359.
The Extraction of Nickel and Cobalt from Laterite
Ores with Concurrent Carbon Sequestration
Fei Wang1 (*) and David Dreisinger1

1
Department of Materials Engineering, The University of British Columbia,
Vancouver, BC, Canada
fei.wang@gmn.ulaval.ca
Abstract. The pathway to net zero carbon emissions requires sequestration of CO2 and
the widespread adoption of electric vehicles to minimize emissions from the transporta-
tion sector. Nickel laterite ores are rich sources of nickel and cobalt for electric vehicle
battery manufacture. These ores are also rich in magnesium and iron silicates in the form
of olivine and serpentine mineralization.
The carbonation of nickel laterite ores has been conducted at elevated temperatures
with an overpressure of carbon dioxide. A soluble complexant, (ethylenediaminetetraace-
tic acid disodium salt (Na2EDTA)) can be added to the slurry during carbonation to affect
concurrent nickel and cobalt extraction. Nickel and cobalt are held in solution and thus
can be recovered from the carbonated residue. This process offers a new route to carbon
sequestration and battery metal recovery with promising results. A sample of olivine
((Mg,Fe)2SiO4) yielded 88% Ni and 62% Co extraction with 88% mineral carbonation
and 2.0% Fe extraction and 0.02% Mg extraction. A sample of nickel laterite ore yielded
73% Ni extraction, 55% Co extraction, 73% mineral carbonation and 6.0% Fe extraction
and 0.5% Mg extraction, respectively.
Keywords: CO2 sequestration · Selective nickel and cobalt extraction · Laterite ·

EDTA · Olivine · Zero carbon emissions · Mineral carbonation
1 Introduction
There is a global consensus that net zero carbon emissions need to be achieved by 2050
[1]. The pathway to achieve the universal goal requires substantial sequestration of CO2
[2–6] and adoption of electrical vehicles (EV) [7] in the transportation sector. With an
increase in EV adoption, the demands for the important metal components of batteries
for EVs are significantly increased. Nickel usually exceeds 60% of metal components
of the nickel-manganese-cobalt cathode (NMC) [8]. The nickel content may further
increase in the future to expand energy capacity [8–10]. Cobalt is also important for
boosting energy density and battery life [8]. However, the current global supply may
not be able to meet the rapid increase in demand for nickel and cobalt.
The majority of worldwide nickel resources are in laterites, accounting for approxi-
mately 70% [11]. With continuous decreasing of nickel sulfide resources [12, 13], the
https://doi.org/10.1007/978-3-031-17425-4_90
798 F. Wang and D. Dreisinger
future increase in the supply of nickel together with cobalt will be mainly from nickel
laterites. High-pressure acid leaching (HPAL) and ferronickel production (pyrometal-
lurgy) are the two dominant processes to produce nickel and cobalt from limonite later-
ite and saprolite laterite, respectively [14]. For nickel and cobalt production, magnesium
and iron from laterites are considered as impurities based on current technologies. In
contrast, magnesium and iron can be significant resources for CO2 mineralization. CO2
mineralization (mineral carbonation) is a method to permanently sequester CO2 gas as
stable mineral carbonates [2–6, 15–18]. The method of CO2 mineralization can match
with the goal of laterite production in terms of removing magnesium and iron as impu-
rities. Additionally, stabilizing magnesium and iron as mineral carbonates can signifi-
cantly reduce CO2 emissions in nickel and cobalt production in the laterite industry.
Previous studies [18, 19] have confirmed the possibility of recovering nickel and cobalt
from olivine-based silicate minerals by utilizing direct aqueous CO2 mineralization.
This work reports the extraction of nickel and cobalt with simultaneous CO2 mineral-
ization from a saprolite sample. A complexant reagent, ethylenediaminetetraacetic acid
disodium salt (Na2EDTA), is used in this work together with sodium bicarbonate to
achieve the selectivity of nickel and cobalt extraction and magnesium and iron carbon-
ation. Since the stability constant of the metal-EDTA is in the forms of Ni2+ > Co2+ >
Fe2+ > Mg2+ [20], it is expected that the released divalent metal ions owing to the dis-
solution of olivine during the CO2 mineralization process can go through a selective
process by a limited Na2EDTA supply. Ni2+ and Co2+ can be stabilized in an aqueous
solution as Ni-EDTA2− and Co-EDTA2− complex ions while Fe2+ and Mg2+ can precipi-
tate with CO2 supply as FeCO3 and MgCO3 in residues. The corresponding proposed
chemical reaction during this process is shown in Reaction (1).
Mg,Fe,Ni,Co 2 SiO4 CO2g Na2 H 2 EDTAa

Na2 Ni,Co EDTA a Mg,Fe CO3 SiO2 H 2 O1 (1)
2.1 Materials
A sample of natural olivine provided by Sibelco Europe and a saprolite laterite were
used in this work. The olivine and the saprolite laterite are magnesium iron silicate and
iron-rich chemically-weathered rock, respectively. The mineral compositions of the
natural olivine sample (as stated as olivine in the following context) and the saprolite
laterite sample (as stated as laterite in the following context) are shown in Fig. 1. Olivine
mainly contains olivine (93%), with a small amount of serpentine and enstatite. The
mineralogy of the laterite is complex, which includes serpentine, olivine, quartz, goe-
thite, enstatite, talc, and montmorillonite. The chemical composition of olivine and
laterite is shown in Table 1. Olivine contains 27.4% magnesium, 7.1% iron, 0.27%
nickel and 0.005% cobalt. Iron (as Fe(II)), nickel and cobalt all replace magnesium in
The Extraction of Nickel and Cobalt from Laterite Ores with Concurrent Carbon… 799
Fig. 1. X-ray diffraction diagram of the pure olivine and the saprolite laterite sample.
Table 1. Chemical composition of the pure olivine and the saprolite laterite sample.
Samples Mg Si Fe Al Cr Ni Co Mn Ca
Olivine, wt% 27.44 20.76 7.09 0.09 0.21 0.27 0.005 0.08 0.11
Laterite, wt% 14.64 17.9 12.14 0.93 1.3 2.28 0.063 0.27 0.20
the crystal structure of olivine. The magnesium content in laterite is lower than olivine,
at 14.6%. In contrast, iron, nickel, and cobalt are much higher than in olivine, occupy-
ing 12.1%, 2.28%, and 0.063%, respectively. The theoretical carbonation capacity of
olivine and the laterite, based on magnesium, iron and calcium (i.e. 1-mole divalent
metal sequesters 1 mole CO2), is 0.54 t CO2/t olivine and 0.36 t CO2/t laterite, respec-
tively. Correspondingly, nickel and cobalt are the target metals to be recovered during
CO2 mineralization. Due to the relatively low content of cobalt, the work focuses on
nickel extraction from olivine and the laterite.
2.2 Methods
Based on the previous studies [16–18, 21], the CO2 mineralization tests were carried
out under the conditions of 175 °C, CO2 pressure (PCO2) = 34.5 bar, and 1.5 molality of
sodium bicarbonate concentration through a 600-mL stainless steel autoclave (Parr No.
5103). Since the carbonation of olivine was still slow [16, 17], the tests were up to
8 hours to achieve higher than 80% carbonation efficiency. The longer duration was not
considered because of high capital costs [22]. Olivine was ground and screened through
a 500-mesh sieve to obtain less than 25 μm particles for the tests with 5% pulp density.
Laterite was pulverized to P80 at 27 μm. The tests for laterite were with the addition of
sodium sulfite to maintain a reductive condition for accelerated CO2 mineralization
[23]. The solution samples were analyzed by atomic absorption spectroscopy (AAS)
(VARIAN-AA240) to determine the concentration of divalent metals and the solid was
analyzed through a LECO C analysis (LECO CS230) to determine the increased carbon
content owing to carbonation precipitation.
Our previous study on natural silicate samples [21] has confirmed that serpentine
(hydrated silicate), enstatite, and magnetite were difficult to react with CO2 to form
carbonates under the test conditions. Olivine was the dominantly reactive mineral.
Therefore, laterite needed calcination to convert all hydrated silicate minerals including
serpentine and montmorillonite to synthetic olivine. Some nickel and cobalt may also
exist in goethite and thus a reductive atmosphere was required to convert goethite to
wustite where iron is reduced to the divalent oxidation state. The calcination conditions
for laterite were 700 °C, 300 mL/min flow of a gas mixture containing 7% CO and 93%
N2 for 1 hour through a rotary kiln – electrical furnace (Nabertherm RSRB 80-750/11).
The calcined laterite was then used for CO2 mineralization and concurrent nickel and
cobalt extraction tests with 0.5% pulp density.
3.1 Nickel and Cobalt Extraction from Olivine and Concurrent Carbon

Sequestration
Olivine was tested with a dosage of EDTA/total nickel molar ratio = 1.2/1. The test
results are shown in Fig. 2. With the CO2 mineralization reaction progressing, the min-
eral carbonation efficiency gradually increased to 88% within 8 hours. Correspondingly,
the nickel leaching efficiency also gradually increased with the identical trend as car-
bonation and reached 88% within 8 hours as well. The cobalt leaching efficiency gradu-
ally increased as well, with identical trends of nickel and carbonation within the first
3 hours and then slowed down to reach 62% of total cobalt. In contrast, iron and mag-
nesium leaching was less than 2% and 0.02%, respectively. The selective nickel and
cobalt extraction and CO2 mineralization of iron and magnesium can be achieved
simultaneously in accordance with Reaction (1). It also means that each tonne of oliv-
ine reacted can permanently sequester 476 kg CO2 as stable mineral carbonates and
Fig. 2. CO2 mineralization and nickel and cobalt extraction from olivine with Na2EDTA at the dosage
of EDTA/total nickel molar ratio = 1.2/1.
recover 2.4 kg nickel and 0.03 kg cobalt into solution. This method may make signifi-
cant contributions to CO2 sequestration and enhanced nickel and cobalt recovery. The
leached divalent metals were in the order of metal-EDTA complex stability:
Ni-EDTA2− > Co-EDTA2− > Fe(II)-EDTA2− > Mg-EDTA2− [20]. In addition, it is also
noted that the iron leaching efficiency firstly reached 6% and then gradually decreased
to less than 2% due to the increase of nickel and cobalt extracted, which corresponds to
the priority of EDTA complexation. Different from CO2 mineralization and concurrent
sulfidization [18], it indicates that pre-adding sufficient EDTA prior to reaction is suit-
able to achieve highly selective nickel and cobalt extraction and concurrent CO2
mineralization.
3.2 Comparison of Extraction and Carbonation Between Raw and Calcined

Laterite
The CO2 mineralization and metal extraction process was further tested on the raw and
calcined laterite at identical conditions except for 0.5% pulp density and the addition of
0.59 molality sodium sulfite. The addition of sodium sulfide was to maintain reductive
conditions for accelerated mineral carbonation [23]. The results are shown in Fig. 3.
Without calcination, the nickel leaching efficiency and CO2 mineralization efficiency
only reached 27% and 7.7% within 2 hours, respectively. In contrast, with calcination
using 7% CO and 93% N2, the nickel leached in an aqueous solution reached 73% the
same as the CO2 mineralization efficiency within 2 hours. During the step of simultane-
ous CO2 mineralization and metal extraction, 261 kg CO2 was converted to stable min-
eral carbonates by using each tonne laterite together with that 16.7 kg nickel was
extracted. The further increase in reaction hours did not obviously increase the nickel
leaching and carbonation. The significant increase in both carbonation and nickel leach-
ing was due to calcination where non-reactive hydrated silicate minerals and goethite
converted to reactive olivine and wustite for the mineralization reaction.
3.3 Effect of EDTA
Since EDTA played an important role in the selective CO2 mineralization and nickel
and cobalt extraction, the effect of EDTA dosage was tested on the calcined laterite, as
shown in Fig. 4. Similar to leaching from olivine, with gradual increase of EDTA to the
molar ratio of EDTA/Nitotal = 1.2/1, the nickel and cobalt leaching efficiency rose to
73% and 55% within 2 hours, respectively. In contrast, there was no iron and magne-
sium leached in the aqueous solution at molar ratio of EDTA/Nitotal < 0.5/1. The supply
of EDTA to the CO2 mineralization system was not sufficient. The increase in EDTA
dosage to molar ratio of 1.2/1 led to around 6% of iron and 0.5% of magnesium in an
aqueous solution as soluble EDTA complex ions. The further increase in EDTA did not
Fig. 3. Comparison of nickel extraction and CO2 mineralization between raw and calcined saprolite
laterite.
significantly increase nickel and cobalt recovery but considerably increased iron leach-
ing efficiency to 22%. The selectivity was weakened by excessive EDTA in the leaching
system. Therefore, it is recommended to maintain EDTA dosage at EDTA/Nitotal = 1.2/1.
3.4 Effect of Sodium Bicarbonate
Sodium bicarbonate is an important reagent to accelerate the CO2 mineralization of

olivine [15, 16, 24] through the Reactions (2)–(8) [17]. Therefore, the addition of
sodium bicarbonate has also been tested to affect the ability of CO2 mineralization and
nickel and cobalt leaching from the calcined laterite, as shown in Fig. 5. In the absence
of sodium carbonate addition, there was no obvious effect on metal extraction. However,
the CO2 mineralization efficiency was significantly decreased from 73% to only 23%.
Correspondingly, the nickel in solids increased from 0.6% to 0.9%. To achieve maxi-
mum CO2 mineralization, a high concentration of sodium bicarbonate, e.g. 1.5 molality,
should be maintained.
Fig. 4. Effect of EDTA dosage on selective nickel and cobalt extraction from the calcined saprolite
laterite.
4H Mg 2 SiO4 2 Mg 2 H 4 SiO4a (2)
Mg 2 CO32 MgCO3 (3)
HCO3 H CO32 (4)
H 2CO3a H HCO3 (5)
CO2a H 2 O1 H 2 CO3a (6)
CO2 g CO2a (7)
H 4 SiO4a SiO2s H 2 O1 (8)
3.5 Effect of Sodium Sulfite
It is known that CO2 mineralization may be accelerated in a slightly reductive atmo-

sphere together with CO2 gas supply [18, 23]. Additionally, sodium sulfite, a reducing
agent, may also be used to enhance the dissolution of goethite in case there was any
Fig. 5. Effect of sodium bicarbonate on CO2 mineralization and metal extraction from the calcined
saprolite laterite.
unreacted nickel-containing goethite left after calcination. Therefore, sodium sulfite

was also added to the CO2 mineralization and leaching system of the calcined laterite as
shown in Fig. 6. It is found that the addition of 0.59 molality sodium sulfite can increase
nickel and cobalt leaching efficiency together with mineral carbonation efficiency by
16%, 9%, and 10%, respectively. Further increase in sodium sulfite concentration was
not important to enhance leaching and CO2 mineralization. Herein, sodium sulfite may
also be also needed at 0.59 molality to enhance the reaction together with sodium
bicarbonate.
3.6 Effect of CO2 Pressure
The effect of CO2 pressure was also tested at 1 bar, 20.7 bar, and 34.5 bar for CO2 min-
eralization and metal extraction from the calcined laterite as shown in Fig. 7. At
PCO2 = 1 bar, the CO2 supply may be suitable for carbon sequestration, but the nickel
and cobalt extraction and mineral carbonation efficiency were low, at only 55%, 48%,
and 34% respectively. With PCO2 increasing to 20.7 bar, the nickel and cobalt extraction
rate has reached 71% and 52%, respectively, together with the decrease of iron leaching
efficiency from 12% to 6%. The further increase of PCO2 slightly improved the nickel
and cobalt leaching. The CO2 mineralization efficiency showed a linearly increasing
relationship with PCO2 from slightly above 30% to 73% at PCO2 = 34.5 bar within 2 hours.
Fig. 6. Effect of sodium sulfite on CO2 mineralization and metal extraction from the calcined saprolite
laterite.
Fig. 7. Effect of CO2 pressure on CO2 mineralization and metal extraction from the calcined saprolite
laterite.
Therefore, high CO2 pressure greater than 20.7 bar may be still needed for fast reaction
and enhanced CO2 mineralization and nickel and cobalt leaching.
3.7 Effect of Temperature
Since the high temperature and the usage of autoclave are important for high capital
costs, the suitability at lower temperatures (75, 125 °C) has been tested compared to at
175 °C, as shown in Fig. 8. At 75 °C, the leaching efficiency of divalent metals was 56%
Ni and 51% Co versus 12% Fe and 27% Mg, respectively. The increase of temperature
to 125 °C can increase nickel and cobalt leaching to 66% and 54% while decreasing
iron and magnesium leaching to 7% and 1.7% within 2 hours, respectively. Compared
to the results at 175 °C, the decrease in temperature to 75 and 125 °C lowered down Ni
leaching efficiency by 16% and 6%, respectively. It is still required to maintain the
temperature at 175 °C and use an autoclave for the process.
Fig. 8. Effect of carbonation temperature on CO2 mineralization and metal extraction from the cal-
cined saprolite laterite.
4 Conclusion
This work has confirmed the suitability of CO2 mineralization of olivine and laterite for
concurrent and selective leaching of nickel and cobalt. CO2 mineralization releases the
divalent metal ions from silicate minerals for selective carbonate precipitation of mag-
nesium and iron and leaching nickel and cobalt into an aqueous solution. A suitable
complexant, Na2EDTA, is required to achieve selective extraction of nickel and cobalt.
The saprolite laterite required calcination in a reductive atmosphere as a pretreatment
step to convert hydrated silicate minerals and goethite to olivine and wustite for CO2
mineralization and nickel and cobalt extraction. At the optimized conditions thus far,
for olivine, 88% Ni and 62% Co extraction with 88% mineral carbonation can be
achieved versus 2% Fe extraction and 0.02% Mg extraction; for the saprolite laterite,
73% Ni extraction, 55% Co extraction, and 73% mineral carbonation can be achieved
versus 6% Fe extraction and 0.5% Mg extraction. Correspondingly, each tonne of the
material can sequester 476 kg CO2 and extract 2.4 kg Ni and 0.03 kg Co for olivine,
261 kg CO2 and extract 16.7 kg Ni and 0.36 kg Co for the laterite, respectively. This
CO2 mineralization and concurrently enhanced metal recovery can make a significant
contribution to CO2 emission reduction and critical metals supply.
Acknowledgment The authors thank Mitacs Accelerate and LeadFX Inc. for the financial supports.
The authors also thank Sibelco Europe for supplying pure olivine sample.
References
1. UNFCCC: Paris Agreement, Conference of the Parties on its twenty-first session 32. FCCC/
CP/2015/L.9/Rev.1. 2015.
2. Xi F, Davis SJ, Ciais P, Crawford-Brown D, Guan D, Pade C, Shi T, Syddall M, Lv J, Ji L, Bing L,
Wang J, Wei W, Yang KH, Lagerblad B, Galan I, Andrade C, Zhang Y, Liu Z. Substantial global car-
bon uptake by cement carbonation. Nat Geosci. 2016;9:880–3. https://doi.org/10.1038/ngeo2840.
3. Sanna A, Uibu M, Caramanna G, Kuusik R, Maroto-Valer MM. A review of mineral carbon-
ation technologies to sequester CO2. Chem Soc Rev. 2014;43:8049–80. https://doi.org/10.1039/
C4CS00035H.
4. Olajire AA. A review of mineral carbonation technology in sequestration of CO2. J Pet Sci Eng.
2013;109:364–92. https://doi.org/10.1016/j.petrol.2013.03.013.
5. Azdarpour A, Asadullah M, Mohammadian E, Hamidi H, Junin R, Karaei MA. A review on carbon
dioxide mineral carbonation through pH-swing process. Chem Eng J. 2015;279:615–30. https://
doi.org/10.1016/j.cej.2015.05.064.
6. Aminu MD, Nabavi SA, Rochelle CA, Manovic V. A review of developments in carbon dioxide
storage. Appl Energy. 2017;208:1389–419. https://doi.org/10.1016/j.apenergy.2017.09.015.
7. Cano ZP, Banham D, Ye S, Hintennach A, Lu J, Fowler M, Chen Z. Batteries and fuel cells for emerg-
ing electric vehicle markets. Nat Energy. 2018;3:279. https://doi.org/10.1038/s41560-018-0108-1.
8. Patel P, special to C & EN. Lithium-ion batteries go cobalt-free. C & EN Glob Enterp. 2020;98:9.
https://doi.org/10.1021/CEN-09829-SCICON5.
9. Li W, Lee S, Manthiram A. High-nickel NMA: a cobalt-free alternative to NMC and NCA cathodes
for lithium-ion batteries. Adv Mater. 2020;32:e2002718. https://doi.org/10.1002/adma.202002718.
10. Shafiei Sabet P, Sauer DU. Separation of predominant processes in electrochemical imped-
ance spectra of lithium-ion batteries with nickel-manganese-cobalt cathodes. J Power Sources.
2019;425:121–9. https://doi.org/10.1016/J.JPOWSOUR.2019.03.068.
11. Norgate T, Jahanshahi S. Assessing the energy and greenhouse gas footprints of nickel laterite
processing. Miner Eng. 2011;24:698–707. https://doi.org/10.1016/j.mineng.2010.10.002.
12. USGS: U.S. Geological Survey: Germanium statistics and information. Available at: https://
www.usgs.gov/centers/national-minerals-information-center/nickel-statistics-and-information.
Accessed 16 Feb 2022.
13. Mudd GM, Jowitt SM. A detailed assessment of global nickel resource trends and endowments.
Econ Geol. 2014;109:1813–41. https://doi.org/10.2113/econgeo.109.7.1813.
14. Dalvi AD, Bacon WG, Osborne RC. The past and the future of nickel laterites. In: PDAC 2004
international convention trade show investors exchange. Toronto: The Prospectors and Developers
Association of Canada; 2004. p. 1–27.
15. Wang F, Dreisinger DB, Jarvis M, Hitchins T. The technology of CO2 sequestration by mineral
carbonation: current status and future prospects. Can Metall Q. 2018;57:46–58. https://doi.org/1
0.1080/00084433.2017.1375221.
16. Wang F, Dreisinger D, Jarvis M, Hitchins T. Kinetics and mechanism of mineral carbon-
ation of olivine for CO2 sequestration. Miner Eng. 2019;131:185–97. https://doi.org/10.1016/j.
mineng.2018.11.024.
17. Wang F, Dreisinger D, Jarvis M, Hitchins T, Dyson D. Quantifying kinetics of mineralization of
carbon dioxide by olivine under moderate conditions. Chem Eng J. 2019;360:452–63. https://doi.
org/10.1016/j.cej.2018.11.200.
18. Wang F, Dreisinger D, Jarvis M, Hitchins T, Trytten L. CO2 mineralization and concurrent utiliza-
tion for nickel conversion from nickel silicates to nickel sulfides. Chem Eng J. 2021;406:126761.
https://doi.org/10.1016/j.cej.2020.126761.
19. Wang F, Dreisinger D, Barr G, Martin C. Utilization of copper nickel sulfide mine tailings for CO2
sequestration and enhanced nickel sulfidization. In: REWAS 2022: developing tomorrow’s techni-
cal cycles, vol. I. Cham: Springer; 2022. p. 227–39. https://doi.org/10.1007/978-3-030-92563-5_24.
20. Martell AE, Smith RM. Critical stability constants. New York: Plenum Press; 1974. https://doi.
org/10.1007/978-1-4615-6761-5.
21. Wang F, Dreisinger D, Jarvis M, Hitchins T. Kinetic evaluation of mineral carbonation of natural
silicate samples. Chem Eng J. 2021;404:126522. https://doi.org/10.1016/j.cej.2020.126522.
22. Santos RM, Knops PCM, Rijnsburger KL, Chiang YW. CO2 energy reactor – integrated mineral
carbonation: perspectives on lab-scale investigation and products valorization. Front Energy Res.
2016;4:1–6. https://doi.org/10.3389/fenrg.2016.00005.
23. Wood CE, Qafoku O, Loring JS, Chaka AM. Role of Fe(II) content in olivine carbonation in
wet supercritical CO2. Environ Sci Technol Lett. 2019;6:592–9. https://doi.org/10.1021/acs.
estlett.9b00496.
24. Gadikota G, Matter J, Kelemen P, Park AHA. Chemical and morphological changes during oliv-
ine carbonation for CO2 storage in the presence of NaCl and NaHCO3. Phys Chem Chem Phys.
2014;16:4679–93. https://doi.org/10.1039/c3cp54903h.
II-VI’s Capabilities in Scandium Recovery
and Purification
Gomer Abrenica1, Ghazaleh Nazari1 (*), and Shailesh Patkar2
1
II-VI Performance Metals, Rosario, Cavite, Philippines
Ghazaleh.Nazari@ii-vi.com
2
II-VI, Saxonburg, PA, USA
Abstract. Scandium (Sc) is one of the highest valued elements in the periodic table and
a critical raw material essential to several emerging applications. Sc’s technological appli-
cations are unique as it is a key component in producing solid oxide fuel cells and high-
strength aluminum alloys used in aerospace and 3D printing. Historically, the supply of
Sc has been limited due to its scarcity and the high cost of production. Most Sc is pro-
duced today as a by-product of other mineral refining processes, such as bauxite residues
from alumina production and acid wastes from titanium dioxide pigment production.
These sources have Sc concentrations that are at the levels of milligrams per kilogram
(mg/kg) and are considered exploitable but technologically complicated to recover. II-VI
has significantly advanced the availability of Sc through a series of patented processes for
the economical recovery of the element. II-VI has the expertise and experience to tailor
the Sc recovery process flowsheets to meet customers’ requirements and uses its advanced
technologies to further purify and produce greater than 99.9% scandium oxide.
Keywords. Scandium · Rare earth · II-VI · Solvent · Bauxite residue · Acid wastes
1 Introduction
1.1 Scandium Supply and Demand

Sc is sparsely distributed in the earth’s crust and occurs in trace amount in many miner-
als. These minerals are concentrated in Australia, the Philippines, China, Kazakhstan,
Norway, Russia, Ukraine, and Madagascar. In Australia and the Philippines, Sc is found
in nickel and cobalt ores, in Russia and Kazakhstan in uranium ores, and in Madagascar
and Norway in pegmatite rocks. In Ukraine, scandium is contained in iron ores. In
China, Sc is contained in tin, tungsten, zirconium and iron ores [1].
Sc is usually found as a by-product of other metal processing such as bauxite residue
from alumina production, acid waste from titanium dioxide (TiO2) pigment production,
tailings from nickel/cobalt (Ni/Co) recovery processes, and tailings from tantalum (Ta),
niobium (Nb) and tungsten (W) metal recovery plants. From these streams, Sc can be
recovered by various processes such as leaching, chemical precipitation, solvent extrac-
tion and/or ion exchange.
https://doi.org/10.1007/978-3-031-17425-4_91
812 G. Abrenica et al.
With the global generation of bauxite residues, highly alkaline waste solids from the
extraction of alumina from bauxite ore using the Bayer process are estimated at 150
million tons per year [2], and the accumulated inventory has been estimated at 3 billion
tons [3]. This is equivalent to a potential additional Sc supply of about 270 thousand
tons. Not only will it provide a substantial Sc source, but it would also mitigate the
environmental concerns of fertile soil and groundwater contamination, and risks to
human health. Another substantial source of Sc is the acid waste from TiO2 pigment
production plants, via chloride and sulfate routes, which is estimated to reach a total
capacity of 8.4 million tons worldwide. Based on the average Sc concentration of
15–25 mg/kg, this waste stream can potentially provide about 200 tons of Sc per year
[4].
Sc is still not widely used due to its scarcity and high cost of production. The current
world’s supply and demand of Sc in the form of scandium oxide (Sc2O3) are only about
20–25 tons per year, and the demand is projected to increase to 1000 tons per year in
2025. This sharp increase in Sc demand in the very near future is driven by major fac-
tors such as accelerating usage in solid oxide fuel cells (SOFCs) and the rising demand
for aluminum-scandium alloys (Al-Sc). The rising demand for clean energy due to
environmental concerns associated with the use of coal and natural gas is expected to
propel the Sc demand for SOFCs. Meanwhile, the growing application of Al-Sc light-
weighting in the aerospace industry and the fast-growing electric vehicle market are
both expected to take a large share in this demand.
1.2 II-VI Capabilities
II-VI, headquartered in Saxonburg, Pennsylvania, has research and development, man-
ufacturing, sales, service, and distribution facilities worldwide. The Company produces
a wide variety of application-specific photonic and electronic materials and compo-
nents and deploys them in various forms. With its mission of “enabling the world to be
safer, closer, healthier, and more efficient”, is also a global leader in rare metals refin-
ing. II-VI has an excellent track record in environmental, health and safety (EHS) and
quality excellence. II-VI has more than 30 years of extensive expertise in extractive
metallurgy with competent and experienced engineers and chemists as the centerpiece
of various new product and process developments.
One such project that had gone to successful commercialization was the recovery of
Sc from fluorite mine tailings. The developed process which is a combination of pyro-
metallurgical and hydrometallurgical concepts starts from reacting the insoluble sili-
ceous fluorite mine tailings.
II-VI’s Capabilities in Scandium Recovery and Purification 813
2 II-VI R&D and Process Developments in Hydrometallurgy
2.1 Development of II-VI Selective-Ion Recovery for the Recovery of Sc

Wastes Materials
In the pursuit of recovering Sc, II-VI developed the Selective-Ion Recovery (SIR)
Technology [5, 6] that includes the use of a composite extractant-enhanced ion-
exchange resin to extract Sc from acidic solution or slurries, and its subsequent recov-
ery as an Sc concentrate. The factors that contribute to its performance have been
studied to a great extent using pregnant leach solution (PLS) from various Sc-containing
sources such as bauxite residues and acid waste from TiO2 production.
Recovery of Sc from Bauxite Residues

Bauxite residue (BR) contains various major metal oxides of iron (Fe), aluminium (Al),
Ti, calcium (Ca), silicon (Si), and sodium (Na), as well as minor metal oxides like vana-
dium (V), gallium (Ga), Sc and other rare earth elements (REE). The typical concentra-
tion of Sc in the bauxite residue is 70–260 mg/kg [7] while the REE is 800–2500 mg/
kg [8]. Several experiments were conducted on the laboratory scale to leach Sc from
this residue. A combination of parameters such as acid concentration, leaching tem-
perature, leaching time, and pulp density was tested. The required amounts of acid were
calculated based on the volume of the leaching solution and the target acid concentra-
tions. The initial acid concentrations were determined by acid-base titration. The leach-
ing temperature and time were monitored during the entire duration of the tests. The
concentrations of Sc and critical elements such as Ti, Fe and Si in the solution were
measured using inductively coupled plasma – optical emission spectrophotometry
(ICP-OES).
High acid leaching tests at 3 M H2SO4 initial acid concentration in the leachant were
carried out to increase Sc recovery while reducing Si concentration in the PLS and
avoiding silica gel formation during the Sc loading on the SIR resin. However, that
resulted in the generation of a PLS with very high impurity such as Fe and Ti levels
relative to Sc that interfered with Sc loading. Various methods such as selective precipi-
tation, reduction-oxidation reaction, and chelation were applied to reduce the concen-
tration of impurities while maintaining Sc in the solution. Considering the results of
these experiments, impurity removal from the high acid PLS was not efficient.
Low acid leaching tests at 1 M H2SO4 initial acid concentration in the leachant were
also carried out with the aim of reducing levels of impurities in the PLS. The levels of
impurities were low under low acid leaching conditions; however, it was found that the
Si concentration was very high. The high silica concentration led to silica precipitation
in the columns and created pressure drop and fouling. For this reason, a series of tests
were conducted to remove silica from the solution during leaching. It was found that
leaching the BR for an extended time at higher temperatures causes silica to precipitate
from the solution together with Ti without significantly affecting Sc.
The concentrations of Sc, Fe, Ti and Si in the PLS from the high acid and low acid
leaching tests are shown in Table 1. The solutions were fed to the SIR column at 2 bed
volumes per h (BV/h) and the raffinates were sampled and analysed using ICP-OES. The
Table 1 Concentrations of Sc, Fe, Ti, and Si in the PLS from high acid and low acid leaching tests
High acid leaching Low acid leaching
Initial H2SO4 concentration, M 3 1 1
Leaching time, hour 1 1 4a
Leaching temperature, °C 80–90 80–90 80–90
Concentration in the PLS:
Sc (mg/L) 9 11 10
Fe (g/L) 19 3 3
Ti (g/L) 3 0.5 0.04
Si (g/L) 0.068 4.2 0.8
Plus 16 hours at 95 °C for maximum removal of silica and Ti
a
Fig. 1 Process flowchart of II-VI developed optimum leaching process

loading graphs at high and low acid concentrations are shown in Figs. 1 and 2. Sc con-
centration was around 10 ppm regardless of acid concentrations. The loading of Sc to
the SIR resin was low at 60 and 500 mg/L resin, respectively. This was assessed to be
mainly due to the significantly higher amount of Fe, and to some extent the amount of
Ti, in the presence of extremely low Sc concentration.
II-VI developed a two-stage optimum leaching process to maximize the dissolution
of Sc while minimizing the dissolution of the other impurities. BR was leached at 0.4 M
H2SO4, 20% pulp density, 80–90 °C, and 1 hour, leaching time. At this acid concentra-
tion, the impurity levels were sufficiently low. In order to increase Sc concentration, the
filtrate was recycled for leaching and used as starting solution for BR leaching. The acid
concentration was adjusted to achieve a 0.4 M initial concentration. In order to increase
the recovery of Sc from the BR, the residue was re-leached. To avoid problems with the
precipitation of metals and/or silica gel formation, the PLS was acidified prior to load-
ing to the SIR. The general flow chart for the said concept is shown in Fig. 1.
The analysis of Sc, Fe, Ti and Si in the PLS from this two-stage leaching process is
shown in Table 2.
The final PLS, in this case, from the first stage of the two-stage optimum leaching
conditions proceeded to the SIR for Sc recovery. As shown in Figs. 1 and 2, the SIR
loading capacity increased dramatically to 5000 mg/L, and the breakthrough point for
Sc up to 200 BVs for the PLS generated from leaching BR at optimum conditions
(Fig. 3).
The eluate proceeded to the downstream treatment steps of acidification followed by
neutralization to precipitate Sc. The crude Sc concentrate contained 20–25 wt% Sc.
Additionally, the SIR Technology pilot plant was designed and constructed. The
pilot plant was composed of three skids that contained the SIR resins and its columns,
tanks for loading, elution and regeneration. The recovery of Sc from BR was evaluated
Fig. 2 Loading of Sc on to II-VI SIR form PLS leached at different conditions

Table 2 Concentrations of Sc, Fe, Ti, and Si in the PLS from II-VI developed two-stage optimum
leaching process
First stage Second stage (Re-leaching)
Initial PLS Final PLS Initial PLS Final PLS
Sc (mg/L) 5 9 3.3 6
Fe (mg/L) 42 99 168 414
Ti (mg/L) 0.53 3 8 16
Si (mg/L) 210 258 334 356
Fig. 3 Sc loading isotherm
at the pilot scale. 10 m3 PLS was generated and fed to the SIR columns. The Sc was then
eluted from the resin and a crude concentrate containing 22% Sc was generated.
Through the leaching and purification, Sc was concentrated more than 2500 times
from BR to a crude Sc concentrate. The significant increase in Sc content substantially
reduces the sizing requirements for downstream purification steps and presents a prag-
matic approach to the recovery of Sc from BR.
Recovery of Sc from Acidic Wastes from TiO2 Pigment Production

The acidic waste or Ti tailings solution was obtained from a TiO2 production plant
employing carbothermal chlorination processing of Ti ores. As a preliminary evaluation
of the effect of the other metal cations particularly Ti on the loading capacity of the SIR
for Sc, the Ti tailings solution containing about 80 g/L hydrochloric acid (HCl), 80 mg/L
Sc, 5000 mg/L Ti, and other transition metal ions was contacted with the SIR. The
adsorption capacity of the SIR for Sc and Ti was 1000 and 51,000 mg/L, respectively.
Considering the high loading of Ti, it was decided to implement a pre-treatment step for
Ti removal to maximize the loading efficiency of the SIR for Sc.
Fig. 4 Loading of Sc onto II-VI SIR from Ti tailing solutions
For the separation of Ti and Sc, precipitation experiments were conducted. It was
found that the optimum pH for the preferential precipitation of Ti was at 0.70 ± 0.10
with more than 98% of the Ti precipitated while more than 80% of Sc stayed in the
solution. This solution was subjected to the succeeding adsorption experiments to eval-
uate the performance of the SIR. As shown in Fig. 4, the SIR resin was loaded to
exhaustion using Ti tailings solution and the loading capacity of 20,000 mg Sc/L was
obtained. The SIR was eluted and regenerated according to the established parameters.
The eluate was further treated as previously described to recover the crude Sc concen-
trate which contained about 20% Sc and 0.5% Ti. The recovery of Sc in the elution and
Sc hydroxide precipitation steps were 99% and 97%, respectively.
Recovery of Sc from Other Waste Streams

In addition to exploring the applicability of the SIR process to recover Sc from Ti tail-
ings solution in chloride medium, recovery of Sc from Ti tailings in sulfate medium was
explored. The process was tailored to meet the criteria for existing plant design and
customer’s requirements. Similarly, Sc recovery from Ni/Co solutions was also
explored. Overall, successful Sc recovery via SIR process using various sources was
demonstrated.
Fig. 5 Incorporation of the II-VI developed recovery and purification processes
2.2 Purification of Crude Sc Concentrate Generated from the Developed

Process
The crude Sc concentrate generated from the SIR process can be processed via the II-VI
developed purification process as shown in Fig. 5. Depending on the existing impuri-
ties, modifications in some process steps might be necessary.
2.3 Recovery and Purification of Sc from Fluorite Mine Tailings

The fluorite mine tailings are composed of thortveitite, quartz, fluorite, feldspars, horn-
blende and other oxide and silicate materials, with an average Sc concentration of
8000 mg/kg. As Sc is present in the silicate matrix, its analysis presented the first chal-
lenge. Various analytical techniques such as dissolution with mixed strong acids and
fusion were tested and developed. It was the latter with relatively less risks to health and
environmental safety that provided the direction that the process development would go
to free up the Sc from the matrix.
As shown in Fig. 4, after reacting the raw material with caustic soda, the Sc-containing
solid is then leached in acid to solubilize the Sc along with the impurities such as other
rare earth metals, Fe, Ti and trace amounts of Th and U. After the reaction and stabiliza-
tion time is achieved, the Sc-containing liquor is separated from the residue. This liquor
is then contacted with an organic extractant with which Sc forms an acid-insoluble
complex. The Sc-containing organic is contacted with a series of scrubbing stages using
various acid solutions that completely remove co-extracted REEs, Fe, Th, U and Ti,
then with an alkaline solution to strip Sc in the form of Sc hydroxide precipitate. The
Sc is dissolved from the said precipitates, further purified by oxalation, and calcined to
produce >99% Sc2O3. The facility, located in the Philippines, includes a rotary-tilting
furnace, mixer-settlers, agitated tanks and calciners and has a production capacity of
7200 kg Sc2O3 per year.
3 Technical Summary and Future Prospects
The increasing demand for Sc in many fields of application has caused numerous pro-
cess development efforts toward recovering Sc from readily available sources such as
bauxite residues, Ti tailings and Ni/Co solutions. Overcoming the technical challenges
posed by the low concentrations of Sc and complex chemistries of these sources while
aiming for economically viable recovery & purification processes has been the main
objective of the development of a series of patented processes by II-VI.
The applicability of the SIR process in the recovery of Sc from BR was investigated.
High acid leaching tests were carried out to increase Sc recovery while reducing Si
concentration in the PLS and avoiding silica gel formation during the SIR loading.
However, that resulted in the generation of a PLS with high impurity levels that inter-
fered with Sc loading. Low acid leaching tests were also carried out with the aim of
reducing levels of impurities; however, it was found that the Si concentration was very
high, which led to silica precipitation in the columns and created pressure drop and
fouling. To improve Sc loading, the leaching conditions were optimized such that the
PLS contained the maximum concentration of Sc while maintaining the impurities to a
minimum. The silica precipitation and fouling issue was also addressed in this process.
The PLS produced under optimum conditions proceeded to the SIR for Sc recovery.
The loading capacity increased drastically. This increase in the loading capacity trans-
lates to significant improvement in the economics and viability of the process. The Sc
recovery from BR was validated at the pilot scale.
II-VI also evaluated the applicability of the SIR process in the recovery of Sc from
the Ti tailings solution. A pretreatment stage was implemented to further improve Sc
recovery from the waste solution. A significant increase in Sc loading capacity was
observed.
The crude Sc concentrates produced by eluting the Sc from the loaded resin and
subsequently precipitating it had 20–25% Sc. This significant increase in Sc content
substantially reduces the sizing requirements for the downstream purification step and
presents a pragmatic approach to the recovery of Sc from the said complex sources.
These concentrates can be processed via the II-VI-developed purification process to
produce >99% Sc2O3.
II-VI welcomes opportunities to license or partner with companies interested in
recovering Sc from industrial streams. II-VI’s vision is to expand industrial applications
for Sc and its alloys as key materials for high energy efficiency and mechanical strength.
II-VI’s R&D is engaged in developing innovative ideas to address the current and
potential opportunities to satisfy the demands of the growing world and is continually
exploring different avenues for applying its refining and purification experience and
expertise.
Acknowledgments The research leading to these results has received funding from the European
Community’s Horizon 2020 SCALE Programme (H2020/2014-2020/No. 730105). This publication
reflects only the authors’ views, exempting the SCALE Consortium from any liability for the informa-
tion presented herein.
References
1. U.S. Geological Survey. Mineral commodity summaries 2021. January 2021.

2. Evans K. The history, challenges, and new developments in the management and use of bauxite
residue. J Sustain Metall. 2016;2:316–31.
3. Binnemans K, Jones PT, Blanpain B, Van Gerven T, Pontikes Y. Towards zero-waste valorisation
of rare-earth-containing industrial process residues: a critical review. J Clean Prod. 2015;99:17–38.
4. Yagmurlu B, Orberger B, Dittrich C, Croisé G, Scharfenberg R, Balomenos E, Panias D, Mikeli E,
Maier C, Schneider R, Friedrich B, Dräger P, Baumgärtner F, Schmitz M, Letmathe P, Sakkas K,
Georgopoulos C, van den Laan H. Sustainable supply of scandium for the EU industries from liquid
iron chloride based TiO2 plants. Mater Proc. 2021;5:86.
5. Xu W-Q, Mattera Jr. V, Abella MYR, Abrenica GM, Patkar S. Selective recovery of rare earth met-
als from an acidic slurry or acidic solution. US20190078175A1 and WO2019099859A1. 2019.
6. Xu W-Q, Mattera Jr. V, Abella MYR, Abrenica GM, Patkar S. Composite extractant-enhanced
polymer resin, method of making the same, and its usage for extraction of valuable metal(s). WO
2017/074921A1. 2017.
7. Chaikin L, Shoppert A, Valeev D, Loginova I, Napol’skikh J. Concentration of rare earth elements
(Sc, Y, La, Ce, Nd, Sm) in bauxite residue (red mud) obtained by water and alkali leaching of baux-
ite sintering dust. Minerals. 2020;10:500.
8. Deady ÉA, Mouchos E, Goodenough K, Williamson BJ, Wall F. A review of the potential for
rare-earth element resources from European red muds: examples from Seydişehir, Turkey and
Parnassus-Giona, Greece. Mineral Mag. 2018;80(01):43–61.
Part VII Towards Sustainable Circularity:
Mining to Materials
Selective Depression of Talc in Cuprite Sulfidation
Flotation by New Depressant Astragaloside
Qiuyue Sheng1, 2, Wanzhong Yin1 (*), Yahui Zhang2 (*), and Kelly Hawboldt2
1
School of Resources and Civil Engineering, Northeastern University, Shenyang,
Chinayinwanzhong@mail.neu.edu.cn
2
Faculty of Engineering and Applied Science, Memorial University of Newfoundland,
St. John’s, NL, Canada
yahuiz@mun.ca
Abstract. Talc is a common gangue mineral with natural hydrophobicity, which may
lower grades of concentrates in metal mineral flotation and cause problems during smelt-
ing. The depression of talc is a common issue in flotation processes and has attracted
attention, the separation between talc and molybdenite ores, nickel sulfide ores has been
reported. As an efficient and environmentally friendly depressant of talc, astragaloside
was firstly used to separate cuprite and talc in this study. The effect of astragaloside on talc
depression and cuprite flotation was studied using Na2S as sulfidation agent and sodium
butyl xanthate (NaBX) as collector. The flotation and contact angle test results indicated
that the hydrophobicity and floatability of talc were weakened after astragaloside addi-
tion. The recovery of talc decreased from 83.46% to 37.98% with the addition of 40 mg/L
astragaloside, but the floatability of cuprite was hardly affected. To explore the adsorption
mechanism of astragaloside, X-ray photoelectron spectroscopy and Fourier transform
infrared spectroscopy analyses were carried out. The results indicated that astragaloside
was selectively adsorbed on the talc surface, thus the flotation separation of cuprite and
talc was achieved, and the concentrate grade in the sulfidation flotation of cuprite was
improved.
Keywords: Astragaloside · Depressant · Talc · Cuprite · Flotation separation
1 Introduction
As a common gangue mineral with natural hydrophobicity, talc is normally found in

nickel sulfide ores, molybdenite ores and copper oxide ores [1–3]. In fact, when talc
particles are crushed and ground as flotation feed, the basal surface accounts for about
90% of the total surface of talc particles, and the basal face is hydrophobic [4, 5], which
does not have hydrophilic ions, thus talc is easily concentrated with excellent floatabil-
ity [6].
Talc may influence the grade of concentrate during the flotation of a metal mineral.
At present, the depression of talc has been studied, and mostly through the addition of
selective depressant of talc. According to previous reports, the separation between talc
https://doi.org/10.1007/978-3-031-17425-4_92
824 Q. Sheng et al.
and chalcopyrite, molybdenite, scheelite as well as pyrite has been studied [7–10].
Several reagents, including organic and inorganic depressants, have been used to
weaken the floatability of talc. Selective adsorption of depressants on talc mineral sur-
face is the key to separate it from other minerals. Based on current studies, organic
reagents such as guar gum, carrageenan, locust bean gum, xanthan gum, carboxymethyl
cellulose and lignosulfonates can be used to depress talc effectively [6, 11–15]. Some
researchers also indicated that the presence of metal cations such as Ca2+ and Al3+ can
increase the adsorption sites of reagents on talc surface [9, 16], thus promote the flota-
tion separation between minerals.
Talc also often exists in copper oxide ores and may decrease the Cu grade of concen-
trate during the flotation of copper oxide minerals [17]. At present, although the separa-
tion between copper oxide minerals and gangue minerals such as calcite as well as
quartz has achieved certain success [18, 19], further exploration of flotation separation
between copper oxide minerals and talc by more efficient and environmentally friendly
depressants is still required. In practice, the concentration of copper oxide minerals is
mainly by direct or sulfidation flotation [20, 21]. Due to the poor selectivity, the appli-
cation of direct flotation is limited [22]. Sulfidation flotation is the more widely used
progress for the recovery of copper oxide minerals [23, 24]. Up to now, sodium sulfide
(Na2S) and sodium hydrosulfide (NaHS) have been used as sulfidation reagents for the
sulfidation of oxide mineral surfaces [25, 26]. Therefore, it is of significance to study
the depression of talc during the sulfidation flotation of copper oxide minerals.
In this work, astragaloside was used as a depressant for talc in the sulfidation flota-
tion of cuprite with the sulfidation agent Na2S and collector NaBX. Flotation tests were
carried out to identify the selective depression of astragaloside on talc. Further, contact
angle measurements, Fourier transform infrared (FTIR) spectroscopy and X-ray photo-
electron spectroscopy (XPS) analysis were conducted to study the adsorption mecha-
nism of astragaloside on the surface of talc. This work can be a reference for improving
concentrate grade during the flotation of copper oxide minerals.
2.1 Samples and Reagents
Cuprite (Cu2O) and talc (3MgO⋅4SiO2⋅H2O) mineral samples were crushed, ground,
screened, and classified into the desired particle sizes. The −106 + 45 μm fraction was
selected for the flotation tests and x-ray photoelectron spectroscopy (XPS) analysis,
and size fractions of −45 μm were ground to −5 μm for FTIR measurements. The X-ray
diffraction (XRD) analysis and chemical compositions of the cuprite and talc samples
are shown in Fig. 1 and Table 1, respectively, which indicate the high purity of mineral
samples.
In this work, analytical grade Na2S⋅9H2O and commercial grade sodium butyl xan-
thate (NaBX, C4H9OCSSNa) were used as the sulfidation agent and the collector,
respectively. Astragaloside with analytical grade was used as the depressant, its chemi-
Selective Depression of Talc in Cuprite Sulfidation Flotation by New Depressant… 825
cal structure is shown in Fig. 2. The pulp pH was regulated by hydrochloric acid
(analytical grade) and NaOH (analytical grade) solutions. Deionized water was used in
all tests.

(a) Cuprite
Intensity (Counts)

10 20 30 40 50 60 70 80
Two-theta (deg)
(b)
• • Talc
Intensity (Counts)
•
•
10 20 30 40 50 60 70
Two-theta (deg)
Fig. 1. XRD patterns of cuprite (a) and talc (b) samples.
826 Q. Sheng et al.
Table 1. Chemical analysis of cuprite and talc samples (%).

Minerals Cu Fe S MgO SiO2
Cuprite 81.03 2.07 0.21 0.19 0.55
Talc – 1.10 – 30.14 60.71
OH
HO OH
OH
HO
O O
O OH
HO O OH
O
O
H3C O OH
HO
Fig. 2. Chemical structure of astragaloside.
2.2 Micro-Flotation Tests
Micro-flotation tests were performed in an XFG laboratory flotation cell (Jilin

Exploration Machinery Plant, China) with a volume capacity of 40 mL and a stirring
speed of 1700 rpm. As shown in Fig. 3, 1.0 g of cuprite and 1.0 g of talc were used as
the flotation sample, which was mixed with 40 ml of deionized water in the flotation
cell, and the depressant astragaloside was added. The pulp was firstly stirred for 3 min,
and the pH was adjusted to the desired value by the addition of HCl or NaOH solutions.
Then, sulfidation agent Na2S solution and collector NaBX solution were added sequen-
tially and allowed to react for 3 min in each step. After that, frother 2# oil was added
and allowed to adjust for 1 min. Finally, flotation was conducted for 3 min, the concen-
trate was collected, dried and weighed, and the Cu and MgO content in the concentrate
was assayed. Based on Eq. (1), the mineral flotation recovery was calculated. Each
flotation test was carried out at least three times, and the average values and standard
deviations were recorded.
mc 1
M 100% (1)
mc 1 mt 2
Where εM represents the recovery of cuprite or talc. β1 (wt%) and β2 (wt%) express the
grade of Cu or MgO in the concentrate and tailing product, respectively. mc (g) and mt
(g) are the mass of the concentrate and tailing, respectively.
2.0 g Sample Astragalus

polysacharin
Stirring, pH
3′
Regulator
3′ Na2S
3′ NaBX
1′ 2# oil
3′ flotation
Concentrate Tailing
Fig. 3. Flowsheet of flotation test.
2.3 Contact Angle Measurements
The mineral sample used for contact angle measurement was pressed with a jack to
make a smooth mineral sheet, and then dropping a water droplet with a diameter of
about 2–3 mm on it through a microsyringe on a JC2000A contact angle measuring
instrument (Powereach Instruments, Shanghai, China) [27]. Before that, the mineral
sample in each measurement was prepared as follows: 2.0 g of talc or cuprite mineral
sample was treated with desired reagents, which was consistent with flotation tests.
Then the solid-liquid separation was conducted, and the solid particles were collected
and dried for the test. Each contact angle measurement was performed three times, and
the average value and standard deviations were reported [27].
2.4 FTIR Measurement
In each measurement, 2.0 g of mineral sample was conditioned with desired reagents
under a specific pH value and settled for about 5 min. After the solid–liquid separation,
the obtained solid fraction was filtered, washed and dried for subsequent FTIR mea-
surements. The FTIR spectra of astragaloside and mineral samples in the absence and
presence of astragaloside were measured. The infrared spectra were measured through
a Nicolet 380 FT-IR spectrometer at 298 ± 2 K in the range of 4000–500 cm−1 [28].
828 Q. Sheng et al.
2.5 XPS Measurement
The XPS characterization was performed using a Thermo VG ESCALAB 250Xi spec-
trometer with Al α X–rays (1486.6 eV) as the sputtering source at a power of 150 W
(15 kV, 10 mA), and the binding energy of C1s at 284.8 eV was used to correct charge.
Mineral sample for each XPS measurement was prepared as follows: 2.0 g of mineral
samples were conditioned with 40 mL of deionized water, and astragaloside solution
with a desired concentration was added and allowed to react for 3 min. Finally, solid–
liquid separation was conducted, and the solid mineral particles were collected, dried
and sent for measurements [29].
3.1 Micro-Flotation Tests
To verify the selective depression of astragaloside on talc during the sulfidation flota-
tion of cuprite, micro-flotation tests were performed with artificial mixtures of minerals
at various astragaloside dosages using 100 mg/L Na2S and 200 mg/L NaBX as sulfida-
tion agent and collector, respectively. The results are shown in Fig. 4. When the dosage
of astragaloside increased from 10 to 20 mg/L, the flotation recovery of talc decreased
from 83.46% to 60.05%, and as 40 mg/L astragaloside was added, the recovery of talc
was only 37.98%. Therefore, the floatability of talc was weakened remarkably in the
Cuprite
80 Talc
60
Recovery (%)
40
20
0
10 20 30 40 50
Astragaloside (mg/L)
Fig. 4. Effect of astragaloside dosage on the floatability of cuprite and talc from artificial cuprite–talc
mixtures.
presence of astragaloside. However, when the amount of astragaloside was in the range
of 10–50 mg/L, the recovery of cuprite was around 65%, the floatability of which was
slightly influenced by the addition of astragaloside. Thus, the separation between talc
and cuprite can be realized effectively.
3.2 Contact Angle Tests
Based on the results of flotation tests, astragaloside can selectively depress the flotation
of talc and realize the separation between it and cuprite. To study the influence of
astragaloside on the properties of mineral surfaces, contact angle measurements were
conducted.
As shown in Table 2, in the presence of 100 mg/L Na2S and 200 mg/L NaBX, the
contact angle of talc surface decreased from 68.13° to 40.90° with the addition of
40 mg/L astragaloside, its hydrophobicity was weakened. Thus, the floatability of talc
was decreased. According to the flotation test, the recovery of talc was 37.89% under
this condition. By comparison, in the absence and presence of astragaloside, the contact
angles of cuprite surface were 66.70° and 65.18°, respectively. Its hydrophobicity was
slightly changed, which was consistent with the test result of flotation.
3.3 FTIR Analysis
In order to determine the adsorption mechanism of astragaloside on the surface of talc,

FTIR analysis was conducted. The FTIR spectra of astragaloside and the talc samples
before and after the treatment with astragaloside are shown in Figs. 5 and 6,
respectively.
According to the measurement result and previous reports, the peak located around
3408.65 cm−1 in Fig. 5 contributed to the stretching vibration of −OH in astragaloside
[30]. As shown in Fig. 6a, FTIR spectra of the talc surface were depicted. For the talc
surface, the peaks at 1022.53 and 664.94 cm−1 were attributed to the vibration of Si-O
and Mg-O, respectively [6]. According to Fig. 6b, the addition of astragaloside brings
up a new absorption peak at 3410.3 cm−1, corresponding to the peak of −OH in Fig. 5,
which identified the adsorption of astragaloside on the talc surface.
3.4 XPS Analysis
XPS analysis was implemented to further identify the adsorption mechanism of astraga-
loside on the talc surface. The XPS narrow scanning spectra of C, Mg and Si were
obtained before and after the treatment with astragaloside and the results are given in
Figs. 7, 8 and 9.
Table 2. Contact angles of cuprite and talc before and after the addition of astragaloside (°).
Minerals In the absence of astragaloside In the presence of astragaloside (40 mg/L)
Cuprite 66.70 65.18
Talc 68.13 40.90
830 Q. Sheng et al.
3408.65
3500 3000 2500 2000 1500 1000 500

Wavenumber (cm-1)
Fig. 5. FTIR spectra of astragaloside.
(b)
(a)
664.94
3410.3
3418 3416 3414 3412 3410 3408 3406 3404 3402

Wavenumber (cm-1) 1022.53
3500 3000 2500 2000 1500 1000 500

Wavenumbers (cm-1)
Fig. 6. FTIR spectra of the talc surface before (a) and after (b) the treatment with astragaloside.
As shown in Fig. 7a and b, the peaks that emerged around 284.82 and 284.81 eV
were attributed to the organic contamination [7]. In Fig. 7b, in the presence of astraga-
loside, two new peaks appeared at 286.39 and 287.84 were corresponding to C peaks
of C–O–C and C=O, respectively [31, 32], which identified the adsorption of astraga-
loside on the surface of talc. In addition, from Figs. 8 and 9, before and after the addi-
Fig. 7. XPS narrow scanning spectra of C on talc surface before (a) and after (b) astragaloside
treatment.
tion of astragaloside, the binding energy spectra of Mg and Si experienced few shifts,
which may suggest that the adsorption of astragaloside on the talc surface was physical
[2].
832 Q. Sheng et al.
(a) 1304.77
1308 1306 1304 1302 1300 1298

Binding energy (eV)
(b) 1304.41
1308 1306 1304 1302 1300 1298

Binding energy (eV)
Fig. 8. XPS narrow scanning spectra of Mg on talc surface before (a) and after (b) astragaloside
treatment.
(a) 103.77
110 108 106 104 102 100 98 96

Binding energy (eV)
(b) 103.43
110 108 106 104 102 100 98 96

Binding energy (eV)
Fig. 9. XPS narrow scanning spectra of Si on talc surface before (a) and after (b) astragaloside
treatment.
4 Conclusion
Copper oxide minerals can be recovered by sulfidation flotation effectively, but gangue
mineral talc may influence the concentrate grade in this process. In this work, astraga-
loside, an environmentally friendly and highly efficient organic reagent, can selectively
weaken the hydrophobicity of talc surface, thus decreasing the flotation recovery of talc
during the sulfidation flotation of cuprite. The recovery of talc decreased to 37.98%
with the addition of 40 mg/L astragaloside, but the floatability of cuprite was little
834 Q. Sheng et al.
influenced. The results of X-ray photoelectron spectroscopy and Fourier transform

infrared spectroscopy analyses indicated that astragaloside adsorbed on the talc surface,
thus the flotation separation of cuprite and talc was achieved. This study can provide a
reference for improving the recovery of copper oxide minerals. But in practice, the talc
slimes (−10 μm) may be difficult to be depressed. Talc slimes always influence the
floatability of copper oxide minerals, including the effects on concentrate recovery and
Cu grade. In the future, we will research the influence mechanism of talc slimes on the
flotation of copper oxide minerals (malachite, azurite and cuprite), and study the effec-
tive inhibition of talc slimes.
Acknowledgments This work was financially supported by the National Natural Science Foundation
of China (Grant Nos. 51874072 and 51404300).
References
1. Feng B, Lu Y, Feng Q, Zhang M, Gu Y. Talc–serpentine interactions and implications for talc

depression. Miner Eng. 2012;32(4):68–73.
2. Feng B, Peng J, Zhang W, Ning X, Guo Y, Zhang W. Use of locust bean gum in flotation separation
of chalcopyrite and talc. Miner Eng. 2018;122:79–83.
3. Yuan D, Xie L, Shi X, Yi L, Zhang G, Zhang H. Selective flotation separation of molybdenite and
talc by humic substances. Miner Eng. 2018;117:34–41.
4. Morris GE, Fornasiero D, Ralston J. Polymer depressants at the talc–water interface: adsorption
isotherm, microflotation and electrokinetic studies. Int J Miner Process. 2002;67(1):211–27.
5. Jin S, Shi Q, Li Q, Ou L, Ouyang K. Effect of calcium ionic concentrations on the adsorption of
carboxymethyl cellulose onto talc surface: flotation, adsorption and AFM imaging study. Powder
Technol. 2018;331:155–61.
6. Zhou H, Zhang ZJ, Oua LM, Mai QY. Flotation separation of chalcopyrite from talc using a new
depressant carrageenan. Colloids Surf A Physicochem Eng Asp. 2020;603:125274.
7. Zhong CH, Wang HH, Zhang LZ, Guo MC, Feng B. Flotation separation of molybdenite and talc
by xanthan gum. Powder Technol. 2021;388:158–65.
8. Liu C, Zhang WC, Song SX, Li HQ. A novel method to improve carboxymethyl cellulose perfor-
mance in the flotation of talc. Miner Eng. 2019;131:23–7.
9. Jin S, Shi Q, Feng QM, Zhang GF, Chang ZY. The role of calcium and carbonate ions in the separa-
tion of pyrite and talc. Miner Eng. 2018;119:205–11.
10. Zhao KL, Gu GH, Wang XH, Yan W, Hu YH. The effect of depressant sesbania gum on the flota-
tion of a talc-containing scheelite ore. J Mater Res Technol. 2019;8(1):14–21.
11. Ma XD, Pawlik M. The effect of lignosulfonates on the floatability of talc. Int J Miner Process.
2007;83:19–27.
12. Claudio A, Garcia V, Marek P. Molecular weight effects in interactions of guar gum with talc. Int
J Miner Process. 2015;138:38–43.
13. Agnieszka MV, David AB. Adsorption of tailored carboxymethyl cellulose polymers on talc and
chalcopyrite: correlation between coverage, wettability, and flotation. Miner Eng. 2010;23:985–93.
14. Pan GJ, Shi Q, Zhang GF, Huang GH. Selective depression of talc in chalcopyrite flotation by
xanthan gum: flotation response and adsorption mechanism. Colloids Surf A Physicochem Eng
Asp. 2020;600:124902.
15. Feng B, Zhang WP, Guo YT, Peng JX, Ning XH, Wang HH. Synergistic effect of acidified
water glass and locust bean gum in the flotation of a refractory copper sulfide ore. J Clean Prod.
2018;202:1077–84.
16. Fu YF, Zhu ZL, Yao J, Han HL, Yin WZ, Yang B. Improved depression of talc in chalcopyrite flo-
tation using a novel depressant combination of calcium ions and sodium lignosulfonate. Colloids
Surf A Physicochem Eng Asp. 2018;558:88–94.
17. Farrokhpay S, Ndlovu B, Bradshaw D. Behavior of talc and mica in copper ore flotation. Appl Clay
Sci. 2018;160:270–5.
18. Liu C, Zhu GL, Song SS, Li HQ. Interaction of gangue minerals with malachite and implications for
the sulfidization flotation of malachite. Colloids Surf A Physicochem Eng Asp. 2018;558:679–84.
19. Liu C, Song SX, Li HQ, Li YB, Ai GH. Elimination of the adverse effect of calcite slimes on the
sulfidization flotation of malachite in the presence of water glass. Colloids Surf A Physicochem
Eng Asp. 2019;563:324–9.
20. Chol J, Chol SQ, Park K, Han Y, Kim H. Flotation behaviour of malachite in mono- and di-valent
salt solutions using sodium oleate as a collector. Int J Miner Process. 2016;146:38–45.
21. Chen Y, Liu M, Chen JH, Li YQ, Zhao CH, Mu X. A density functional based tight binding
(DFTB plus) study on the sulfidization-amine flotation mechanism of smithsonite. Appl Surf Sci.
2018;458:454–63.
22. Huang YG, Niu XX, Liu GY, Liu J. Novel chelating surfactant 5-heptyl-1,2,4- triazole-3-
thione: its synthesis and flotation separation of malachite against quartz and calcite. Miner Eng.
2019;131:342–52.
23. Feng QC, Zhao WJ, Wen SM, Cao QB. Copper sulfide species formed on malachite surfaces in
relation to flotation. J Ind Eng Chem. 2017;48:125–32.
24. Feng QC, Wen SM, Zhao WJ, Deng JS, Xian YJ. Adsorption of sulfide ions on cerussite surfaces
and implications for flotation. Appl Surf Sci. 2016;360:365–72.
25. Sheng QY, Yin WZ, Yang B, Cao SH, Sun HR, Ma YQ, Chen KQ. Improving surface sulfidiza-
tion of azurite with ammonium bisulfate and its contribution to sulfidization flotation. Miner Eng.
2021;171:107072.
26. Liu C, Zhang WC, Song SX, Li HQ, Jiao XK. A novel insight of the effect of sodium chloride on
the sulfidization flotation of cerussite. Powder Technol. 2019;344:103–7.
27. Sheng QY, Yin WZ, Ma YQ, Liu YT, Wang LY, Yang B, Sun HR, Yao J. Selective depression of talc
in azurite sulfidization flotation by tamarind polysaccharide gum: flotation response and adsorp-
tion mechanism. Miner Eng. 2022;178:107393.
28. Wang X, Zhao B, Liu J, Zhu YM, Han YX. Dithiouracil, a highly efficient depressant for the selec-
tive separation of molybdenite from chalcopyrite by flotation: applications and mechanism. Miner
Eng. 2022;175:107287.
29. Tang Y, Yin WZ, Kelebek S. Magnesite-dolomite separation using potassium cetyl phosphate as a
novel flotation collector and related surface chemistry. Appl Surf Sci. 2020;508:145191.
30. Zhou HP, Yang ZZ, Tang XK, Sun W, Gao ZY, Luo XP. Enhancing flotation separation effect of flu-
orite and calcite with polysaccharide depressant tamarind seed gum. Colloids Surf A Physicochem
Eng Asp. 2021;624:126784.
31. Arai J. No-flash-point electrolytes applied to amorphous carbon/Li1+xMn2O4 cells for EV use. J
Power Sources. 2003;119–121:388–92.
32. Li Y, Zhou W, Wang H, Xie L, Liang Y, Wei F, Idrobo JC, Pennycook SJ, Dai H. An oxygen
reduction electrocatalyst based on carbon nanotube-graphene complexes. Nat Nanotechnol.
2012;7:394–400.
Alternative Processing for Silver Extraction
from High Manganese-Silver Ores Using Leaching
with SO2 Pretreatment
Ximena Nevárez1 (*), Daniel Manqueros1, Flor Salazar1, Fisher Wang1,

and Persio Rosario1
1
First Majestic Silver Corp, Vancouver, BC, Canada
xnevarez@firstmajestic.com
Abstract. Silver ores with high manganese contents, in order of 2–10%, are difficult to
process using conventional cyanidation methods. The resulting low silver extraction is
primarily due to the intimate association between silver and manganese minerals, which
prevents valuable minerals from contacting solvent agents. The objective of this study
was to investigate alternative processing methods to improve the silver extraction from
the manganese-silver (Mn-Ag) ores. In this research, an extensive experimental program
was carried out using high Mn-Ag ore samples from the La Encantada mine in Mexico to
evaluate the silver extraction from the cyanidation process with and without the addition
of SO2 gas for pretreatment. The results demonstrated a significant increase in silver
extraction with SO2 injection before the cyanidation process. In addition, the effect of SO2
concentration, slurry solids density, oxidants addition and agitation speed on silver dis-
solution was thoroughly investigated.
Keywords: Manganese-silver · Silver extraction · Hydrometallurgy · SO2

pretreatment · Acid leaching
1 Introduction
In 2008, the La Encantada mine in Mexico started the investigation of reprocessing the
old tailings for silver (Ag) extraction. Satisfactory Ag extraction of 45–50% was
obtained from laboratory cyanidation tests on samples collected from the tailings
deposit. As a result, the mine and processing plant were reconfigured to support the
operation of tailings reprocessing for 5 years. However, it was found that the presence
of Manganese (Mn) became increasing notorious during the operation, resulting in sig-
nificantly lower Ag extraction. Thus, a search was initiated to look for alternative meth-
ods for processing the manganese-silver (Mn-Ag) ores.
In 1981, Chase [1] studied the recovery of Ag from ores with Mn, examining differ-
ent extraction methods, including the method proposed in 1957 by Bender and
Rampacek [2] of using the Sulfur dioxide (SO2) gas to dissolve Mn Oxides and Mn
Sulphates. A similar study was conducted in 1981 at the Millers plant in Nevada by
https://doi.org/10.1007/978-3-031-17425-4_93
838 X. Nevárez et al.
Haydon and Rip [3], where the Sulfurous acid (H2SO3) was used to treat Ag minerals
with Mn associations which increased the Ag dissolution from 40% to over 80%.
Canterford, in 1984, attempted to extract Co, Ni, and Cu through leaching with SO2, but
the research was not specifically focused on the dissolution of the Mn minerals [4].
The University of San Luis Potosí [5] conducted a study in 1997 using ore samples
from the La Encantada mine. In that research, SO2 was used, and the Ag dissolution of
80–90% was achieved. Later a characterization study was conducted on similar ore
samples from the La Encantada mine, indicating that Ag was mainly associated with
Pyrolusite and Rhodochrosite. In 2011, Hazen Research (Denver, CO) was requested to
conduct a study to design a continuous acid leaching circuit. It was also found that the
addition of SO2 generated the maximum Ag dissolution. Due to the need for optimiza-
tion of the SO2 gas consumption, the project was put on hold while the operation was
continued with the fresh ore feed only. In 2018, the research of utilizing the SO2 gas for
processing the Ag-Mn ores was resumed, starting with the comparative study for cyani-
dation with and without the SO2 pretreatment.
This paper shows the effect of SO2 gas pretreatment on the Ag dissolution of the fine
particles (−20 μm) due to the high concentration of Ag in the fines fraction.
Sample Preparation Approximately 2 tonnes of samples were collected from the La

Encantada tailings deposit and shipped to the First Majestic Research and Innovation
Center (CIEIFM, Mexico) for the experimental program. The sample was sieved in the
laboratory to eliminate lumps, homogenized using the conical pile method and sampled
by the cake method to obtain a representative sample of about 200 kg, which was then
dried at 70 °C for 24 h.
The particle size distribution analysis was performed on the subsample, where the
P80 of the sample was determined to be 119 μm, with 46% of the mass recorded below
20 μm. Subsequently, the entire 200 kg sample was screened at #635 mesh (20 μm) to
obtain only the fines portion (−20 μm) for the experiments. The fine fraction was fil-
tered, dried, and split into representative subsamples using the rotary quartering method
and stored in 1.5 kg bags, while a 300 g subsample was obtained for chemical analysis.
Note that the −20 μm sample (P80 of 12.7 μm) was used for this research.
Mineralogical Characterization The qualitative mineralogical analysis identified
that the Ag-bearing minerals were Electrum, Cerargyrite, and Goethite. Manganese
oxides (Pyrolusite and Chalcophanite) were also identified in the sample. The analysis
of the final residues from the acid leaching tests with SO2 showed no presence of man-
ganese oxides.
Chemical Characterization A head sample was sent to the CIEIFM Assay Lab for
chemical analysis, where the sample was digested with Aqua regia. The ICP Optima
equipment was used to analyze the Pb, Zn, Cu, Fe, and Mn content in the sample. The
Ag content was determined by the fire assay method. Table 1 summarizes the results of
the chemical analysis performed on the sample.
Alternative Processing for Silver Extraction from High Manganese-Silver Ores Using… 839
Table 1. Chemical characterization.

Chemical assay
Ag g/t Pb % Zn % Cu % Fe % Mn %
123 0.87 0.97 0.01 3.84 3.43
Leaching and Multivariate Experimental Designs A cyanidation test was per-

formed on the obtained fine fraction (−20 μm) using the operating conditions at the La
Encantada mine to establish the baseline performance and confirm the leaching
kinetics.
The bottle-roll cyanidation testing was conducted at 40% solids density. The slurry
pH was adjusted to 10.5–11.0 using lime (CaO), and then sodium cyanide (NaCN) was
added to obtain a concentration of 2000 ppm. The pH adjustment and NaCN concentra-
tion were monitored at 4, 12, 24, 48, and 72 h.
3 Experimental Program
Agitated Leaching Both acid leaching and cyanide leaching were performed using an
agitation machine, which consisted of a motor attached to a hollow and vertical shaft.
An impeller and a diffuser were attached to the lower end of the shaft. The pulley sys-
tem and tachometer were used to control the agitation speed (in rpm). The machine had
an orifice in the stem for the gas injection (i.e. air, O2, and SO2), as well as rotameters
to measure the flow rate (L/min). The agitation was intended to have the maximum dis-
solution of SO2 gas into the slurry to evaluate the effect on the dissolution of Mn and
the exposure of Ag.
Agitated Acid Leaching – SO2 The tests started with acid leaching to expose Ag for
subsequent dissolution process with NaCN. It is believed that the highest yield of SO2
gas would have the greatest impact on the Ag dissolution but could be overly expensive.
Therefore, it is important to obtain the maximum dispersion of the SO2 gas into the
slurry during the test to achieve high efficiency.
In all the tests, the slurry was introduced into the tank of the agitation machine, and
then the SO2 gas was added at a constant flow rate. A special sleeve was used to main-
tain the tank temperature at 25 °C and to avoid issues resulting from inconsistent gas
dispersion.
The tests began with pre-conditioning the slurry to achieve acidification from an
initial pH of 8.0 until a pH around 4.0 is reached. The SO2 gas addition during the acid
leaching is maintained for a determined timeframe. The pH values ranged from 1.8 to
4.0 depending on the amount of SO2 gas added. Note that obtaining a specific pH value
was not the purpose of the research.
Neutralization At the end of the acid leaching, the slurry was left to rest for 1 h to
stabilize the reaction and then proceed to the neutralization stage using CaO before
cyanidation. Once the stabilization period had elapsed when the pH was raised to 11,
NaCN was then added to start the cyanide leaching process.
Fig. 1. Block diagram of conventional cyanidation and experimental agitated leaching with SO2
addition.
Table 2. Experimental parameters and their values investigated in the study.
Parameter Values
Percent of solids, % (Slurry) 30, 35, 40
Addition of SO2 gas, L/min (Acid leaching 0.3, 0.5, 1.0, 1.3, 1.5
stage)
Addition of oxidant reagent, g/t (Cyanidation 20, 30, 40, 50
stage)
Agitation speed, rpm (Acid leaching stage) 1200, 1500, 1800
Acid leaching time, min 5, 10, 15, 20, 25, 30
Leaching with Sodium Cyanide The cyanidation tests were performed in the agita-
tion tank at 600 rpm. Earlier investigations showed that the same results of final Ag
dissolution could be achieved using the agitated leaching process compared to the con-
ventional bottle-roll procedure. The advantage of using the agitation tank for cyanida-
tion was the reduction of the required retention time (from 72 to 24 h) which allowed
the execution of a broader array of tests.
In the development of the test program, the following stages were carried out: acid
leaching with SO2, Neutralization (pH adjustment) and Cyanidation. Figure 1 compares
the block diagram of the conventional cyanidation tests and the proposed agitated cya-
nidation tests. For the latter test program, the agitation speed of 600 rpm, NaCN con-
centration of 2000 ppm, the retention time of 24 h and pH of 10.5–11.0 were kept
constant in all tests.
As shown in Table 2, variables evaluated in the experiments were the slurry solids
density (%), SO2 gas addition rate (L/min), addition of oxidizing reagent in cyanidation
(g/t), agitation speed in acid leaching (rpm), and acid leaching time (min).
4.1 Analysis of the Variables
Variation of the Slurry Solids Density The research began with varying the slurry
solids density: 30%, 35% and 40%. They were considered closer to the industrial set-
ting at ~43% without the need for a thickening stage before cyanidation. The addition
of SO2 was kept constant during the pre-conditioning (3.0 L/min) and acid leaching
(2.0 L/min) stages.
The slurries were prepared with 500 g of dry sample, adding water to achieve the
slurry solids density targets. The slurry was then conditioned by adding SO2 for 2 min
until a pH of ~4.5 ± 0.5 is reached. Figure 2 illustrates that the pH decreased gradually
during the 30 min of leaching and reached an average acidification pH of 2.1 at the end
of the tests.
As shown in Fig. 3, the leaching test using the plant operating parameters yielded
only 14% of Ag dissolution, while the Ag dissolution with acid leaching increased con-
siderably regardless of the slurry solids density.
Variation of SO2 Addition in Acid Leaching Once the slurry solids density of 40%
was set as a fixed variable, the SO2 dosage variation tests were carried out. Six slurries
were prepared at 40% solids, and the injection of SO2 gas during pre-conditioning was
kept constant at 2.0 L/min. The addition of SO2 gas during the acid leaching was varied
at 0.3, 0.5, 1.0, 1.3, and 1.6 L/min.
The slurries were prepared with 500 g of dry sample, adding water to achieve the
slurry solids density target. The slurry was pre-conditioned by adding SO2 for 2 min
until a pH of ~4.5 ± 0.5 is reached. During the 30 min of acid leaching, the pH decreased
with the increasing addition of SO2 gas.
Although the final average pH was 2.42, the results showed that a more aggressive
acidification reaction was required since the Ag dissolution was directly affected by the
5
pH( 30% solids test)
pH(35% solids test)
4
pH(40% solids test)
pH
1
5 10 15 20 25 30
Time, min
Fig. 2. pH decreases during the addition of SO2..

SO2 dosage. The best results were obtained with dosages of 1.3 L/min and 1.6 L/min,
having a similar Ag dissolution of 91%. Figure 4 shows that a further increase of SO2
addition did not improve the Ag dissolution, suggesting that a saturation point was
reached and recoverable Ag from the Mn-Ag association was released.
Variation in the Addition of Oxidizing Reagent in Cyanidation In this research, a
recommended oxidant was tested as a reducing salt. This reagent contains
M-nitrobenzenesulfonic acid and sodium salt. The reagent was investigated with the
purpose of increasing the yield of NaCN in the Ag solution by oxidizing other NaCN
consumers when dissolved (primarily sulfides). The reagent was added at the beginning
of the cyanidation stage, with the following dosage settings: 10, 30, 40 and 50 g/t.
The slurries were prepared with 500 g of dry samples, considering the previous
results, the slurry solids density of 40%, SO2 dosage of 2 L/min in the pre-conditioning
stage and an additional 1.3 L/min of SO2 in the acid leaching stage. Figure 5 shows that
a maximum Ag dissolution of 93% was obtained with the dosages of oxidant at 40 and
50 g/t, which was 2% higher than the tests without the oxidant.
Variation in the Agitation Speed in Acid Leaching The series of tests consisted of
performing a variation in the agitation speed in the acid leaching stage to observe the
effect of a more aggressive gas-pulp contact. The agitation speeds during the acid leach-
ing stage were adjusted to 1200, 1500, and 1800 rpm, while the cyanidation agitation
speed was kept constant at 600 rpm.
The slurries were prepared with 500 g of dry samples, considering the previous
results, slurry solids density of 40%, SO2 dosage of 2 L/min in the pre-conditioning
stage and additional 1.3 L/min of SO2 in the acid leaching, and the addition of 40 g/t of
oxidant reagent in cyanidation.
As shown in Fig. 6, the effect of the agitation speed during the acid leaching stage
on Ag dissolution was inconclusive. These tests require further investigation because
gas leaks were detected during the 1500 and 1800 rpm tests, as a result of inadequate
dispersion in the slurry.
100
90
80
Ag Disolution (%)
70
60
50
40
30
20
10
43 % solids- 30% solids test 35% solids test 40% solids test
Benefit Plant SO2 SO2 SO2
Test, % of Solids
Fig. 3. Effect of the SO2 pretreatment at different slurry solids density on Ag dissolution.
100
95
90
85
Ag Disolution (%)
80
75
70
65
60
55
50
45
40
0.3 lpm 0.5 lpm 1.0 lpm 1.3 lpm 1.6 lpm
Test, lpm SO2
Fig. 4. Effect of the SO2 dosages on Ag dissolution.
100
95
Ag Disolution (%)
90
85
80
75
70
20 g/t 30 g/t 40 g/t 50 g/t
Test, g/t Oxidant reagent
Fig. 5. Effect of oxidizing reagent dosage on Ag dissolution.
100
95
Ag Disolution (%)
90
85
80
75
70
1200 rpm 1500 rpm 1800 rpm
Test, Agitation speed (rpm)
Fig. 6. Effect of agitation speed in acid leaching on Ag dissolution.

Variation in Agitation Time in Acid Leaching A series of tests were carried out to
observe the effect of the agitation time in acid leaching on Ag dissolution. A longer
agitation time provided the slurry with longer exposure to the acidic environment.
The slurries were prepared with 500 g of dry sample, considering the earlier results,
slurry solids density of 40%, SO2 dosage of 2 L/min in the pre-conditioning stage, addi-
tional 1.3 L/min of SO2 and 1200 rpm agitation speed in the acid leaching, and addition
of 40 g/t of oxidant reagent in cyanidation.
Figure 7 showed that the Ag dissolution increased with longer contact time, with the
best result recorded at 30 min agitation time. The experiment was capped at 30 min
because high consumption of SO2 would imply significantly higher operating costs.
4.2 Analysis of Products and Reactants
Acid Solution Analysis A solution sample was taken at the end of the acid leaching
from the best test (conditions are as follows: a pre-conditioning with 2.0 L/min SO2 gas
dosage, 40% slurry solids density, 1.3 L/min of SO2 addition in the acid leaching at
1200 rpm agitation speed). The solution was analyzed with a PE 500 spectrophotometer
for the Ag, Pb, Zn, Fe, Cu and Mn, and the results are summarized in Table 3. It was
found that Mn minerals were mostly dissolved during the acid leaching process, thus
releasing the Ag for subsequent dissolution in the cyanide environment.
Reagent Consumption SO2 gas consumption was 39 L (or 195.55 kg of SO2 per tonne
of ore processed). During the neutralization stage, high consumption of CaO at 47 kg/t
was recorded. The neutralization was a time-consuming process since the pH took a
long time to stabilize (about 10 h), therefore CaO must be continuously added. This was
likely caused by the precipitation of dissolved metals together with the gypsum
formation.
100
95
90
85
Ag Disolution (%)
80
75
70
65
60
55
50
5 10 15 20 25 30
Test, Agitation Time (min)
Fig. 7. Effect of agitation time in acid leaching on Ag dissolution.

Table 3. Chemical characterization.

Chemical elements (mg/L)
Ag Pb Zn Cu Fe Mn
0.226 348 4602 11.23 1360 28,080
The NaCN consumption in the base test using the plant operating parameters was
1.8 kg/t and achieved a poor Ag extraction. Although the NaCN consumption rate
increased to 5.0 kg/t for the tests performed with acid leaching prior to cyanidation, Ag
extraction was significantly improved.
Although the addition of the oxidizing reagent in cyanidation was 40 g/t, the eco-
nomics is likely favorable when considering the additional 2% of Ag dissolution.
5 Conclusions and Recommendations
The sample from the La Encantada tailings deposit was composed of Electrum,
Cerargyrite, Goethite, Pyrolusite, and Chalcophanite as Ag-bearing minerals. The Mn
content in the sample was 3.43%. It is believed that the Mn-Ag association was hinder-
ing the extraction of Ag in the conventional cyanidation process. To expose the Ag for
dissolution with NaCN, an acidification process that uses the SO2 gas in the slurry to
dissolve the Mn minerals was required.
This study has demonstrated a significant improvement in Ag dissolution by apply-
ing the SO2 pretreatment prior to the conventional cyanidation for processing the high
Mn-Ag ore sample from the La Encantada mine. From the experiment, the Ag dissolu-
tion reached 91% using the SO2 gas pretreatment process, compared to only 14%
obtained with the conventional cyanidation process. To achieve 91% of Ag dissolution,
the key component of the experiment was to maintain a uniform dissolution of the SO2
gas within the slurry using the agitated machine to generate an efficient contact with the
ore particles.
The optimal conditions for Ag dissolution involved a pre-conditioning of the slurry
with 2.0 L/min SO2 gas dosage, 40% slurry solids density, and 1.3 L/min of SO2 addi-
tion in the acid leaching at 1200 rpm agitation speed. An additional 2% increase in Ag
dissolution could be achieved by adding the oxidizing reagent in the cyanidation stage.
This paper has demonstrated the benefits of using SO2 to improve the overall Ag
dissolution for the high Mn-Ag ores on a laboratory scale as a proof of concept. Pilot-
scale testing is recommended to confirm these benefits and determine the overall eco-
nomics of the process. Future work also needs to be done on other mechanisms that
allow the maximum diffusion of the gas in the slurry and its contact with mineral par-
ticles and thus be able to take this concept to an industrial scale in a profitable way.
In addition, the neutralization of the slurry required a high consumption of CaO;
therefore, other reagents that could improve the process should also be investigated.
Acknowledgments The authors would like to express their gratitude to First Majestic Silver
Corporation for the permission to publish this work.
References
1. Chase CK. Treatment of manganiferous silver ores for recovery of silver in view of changed pre-
cious metal economics. In: 110th AIME annual meeting, Chicago, Illinois. February 1981.
2. Bender FN, Rampacek C. Percolation leaching of manganese ores with sulfur dioxide, vol. 5323.
United States Department of the Interior, Bureau of Mines; 1957.
3. Haydon BC, Rip FL. Sulfurous acid pre-leaching for reprocessing manganese silver tailings at
Tonopah, Nevada. Miner Metall Process. 1984;1(3):232–6.
4. Canterford JH. Cobalt extraction and concentration from manganese wad by leaching and precipi-
tation. Hydrometallurgy. 1984;12(3):335–54.
5. Valdivieso AL, Bahena MJLR, Espinosa AZ. Recuperacion de plata en minerales con alto con-
tenido de manganeso. In: XXII Convención AIMMGM. October 1997.
Imaging the Remaining Refractory Lining
in Operating Furnaces
Rachel Santini1 (*), Afshin Sadri1, and Wai Lai (Winnie) Ying1
1
rachel.santini@hatch.com
Abstract. Reliable measurement of remaining refractory lining is essential to ensure the

safe and stable operation of furnaces. Carefully monitoring the integrity of a refractory
lining can allow for pre-emptive detection of refractory wear and avoid sudden furnace
failures or other unscheduled shutdowns. This lining information plays a critical role in
allowing furnace operators to confidently push past originally scheduled reline dates.
Recent technological breakthroughs have now made it possible to image the refractory
lining in operating furnaces for the first time. This paper presents the first images ever
created of the refractory lining of an operating furnace.
Keywords: Operating furnace monitoring · Non-destructive testing · Campaign life

extension · Acousto Ultrasonic-Echo (AU-E) · Imaging
1 Introduction
Reliable measurement of remaining refractory lining is essential to ensure the safe and
stable operation of furnaces. Carefully monitoring the integrity of a refractory lining
can allow for pre-emptive detection of refractory wear and avoid sudden furnace fail-
ures or other unscheduled shutdowns. This lining information plays a critical role in
allowing furnace operators to confidently push past originally scheduled reline dates.
There are three major benefits to having a continuous undisruptive extended furnace
campaign. In terms of sustainability and environmental aspects, relining of metallurgi-
cal furnaces generates tons of refractory contaminated waste that need to be disposed of
and monitored for many years. Operating a furnace with the same refractory lining for
an extended period avoids undesirable waste generation. In terms of economy, continu-
ous operation of a furnace without downtimes caused by metal leaks or metal run-outs
is a huge implication. Last but not the least, measuring and monitoring furnace lining
increase the safety aspects of the operation. Recent technological breakthroughs have
now made it possible to image the refractory lining in operating furnaces for the first
time. This paper presents the first images ever created of the refractory lining of an
operating furnace.
https://doi.org/10.1007/978-3-031-17425-4_94
848 R. Santini et al.
2 AU-E Technology
Acousto Ultrasonic-Echo (AU-E) is a non-destructive method of measuring refractory

thickness and condition in operating furnaces. The development of AU-E began in the
late 1990s based on advances in concrete non-destructive testing (NDT) techniques.
Most notably, Impact-Echo was combined with the knowledge of furnace operation to
successfully model the propagation of an acoustic wave within furnace refractory mate-
rial [1–3]. By accounting for extreme temperatures, furnace design factors, and the
multi-layer nature of refractory linings, AU-E was born [4–6]. The AU-E technology
has been patented in all four continents and has received international awards and rec-
ognitions from various global technical organizations for its contribution to metal pro-
cessing and NDT technologies. Namely the MetSoc Innovation award in 2018 and the
Canadian Institute of Non-Destructive Evaluation Innovation Award in 2019. To date;
various types of furnaces, including but not limited to flash furnaces, electric furnaces,
blast furnaces, converters, reactors, ISASMELT, glass furnaces, chlorinators, etc., from
all corners of the globe, have been inspected using this technique.
AU-E utilizes the properties of refractory materials to measure frequency response
to an induced acoustic wave. During measurement, a stress impulse is generated in the
refractory by a mechanical impactor. This impulse then travels through the refractory
and reflects at any change in acoustic impedance along the path of propagation in the
material. The reflected waves are then captured using a transducer placed on the shell
of the furnace, resulting in time-domain signals that are used as the base of AU-E analy-
sis. A diagram of AU-E data acquisition is given in Fig. 1.
The signal produced from this data collection often follows a decaying exponential
trend as energy is dissipated throughout the refractory. This is particularly important
because the high-energy initial return signal can provide a different insight into refrac-
tory conditions compared to the lower energy signals returned later. This knowledge
has allowed for recent improvements in the signal processing of AU-E signals. An
example AU-E signal collected from an operating furnace is given in Fig. 2.
Time-domain signals, by themselves, are complex and difficult to interpret.
Therefore, the information carried by the stress waves is extracted by converting the
collected time-domain signals into the frequency-domain using a standard Fourier
transform. A sample frequency-domain signal is given in Fig. 3 below:
Frequency-domain plots show the amplitude of the signal response at various fre-
quencies. Using these plots, it is easy to obtain the frequencies at which the signal is
being collected. This information can then be related back to the thickness within the
refractory using Eq. (1) below [7]:
Vpn n n 1 n 1 2Ti
Tn (1)
2 f i 1 Vp i i
i
Where Tn is the thickness of the nth layer, Vp is the P-wave speed, α is the thermal cor-
rection factor, β is the shape correction factor, and f is the frequency of the reflection.
Up until recently, this is the extent to which AU-E signals were collected are processed.
However, recent advancements in data collection, data processing, and information
Imaging the Remaining Refractory Lining in Operating Furnaces 849
Fig. 1. Diagram of AU-E data acquisition.
Fig. 2. The typical time-domain signal collected using AU-E.

Fig. 3. Typical frequency-domain signal collected using AU-E.
visualization have now made it possible to use AU-E technology to create an image of
the remaining refractory lining within operating furnaces.
The pre-inspection calibrations provide data to create thickness-frequency modeling
of the area to be tested. For coarse-grained materials, the reflecting signals in the
frequency-domain are complex. Frequency picks are not necessarily related to thick-
nesses. Refractories are complex ceramic blocks that are subjected to continuous severe
conditions and aggressive chemicals that result in material changes. Material changes
result in density, elasticity, strength, and wave speed changes. Everything in the furnace
refractory lining is dynamic and subject to change.
3 Recent Advancements
There have been many developments in AU-E since the first commercial implementa-
tion of the technique in the 1990s. Some of these techniques have come as a result of an
improved data acquisition system, while others have been incorporated through the
processing of the data collection.
The data acquisition system used during the first commercial implementation of the
technique looked far different from the system that is in use today. Advancements in
computer processing, sensors, and the ruggedization of equipment have allowed for
massive improvements in the accuracy of the technique.
The first big change came in the computer processing of AU-E data. Initially, the
step of converting time-domain signals into frequency-domain was done manually and
offered no live-time feedback to the data collection. The introduction of custom-built
software in the late 2000s allowed for the time-domain and frequency-domain signals
to be displayed simultaneously. Furthermore, live-time data visualization allowed for

immediate feedback on the quality of the signals being captured. This in turn led to
higher quality data sets and the quicker processing of collected data.
Another major change came through the introduction of specialized sensors into the
AU-E data acquisition unit. Piezoelectric transducers are common throughout the NDT
industry. However, none of these sensors are sensitive to the range of frequencies most
commonly seen in furnaces. By specially designing and patenting the sensors to be
most sensitive to the desired frequency ranges, and dampening undesired surface and
shear waves, a greater degree of accuracy can be achieved.
Finally, this has all been possible to implement due to advancements in the rugged-
ization of sensitive equipment. The environment around furnaces is extremely hostile to
sensitive electronics. Dust, extreme heat, extreme cold, power fluctuations and ground-
ing issues are challenges that the system needs to overcome in order to function con-
tinuously and without interruption in extreme conditions. However, the increasing
ruggedization of this equipment has allowed for major improvements in the technology
that can be introduced into the data acquisition process. Figure 4 below shows an exam-
ple of the AU-E data collection.
In addition to the improvements to the data acquisition system, the data processing
component of AU-E has also been greatly improved since its initial commercial
implementation. By incorporating thermal data, signal filters, and an expansive array of
gained knowledge, the accuracy of AU-E measurements has been increased dramati-
cally in recent years.
Starting in 2016, thermal modeling has been increasingly incorporated into AU-E
data processing. Thermal data provides an additional reference check to help with the
interpretation of AU-E signals. This is particularly important in furnaces that have
Fig. 4. Photograph of AU-E data being collected.

undergone a chemical attack on the refractory lining. These chemical attacks can create
a brittle layer of refractory that has opposing effects on AU-E and thermal signals.
Chemically altered bricks act as an insulator resulting in thermal models showing the
refractory as thicker than reality, whereas chemically altered bricks reflect AU-E sig-
nals resulting in the data showing the refractory as thinner than reality. By using each
data set as a check on the other, the accuracy of the measurements is increased.
One of the more recent developments in AU-E technology relates to the filtering of
signals once they have already been collected. It is now possible to apply time-domain
filters to the collected signals to reduce noise and increase the clarity of the signals. This
is done by multiplying the time-domain signal by a predefined set of criteria to place
greater emphasis on the time-domain signals being received at various times. This helps
reduce the amount of noise introduced into the signal through surface waves and
obscure geometrical reflections, therefore allowing the greater focus to be placed on the
intended signal. The result is an increase in the clarity of AU-E signals and a more
accurate representation of refractory conditions. An example of an unfiltered signal
versus a filtered signal is given in Fig. 5 below:
Finally, one major change between the AU-E technique that is in use today, com-
pared with that of the late 1990s, and early 2000s, is the expansive array of knowledge
gained over that time. Through the implementation of AU-E over the last 20+ years, an
expansive array of data has been built that now helps to better identify signals and their
meaning on furnace health. This allows for a better interpretation of the signal and a
more definitive conclusion on the health of the furnace lining.
All these innovations have now allowed for the greatest clarity ever achieved in the
AU-E processing of refractory thickness data. This has paved the way for a new form
of AU-E data processing: imaging of the remaining refractory lining.
Fig. 5. (a) unfiltered frequency domain signal (b) filtered frequency domain signal.
4 Refractory Image Processing
The method of collecting and interpreting AU-E data requires several steps before a
readable output is obtained. First, the time-domain data is converted into the frequency-
domain. The resultant frequencies in this spectrum are then input into the thickness
equation (Eq. 1), with the result being a thickness value. This value can then be related
back to its position in the furnace, which when combined with other measurements can
generate a standard refractory thickness profile. An example of one of these profiles is
given in Fig. 6.
However, during this process, some information is inherently lost because not all
frequencies are represented in the final output. Amplitudes that do not meet a certain
threshold must be omitted to maintain clarity of the results. However, by using colors
to represent amplitude, the format can be changed to allow for more information to be
conveyed at the same time. This is the basic idea behind the imaging of refractory wear
profiles within operating furnaces.
To achieve this goal, the frequency-domain data must first be converted into a pseudo-
thickness domain data set. This ensures that each frequency that has reflected a signal is
mapped to the correct location within the furnace. All frequencies received by the sensor
must be converted into a thickness value since all amplitude values, not just in areas
Fig. 6. Refractory wear profile generated using AU-E.

where the largest amplitude peaks occur, are of interest. Due to this large range of fre-
quency values and based on the refractory layout of a furnace, multiple thickness equa-
tions with different final (nth) layers are utilized. Although the refractory is the final layer
in the furnace, it cannot be the nth layer for all thickness equations used to convert all
frequencies to thicknesses. This is because only a certain lower range of frequencies
reaches the final refractory layer, as higher frequencies may not penetrate through to the
final layer. The example frequency-domain signal shown in Fig. 3 is again used to dem-
onstrate the transformation into a thickness-domain signal shown in Fig. 7.
Once this step is complete, it is now possible to generate the images of individual
data points. The thickness plots are colourized, in which, high amplitude reflections
take on a red colour, and low amplitude reflections take on a blue colour. An example
of one of these plots is given in Fig. 8.
Multiple data points are then shown according to their position within the furnace,
and an outline of the furnace is plotted alongside the data to complete the colourized
data set. By doing this, a direct comparison can be made between the amplitude peaks
and the original thickness values. Figure 9 shows the final step of this process in com-
parison with the refractory profile shown earlier in Fig. 6. The high amplitude peaks are
easily seen, and the outline of the refractory thickness is evident. These images provide
the greatest level of detail ever obtained in the imaging of a remaining refractory
lining.
This imaging technique is also used to image a furnace hearth. Figure 10 below
shows the comparison of amplitude peaks from two furnace hearths and their respective
original thickness values. In both plots, the high amplitude peaks indicate full refrac-
tory thickness. However, there are very few reflections from within the brick in the
hearth on the left compared to the hearth on the right, which contains a higher number
Fig. 7. Thickness-domain plot.

Fig. 8. Colourized pseudo-thickness plot of individual data point.
Fig. 9. Comparison of imaging technique to manual data analysis.
of reflections coming from within the brick. A clear signal reflection from within the
refractory lining where the source of the reflection is unknown is referred to as an
anomaly in the context of the AU-E technique. Anomalies can indicate potential cracks,
gaps, delamination, metal infiltration/impregnation, oxidation, or a combination of
these features. In this case, a distinct line of metal infiltration/impregnation can be
observed from the reflections from within the brick.
The benefits to furnace operators are quite readily apparent: quick feedback on the
condition of their remaining refractory lining, the prediction of future wear patterns in
the furnace, and an overall assessment of furnace health. Furthermore, this will improve
the AU-E technique by allowing for a more in-depth and holistic approach to data anal-
ysis. As well, the combination of this imaging technique and periodic AU-E inspections
will allow for the creation of trends for predictive maintenance of refractory lining in
furnaces.
Fig. 10. Comparison of imaging technique to manual data analysis for a furnace hearth (a) with few
anomalies and (b) with many anomalies.
5 Conclusions
Advancements made since the introduction of the AU-E technique have now made it
possible to generate the first images of the remaining refractory lining in operating
furnaces. These images are unique because they display only the raw data collected
using the technique and require no human input to interpret the data.
Moving forward, an increasing amount of AU-E data processing will be conducted
using the imaging techniques described above. This will allow for a more holistic view
of furnace condition changes over time, and lead to a better understanding of the pro-
cesses of refractory wear in the sidewall and hearth. With periodic AU-E inspections,
trends for predictive maintenance of remaining refractory lining in the furnaces can be
created. This is advantageous for furnace operators as they can confidently operate
without undesirable stoppages, as well as, potentially prolong the campaign life of the
furnaces. This greater understanding of refractory wear processes and trends can ulti-
mately lead to more sustainable furnace operation and possibly minimize unnecessary
relines.
The next steps in imaging the remaining refractory lining in operating furnaces are
implementing three-dimensional lining imaging and the use of artificial intelligence
and/or pattern recognition in signal interpretation.
References
1. Sansalone MJ, Streett WB. Impact-echo: nondestructive evaluation of concrete and masonry.
Ithaca: Bullbrier Press; 1997. p. 339.
2. Sansalone MJ, Carino NJ. Impact-echo: a method for flaw detection in concrete using tran-
sient stress waves, NBSIR 86-3452, NTIS PNB#87-104444/AS, National Bureau of Standards,
Springfield, VA, USA. 1986. p. 222.
3. Carino NJ, Sansalone MJ. Impact-echo: a new method for inspecting construction materials. In:
Proceedings of nondestructive testing and evaluation of materials for construction, University of
Illinois, Urbana and Champaign, IL, USA. August 1988.
4. Sadri A. Non-destructive determination of refractory and build-up thickness in operating furnaces
using an acousto ultrasonic reflection technique. In: Proceedings of the materials degradation:
innovation, inspection, control and rehabilitation symposium, COM2005, Calgary, AB, Canada.
August 2005.
5. Sadri A, Marinelli P, Doro E, Gebski P, Rampersad A. Comparing the accuracy of acousto
ultrasonic-echo (NDT), finite element analysis (FEA), and drilling when obtaining a blast refrac-
tory lining wear profile. In: AISTech 2009, Proceedings of iron and steel technology conference,
St. Louis, MO, USA. May 2009.
6. Sadri A, Lachemi M, Walters G. Determination of refractory lining thickness and quality in oper-
ating industrial furnaces using a stress wave reflection technique. In: Proceedings of first CSCE
specialty conference on infrastructure technologies, management and policy, Toronto, ON, Canada.
June 2005.
7. Sadri A, Gordon I, Rampersad A. Acousto ultrasonic-echo (AU-E): a non-destructive testing tech-
nique for blast furnace hearth refractory condition monitoring. In: AdMet 2007, Proceedings of
international conference advances in metallurgical processes and materials, vol. 2, Dnipropetrovsk,
Ukraine. May 2007. p. 77–85.
Dynamic Renewable Energy-Driven Framework
Development for Mineral Processing Circuits
Ryan Wilson1, Andy Reynolds2, and Alessandro Navarra1 (*)
1
alessandro.navarra@mcgill.ca
2
Inspire Resources Inc., Nanaimo, BC, Canada
Abstract. Evolving global economics and socio-environmental factors are driving

research and development trends in the mining industry towards the use of 100% renew-
able energy sources to support operations. In addition to the reduction of carbon foot-
prints, this will be particularly important to improve the feasibility of remote projects that
often face significant scheduling and cost barriers related to transmission line installation
and/or fuel delivery for generator-based systems. The state of the art is to use rolling ele-
ment mills, dry grinding and process storage systems to enable intermittent operation of
the mill. However, conventional circuits that rely on carbon-based fuels typically require
continuous operation in order to remain economically competitive. The notion of an effi-
cient plant that can start up and shut down at a high frequency (e.g. daily) based on
dynamic fluctuations in the availability of renewables has not been properly tested. To
highlight the performance attributes and constraints that would need to be addressed by
equipment designers, this work adapts a systems approach to support solar energy-driven
mineral processing facilities. The approach is to incorporate mass balancing via linear
circuit analysis into a discrete event simulation framework to evaluate the system-wide
response to changes in energy supply caused by weather patterns and related environmen-
tal uncertainty. Sample simulations are presented in the context of a copper flotation plant
loosely based on the Australian context.
Keywords: Remote mining · Renewable energy · Extractive metallurgy · Mineral

flotation · Linear circuit analysis · System dynamics · Discrete event simulation ·
Mass balance · Environmental uncertainty · System flexibility
1 Introduction
With many of the near-surface, high-grade ore deposits already identified across most
of the easily accessible regions of the world, mining companies are now faced with
increasingly challenging prospects. This is compounded by evolving global economy
and commodity markets, as well as complex socio-environmental and geopolitical fac-
tors. Moreover, accelerated world population growth, coupled with a collective push
toward electrification and automation in the age of Industry 4.0, will inevitably increase
the overall demand for raw materials. With declining trends in the rates, sizes and
grades of new mineral discoveries, this imbalance is expected to drive significant
increases in energy consumption across the mining sector to meet future supply targets
https://doi.org/10.1007/978-3-031-17425-4_95
860 R. Wilson et al.
[1]. This is especially significant given the current state of climate change and related
pressure from community stakeholders to reduce greenhouse gas emissions and transi-
tion away from fossil fuels to a clean energy-based economy.
As a result, recent trends indicate that renewable energy sources are rapidly gaining
traction in the mining industry. For instance, the renewable energy supply serving mine
sites globally grew from just 600 MW in 2015 to 4.9 GW in 2020 [1, 2]. As new tech-
nologies continue to emerge, research and development has also turned to the use of
100% renewables to support operations via a combination of wind, solar, geothermal,
battery and process storage, among other methods. Beyond the clear benefits of reduc-
ing carbon footprints and obtaining the social license to operate (SLO), such an
approach will likely be critical to improve the feasibility of remote projects with mar-
ginal economics. These operations tend to be smaller in nature, have less available
working capital, and commonly face significant scheduling and cost barriers related to
power grid installation and/or fuel delivery constraints for generator-based systems.
In general, conventional mineral processing circuits require around-the-clock con-
tinuous operation to improve system efficiencies and offset the high costs incurred by
carbon-based fuel consumption in order to remain economically viable. However,
recent advances in comminution practices, such as rolling element mills and dry grind-
ing, and process storage systems have enabled the development of intermittent opera-
tion of the mill (e.g. OZ Minerals’ West Musgrave project [3]). Such approaches tend
to rely on capital-intensive energy storage, though projections show the costs for large
utility-scale lithium-ion battery systems are rapidly declining [4]. While technological
advances will continue to play a critical role in the development of cost-effective long-
duration energy storage systems, computational tools will be equally important to eval-
uate the tradeoffs between varied approaches for the design and implementation of new
or modified strategies.
What has not been properly tested to date is the concept of an efficient processing
plant that can start up and shut down at a high (e.g. diurnal) frequency contingent on
dynamic fluctuations in the direct availability of renewables. This work utilizes a sys-
tems approach to analyze the transient response of a remote copper flotation plant sup-
plied purely by available solar energy in order to better understand the tradeoff with
energy storage systems. The ultimate intent is to highlight the performance attributes
and constraints that would need to be addressed by equipment designers to support
plant flexibility under such a strategy. The general approach allows for adaptable min-
ing and mineral processing profile designs by coordinating a range of standardized
models and submodels (i.e. modules) into systems. In this way, relationships between
important variables and processes can be conceptualized and quantified through model-
ling and simulation to capture key decision points within complex dynamic systems [5].
The methodology is especially useful in the design and testing of alternative strategies
and related operational policies, associated tradeoffs and resultant short- to long-term
impacts [5]. A systems approach that supports hierarchical detailing is thus suitable for
smaller-scale projects that are commonly subject to unique requirements and related
project barriers.
Mineral flotation comprises a set of physicochemical processes that separate target
ore minerals from the remaining gangue minerals and impurities that make up the
majority of the host-rock assemblage [6]. This separation occurs on milled aqueous
Dynamic Renewable Energy-Driven Framework Development for Mineral Processing… 861
mineral suspensions, whereby valuable minerals are floated from the tailings (waste)
based on underlying hydrophilic/hydrophobic properties under forceful air bubbling
[6]. The overall process uses a series of interconnected flotation cells (grouped into
banks) to ultimately produce metal concentrate and tailing flows; system performance
is, therefore, dependent on circuit configuration and the nature of the ore feeds [6].
Flotation circuits typically contain a large number of variables that affect overall
system performance to varying degrees [7], and which may be subject to stochastic (e.g.
geological uncertainty) and/or epistemic (lack of knowledge) variations [8]. However,
linear circuit analysis is a modelling tool that evaluates potential separation circuit
designs through simplified algebraic formulations [8–10]. The approach is particularly
effective in the early design stages because it focuses on unit arrangements and stream
interconnections, and only requires limited equipment and feed condition specifications
[8–10]. The methodology is based on separation probabilities, expressed via transfer
functions given by the mass flow ratios of individual variables (e.g. ore vs. gangue par-
ticle classes) within each process unit. This mass balance can be extended to the entire
circuit, from which global performance indicators such as recovery, grade and separa-
tion efficiency can then be calculated [8–10].
Moreover, discrete event simulation (DES) is an important class of computational
methods gaining traction in the integrated modelling of mining and mineral processing
systems. DES is capable of simulating the interactions between critical parameters and
processes within complex systems in response to random variations. The ability of DES
to simulate extended periods of operation allows for the identification of potential defi-
ciencies, bottlenecks, or other risks, which can then be mitigated by adjusting operating
policies, implementing control strategies, or modifying the system design. Recent work
has shown the advantages of a systems approach and the implementation of alternate
modes of operation using mass balancing and mathematical programming [11]. Similar
techniques using DES have also been applied to the concentrator and smelter dynamics
[12–15], heap leach processes [16], tailings retreatment [17], oil sands processing [18],
and reagent control in gold extraction [19]. Specific to mineral flotation, a recent study
used a hybrid approach by combining phenomenological models with DES to analyze
the dynamics of rougher cells [20]; although, more work is required to incorporate the
approach into design methodologies, particularly regarding hierarchical detailing in the
context of small-scale projects.
The current study incorporates mass balancing via linear circuit analysis into a dis-
crete event simulation framework to evaluate the system-wide response to transient
changes in energy supply caused by weather patterns. The coupling of linear circuit
analysis at the unit operation level with DES at the system level can be a useful approach
to assess risk at both short and long timescales (similar to [20]). System performance is
then tracked using a variety of KPIs to help guide the design process. The remainder of
the paper is organized into sections comprising materials and methods, followed by
preliminary results for a copper flotation plant loosely based on the Australian context,
and finally a brief discussion and goals for future work.
The models and simulation tools for this work are being developed using the Matlab®
and Simulink® platforms by Mathworks®, which allow for modifications and custom-
izations to be made directly in the source code.
2.1 Data Acquisition
The initial focus of the study is centered on the development of a generic copper flota-
tion circuit model. For simulation purposes, a proprietary set of ore characterization
data was acquired as part of a partnership with Australian miner OZ Minerals. The set
is comprised of data points for 50,000 randomly selected parcels from a geological
block model of an operating mine and includes tonnages, metal concentrations (Cu, Au,
Ag), and mineral abundances for a variety of dominant sulfide phases, including pyrite,
chalcopyrite, chalcocite and bornite. Descriptive statistics for the dataset are summa-
rized in Table 1.
2.2 Linear Circuit Analysis
Linear circuit analysis is a tool to evaluate the separation performance of multi-unit

circuits based strictly on unit arrangements and related interconnections [9]. The
method relies on separation probabilities (a.k.a. transfer functions) that give the
concentrate-to-feed mass flow ratios for each target component within a given separa-
tion unit process [9]. As such, this approach assumes that these transfer functions are
known (i.e. determined empirically), in addition to process linearity [21].
The mass balance for each separation unit with a transfer function, Pi, can be deter-
mined as follows [9, 10, 21]:
Ci = Fi Pi (1)
where Ci is the mass of component i in the concentrate stream, Fi is the mass of compo-
nent i in the feed stream, and
Table 1. Summary of descriptive statistics for ore characterization dataset (n = 50,000).

Variable Units Min Max Mean St. dev.
Mass tonnes 43.7 191 55.0 17.7
Copper wt.% 0.001 9.73 1.77 1.65
Silver ppm 0.166 42.7 3.76 5.13
Gold ppm 0.005 1.96 0.304 0.249
Pyrite wt.% 0 1.76 0.01 0.07
Chalcopyrite wt.% 0 5.14 0.10 0.43
Chalcocite wt.% 0 12.3 1.93 2.06
Bornite wt.% 0 6.66 0.33 0.69
Ti Fi 1 Pi (2)
where Ti is the mass of component i in the tailings stream. This approach can be
extended to multi-unit configurations by deriving an algebraic function that describes
the concentrate-to-feed ratio (C/F) for the entire circuit (i.e. global recovery). From
this, other performance indicators, such as grade and separation efficiency, can then be
calculated to compare circuits with different configurations [9, 10].
To illustrate the overall technique, Fig. 1 shows the currently modeled flowsheet and
algebraic formulations for a typical three-stage circuit consisting of singular roughing,
cleaning and scavenging (RCS) banks with recycling. This simplified design approach
is important to the present work as it maintains focus on the interconnections between
unit processes rather than detailed physicochemical phenomena. The key is to ensure
hierarchical flexibility within the framework such that more detailed models and sub-
models can be implemented as necessary depending on the scope of a given problem,
or as more data becomes available through continued design and testing work.
Initial simulation work is focused primarily on material flows of ore feed, concen-
trate and tailing streams with respect to individual target components (i.e. Cu, Ag and
Au), in addition to energy consumption. Though alternative configurations will also be
tested in an effort to optimize the generic copper flotation flowsheet, the real crux of this
work is to better understand the time constants that arise from frequent startup and
shutdown. This steady-state model is being incorporated into the greater DES framework
to observe the transient response of the flotation circuit during intermittent operation as
a function of available solar energy. Emphasizing the importance of connections
between unit processes using a simplified flowsheet will ultimately allow for the analy-
sis of the tradeoff between plant flexibility and the need for large capital-intensive
energy storage systems.
Fig. 1. Flowsheet and algebraic formulations derived via linear circuit analysis for a typical rougher-
cleaner-scavenger (RCS) circuit with recycled scavenger floats and cleaner sinks fed back into the
roughing stage. (After Noble et al. [10]).
2.3 Simulation Framework
The context for the present study is a program of work to create the capability to iden-
tify the value of flexibility in mining and mineral processing. At the core of the program
is a simulation platform being developed by Inspire Resources Inc. with Mathworks®
tools. Monte Carlo simulation is an appropriate technique for assessing value across a
wide range of scenarios [22] but requires relatively simple models. To determine the
value of flexibility, these models should include management decision logic. Discrete
event simulation (DES) is an appropriate tool for this purpose because it can easily
implement “if-then” logic. With careful design, DES can balance computational effi-
ciency with adequately representative component models for use in Monte Carlo
simulation.
A process flow diagram for the whole-system model being developed by Inspire
Resources is shown in Fig. 2. In the application relevant to this study, i.e. flexibility for
tolerance of solar energy variability, the model is run with the objective of power bal-
ance, with a load dispatch algorithm determining which unit operations should run
according to available power and stored electrical energy. The unit operation models are
defined with high-level empirical data from equipment supply and engineering compa-
nies participating in the development program. Examples of implementation include
recovery-by-particle-size and recovery-by-grade curves derived from metallurgical test
work specific to the modelled ore body.
3 Preliminary Results
In the course of development of this whole-system model, it became apparent that the
behaviour of flotation plants under transient operating conditions such as short duty
cycles has received little attention. A histogram of daily flotation plant run-times for a
preliminary simulation run is shown in Fig. 3. The model parameters for this distribu-
tion included a 1-year hourly record of solar energy resources in South Australia, and a
Fig. 2. Process flow diagram for the whole-system simulation model being developed by Inspire
Resources Inc. A load dispatch algorithm determines which unit operations should run as a function of
available power and stored electrical energy.
Fig. 3. Histogram of daily flotation plant run-times from a preliminary simulation spanning a 1-year
period (less 35 days for initial start-up) using the solar dispatch algorithm.
geologic block model of an underground copper mine contributed to the program by

OZ Minerals. The mine design and sequencing algorithm are based on all-electric
mechanical cutting methods and are proprietary to Inspire Resources Inc. The load
dispatch algorithm used for this example prioritized mining over mineral processing in
order to create demanding duty cycle conditions for the flotation plant with electric
power as the system bottleneck. It should be noted that a minimum run-time of 4 hours
was imposed on the plant via the dispatch algorithm as a tradeoff against energy
storage.
It is clear from the above histogram that a significant number of simulated duty
cycles (nearly 50%) under the present scenario are very short in duration, with many
instances occurring between just 4 and 8 hours. This is rather telling and is a driving
factor behind the current study to explore the notion of a flexible plant design capable
of responding to frequent changes in renewable energy supply. In other words, the opti-
mal circuit for such a case may necessitate a simpler design focused on rapid startup,
material transfer and shutdown capabilities, rather than detailed phenomenological
models centered on optimal steady-state operating conditions.
4 Discussion and Future Work
As previously discussed, accelerated global population growth and current mining

trends indicate a growing energy demand within the minerals sector. Coupled with
increasing socio-environmental awareness and other geopolitical factors, there is sig-
nificant pressure from stakeholders and the community at large to drastically reduce the
carbon footprints of all industrial operations. As a result, research and development
trends are gradually being driven towards the sustainable development of mining proj-
ects using 100% renewable energy sources to support these operations.
This work is especially interested in the development of mining and mineral pro-
cessing approaches that consider the possibility of flexible operations that can handle
intermittent operating periods as a function of available renewable energy supply. Such
an approach could be particularly important for smaller-scale remote operations facing
significant scheduling and cost barriers related to power grid availability and fuel deliv-
ery logistics. While recent work has shown the ability for intermittent mill operation
(e.g. OZ Minerals’ West Musgrave project), it seems very little work (if any at all) has
gone into a similar approach for flotation plants.
The present study incorporates mass balancing via linear circuit analysis modelling
into a DES-based simulation model to evaluate the transient response of a flotation
plant to fluctuations in solar energy supply. The main idea is to model the potential
tradeoff between plant flexibility and large capital-intensive battery and process storage
systems. Preliminary simulation results from a scenario loosely based on the Australian
context indicate a relatively high proportion of short (<8 hours) flotation plant duty
cycles based on a custom load dispatch algorithm. Realistically, the nature of these
shorter cycles may not allow true steady-state conditions to ever be reached. As such, it
is quite possible that simplified circuit designs focused on the interconnections between
unit processes, rather than the optimal conditions within each unit cell, may be more
effective towards the notion of efficient plant flexibility.
The concept of intermittent operations raises an important question with respect to
idle assets and manpower, which can have significant impacts on operating costs. While
not addressed directly here, this issue of idleness is not limited to the described remote
project scope. In the push towards electrification, i.e. the clean energy transition, all
industries will face challenges related to power grid management and associated elec-
tricity rate design models. For instance, increasing loads on the power grid are already
driving a shift from traditional volumetric (flat rate) to dynamic (time-variant) pricing
models in order to reduce costs and cut down on pollution and environmental disrup-
tions caused by power plants and transmission lines. In any case, the idea of reduced
and/or intermittent operations will need to be evaluated across the industrial spectrum,
especially in the mining sector. As new technologies and computational capabilities
continue to emerge in the age of Industry 4.0, it is likely that automation will play a key
role in reducing these effects moving ahead.
Future work will focus on analyzing the time constants that result during frequent
startups and shutdowns using different circuit configurations. The intent is to identify
the most important parameters affecting the rate of the recovery curve in order to open
the discussion with equipment designers and manufacturers. The novelty of this
research lies in the fact that market demand is more or less exclusively centered on
continuous steady-state operating conditions. The development of new benchmarking
tools that measure the speed of response will be important for the possibility of efficient
and flexible flotation plant design.
Acknowledgments Partial funding for this work was provided by NSERC, grant no. 2020-04605,
supported by the Government of Canada. Support was also provided to Ryan Wilson through the
Mitacs Accelerate Internship program, with matching funds from industry partner Inspire Resources
Inc.
References
1. Igogo T, Lowder T, Engel-Cox J, Newman A, Awuah-Offei K. Integrating clean energy in min-

ing operations: opportunities, challenges, and enabling approaches. Technical report, National
Renewable Energy Laboratory. NREL/TP-6A50-76156. 2020. https://www.nrel.gov/docs/fy20o-
sti/76156.pdf. Accessed 28 Feb 2022.
2. Kirk T, Cannon C. Sunshine for mines: a brighter vision for sustainable resources. Rocky Mountain
Institute; 2020. https://rmi.org/sunshine-for-mines-a-brighter-vision-for-sustainable-resources/.
Accessed 28 Feb 2022.
3. Gleeson D. OZ Minerals, Loesche team up following West Musgrave vertical roller mill test work.
International Mining Team Publishing Ltd; 2020. https://im-mining.com/2020/06/25/oz-minerals-
loesche-team-following-west-musgrave-vertical-roller-mill-test-work/. Accessed 28 Feb 2022.
4. Hoff S, Mey A. Utility-scale battery storage costs decreased nearly 70% between 2015 and
2018. U.S. Energy Information Administration; 2020. https://www.eia.gov/todayinenergy/detail.
php?id=45596. Accessed 28 Feb 2022.
5. Alpagut MA, Çelebi N. System dynamics applications in the mining industry. In: Ozbayoglu G,
editor. Proceedings of the 18th International Mining Congress and Exhibition of Turkey (IMCET).
Antalya: Chamber of Mining Engineers of Turkey; 2003. p. 6.
6. Cisternas LA, Acosta-Flores R, Lucay F, Gálvez ED. Mineral concentration plants design using
rigorous models. In: Kravanja Z, Bogataj M, editors. Proceedings of the 26th European Symposium
on Computer Aided Process Engineering (ESCAPE 26). Portoroz: Elsevier B.V.; 2016. p. 1461–6.
https://doi.org/10.1016/B978-0-444-63428-3.50248-4.
7. Jovanovic I, Miljanovic I. Modelling of flotation processes by classical mathematical methods – a
review. Arch Min Sci. 2015;60(4):905–19.
8. Cisternas LA, Lucay FA, Botero YL. Trends in modeling, design, and optimization of multiphase
systems in minerals processing. Fortschr Mineral. 2020;10(22):1–28.
9. Amini SH, Noble A. Application of linear circuit analysis in the evaluation of mineral processing
circuit design under uncertainty. Miner Eng. 2017;102:18–29.
10. Noble A, Luttrell GH, Amini SH. Linear circuit analysis: a tool for addressing challenges and iden-
tifying opportunities in process circuit design. Min Metall Explor. 2018;36:159–71.
11. Navarra A, Rafiei A, Waters K. A systems approach to mineral processing based on mathematical
programming. Can Metall Q. 2017;56(1):35–44.
12. Navarra A, Marambio H, Oyarzun F, Parra R, Mucciardi F. System dynamics and discrete event
simulation of copper smelters. Miner Metall Process. 2017;34(2):96–106.
13. Navarra A, Alvarez M, Rojas K, Menzies A, Pax R, Waters K. Concentrator operational modes in
response to geological variation. Miner Eng. 2019;134(1):356–64.
14. Navarra A, Wilson R, Parra R, Toro N, Ross A, Nave J-C, Mackey PJ. Quantitative methods to sup-
port data acquisition modernization within copper smelters. Processes. 2020;8(11):1478.
15. Wilson R, Perez K, Toro N, Parra R, Mackey PJ, Navarra A. Mine-to-smelter integration frame-
work for regional development of porphyry copper deposits within the Chilean context. Can Metall
Q. 2021;61(1):48–62.
16. Saldaña M, Toro N, Castillo J, Hernandez P, Navarra A. Optimization of the heap leaching process
through changes in modes of operation and discrete event simulation. Minerals. 2019;9(7):421.
17. Wilson R, Toro N, Naranjo O, Emery X, Navarra A. Integration of geostatistical modeling into
discrete event simulation for development of tailings dam retreatment applications. Miner Eng.
2021;164:106814.
18. Wilson R, Mercier PHJ, Patarachao B, Navarra A. Partial least squares regression of oil sands
processing variables within discrete event simulation digital twin. Fortschr Mineral. 2021;11:689.
19. Órdenes J, Wilson R, Peña-Graf F, Navarra A. Incorporation of geometallurgical input into gold
mining system simulation to control cyanide consumption. Fortschr Mineral. 2021;11:1023.
20. Saldaña M, Neira P, Flores V, Moraga C, Robles P, Salazar I. Analysis of the dynamics of rougher
cells on the basis of phenomenological models and discrete event simulation framework. Metals.
2021;11:1454.
21. Gálvez ED. A shortcut procedure for the design of mineral separation circuits. Miner Eng.
1998;11(2):113–23.
22. Kleijnen JPC. Chapter 18 – Design and analysis of Monte Carlo experiments. In: Gentle JE,
Hardle WK, Mori Y, editors. Handbook of computational statistics. New York: Springer Science +
Business Media; 2012. p. 529–47.
The Role of Cold Bonding in Sustainable Steelmaking
R. Elliott1 (*), B. MacKinnon1, and I. Cameron1

1
richard.elliott@hatch.com
Abstract. All iron and steelmaking processes generate solid ferrous waste. These dusts,
sludges, and scales represent a significant environmental liability if not reclaimed and
recycled. At present, the most common strategy to recover these iron units in a blast
furnace-basic oxygen furnace steel plant is to produce sinter, a technology developed over
100 years ago for this purpose. As pressure to decarbonize intensifies, integrated steel-
makers will increasingly seek alternatives to replace carbon-intensive sinter plants. Cold
bonding is one possible alternative and has been demonstrated by blast furnace operators
in Scandinavia and elsewhere, although complete replacement of the waste recycling
infrastructure for an integrated steel plant is a complex task. A parallel challenge exists for
direct reduced ironmaking (DRI) plants, which typically do not possess an equivalent
integrated waste recycling route. As DRI plants proliferate as a decarbonization strategy,
so will demand for low-emissions waste solids recycling. This extended abstract dis-
cusses the prospects for the widespread adoption of cold bonding technology within the
iron and steel industry.
Keywords: Cold bonding · Briquetting · Decarbonization · GHG emissions · Iron

ore · Sintering · Pelletizing · Waste solids
1 Introduction
Steelmaking is a waste-intensive process. A modern integrated steel plant may lose up

to 10% of the iron units in its feed to various waste streams, in addition to the significant
volumes of slag and other non-ferrous waste generated in the process (Table 1).
Recovering these iron units in the process is one of the primary functions of a sinter
plant. However, sintering requires the use of solid carbon and emits on the order of
300 kg CO2e/t sinter according to the IPCC [1].
As blast furnace operators face pressure to decarbonize their operations, many are
considering a shift towards pellets as the primary burden material, as is already the case
in North America. Pelletizing is less carbon-intensive than sintering, capable of achiev-
ing an emissions intensity of <100 kg CO2e/t pellet. Increasing the fraction of pellets in
a blast furnace burden has also been shown to increase blast furnace efficiency, decrease
the coke rate, and contribute to a reduction in direct emissions [3]. However, pellet
plants are typically not colocated with the plants(s) they supply, and are not generally
effective at recycling ferrous waste materials into materials suitable for charging to a
https://doi.org/10.1007/978-3-031-17425-4_96
870 R. Elliott et al.
Table 1. Typical solid ferrous waste generation in the BF-BOF and DRI-EAF routes [2].
Ferrous solid waste Units BF-BOF DRI-EAF
Iron ore fines kg/tls – 30–45
Dusts & sludges kg/tls 30–80 20–60
Metallic fines kg/tls – 10–50
Mill scale kg/tls 30–70 30–70
Total kg/tls 60–150 80–215
blast furnace. Blast furnace operators pursuing this transition to pellets will require an
alternate solid waste disposal strategy.
Direct reduction (DR) plant operators face a similar challenge. These DR plants also
generate a significant volume of waste (Table 1) but do not have an integrated recycling
strategy. Typically, these plants sell as much of their waste as possible (frequently to
local sinter plants) and landfill everything that cannot be sold. As DR plants proliferate
as part of the decarbonization of the iron and steel industry and sinter plants are idled,
alternative strategies for recycling ferrous waste will be required.
2 Cold Bonding as Waste Management for Integrated Steel

Plants
Sintering is an agglomeration process used to convert a mixture of iron ore fines, recy-
cled ironmaking by-products, fluxes, slag-forming agents, and solid fuel (coke breeze
or anthracite) into sinter. This sinter is produced to have suitable characteristics (chemi-
cal, mechanical, physical, and thermal) for use as a blast furnace burden material.
However, the sinter is not strong enough to withstand long-distance transportation, and
sinter plants are colocated with the blast furnace. This colocation allows ferrous wastes
to be readily blended with sinter fines as feed for the sintering process, facilitating the
recycling of these iron units.
It is also possible for a blast furnace to operate on a 100% pellet burden, as is done
in several plants in North America and Europe. In these cases, no sinter plant is oper-
ated onsite and a different waste management strategy is required. For most North
American plants operating on pellets, the solution is to sell as much of the waste as
possible to one of the few sinter plants operated in the region and landfill the rest. This
strategy is driven by cost-minimization, not sustainability.
Some blast furnace operators have taken a different approach, cold bonding their
ferrous wastes for recycling to the BF and/or BOF. SSAB Luleå in Sweden operates a
cold briquetting facility for ferrous wastes. Dust, sludges, and mill scale are blended
with cement and formed into briquettes, then cured for up to 30 days to develop the
strength needed to use as a blast furnace burden material. These briquettes can be
charged to the BF at rates up to 80–100 kg/thm without observed impact on BF perfor-
mance. In Sweden, the tonnage of BF sludge produced amounted to 22,700 metric ton
in 2012 [4, 5]. This system does not recycle all ferrous waste, and a significant fraction,
including the fine fraction of the sludges, is sent to landfills. In 2014, SSAB Luleå
The Role of Cold Bonding in Sustainable Steelmaking 871
started using a new sludge drying method to recycle 10,000–20,000 tonnes of BOF
sludge into briquettes for use as a raw material instead of being landfilled [6]. Overall,
roughly half of all waste solids are recycled.
Cold-bonded, cement-based briquettes can also be used as a charge material for the
BOF. USS Kosice and others currently briquette much of their ferrous oxide wastes in
this manner and charge them to the BOF as a partial scrap replacement: the reduction
of the iron oxide acts as a coolant and also supports the decarburizing of the hot metal.
Cold bonding as a waste management strategy for integrated steelmakers is a techni-
cally mature technology that has been in commercial use for decades. It has been
adopted only in a limited number of cases as most integrated sites have a sinter plant
that serves the dual function of feed preparation and waste management. Those cases
where cold bonding has been implemented have been driven primarily by environmen-
tal regulations. As operators, particularly in Europe, seek to decrease their carbon emis-
sions while continuing to operate their major assets, it is likely that some will elect to
replace their sinter plants with imported pellets and cold-bonded waste briquetting.
With current cold-bonding technology, approximately half of the waste solids gener-
ated in an integrated steel plant could be recycled without a sinter plant. This represents
a significant decarbonization opportunity but leaves a significant volume of waste to be
addressed by landfilling or recycling at another sinter plant. Thus, cold-bonding is
likely to be only one component of a multi-faceted decarbonization strategy for sinter
plants.
3 Cold Bonding as Waste Management for Direct Reduction

Plants
Direct reduction plants do not possess any form of integrated ferrous waste recycle
system. As a result, their waste must be sold (frequently to local sinter plants) or accu-
mulated in stockpiles or landfills. These waste volumes are significant and represent a
significant ferrous yield loss for the DR-EAF steelmaking route.
Strategies for recycling these ferrous wastes into the DR process have been an area
of interest for the past several decades. Several operating DR plants, including voestal-
pine Texas, Emirates Steel, and Qatar Steel, have invested in briquetting projects with
only limited success. When the DR plant is colocated with an EAF melt shop, some
ferrous wastes can be charged to the steelmaking process. For example, Qatar Steel
reported in 2019 that 100,000 t of DR fines and dusts were briquetted and recycled to
their melt shop [7]. When the DR plant is an independent facility, this is not possible.
In both cases, the oxide fines require an alternate disposal method. As with the inte-
grated steel plants, this alternate strategy is typically sold to a sinter plant or
landfilling.
As additional DR plants are constructed to meet global climate policy targets, this
waste solids management challenge will grow. Truly sustainable steelmaking requires
that process yield is maximized, minimizing primary mining requirements as well as
the environmental impacts of waste disposal. DR technology is well suited to support-
ing the decarbonization challenges facing the iron and steel industry, but without the
development of a sustainable waste management strategy may create a new liability for
the industry.
4 Cold Bonding as a Replacement for Sintering and Pelletizing
The global seaborne iron ore market is comprised of sinter feed, lump ore, pellets and
pellet feed. Sinter feed sales are the largest volume of the global seaborne trade,
although pellet feed and fired pellet sales have experienced greater growth over the past
decade. Despite this growth in pellet feed and pellet sales, sinter feed represents about
70% of the global seaborne iron ore trade. As a result, the global supply of pellets is
insufficient to support a significant transition from sinter-based ironmaking to pellet-
based ironmaking, and blast furnace ironmaking will continue to rely heavily on sinter
fines.
Cold bonding has potential application as a partial replacement for indurated pri-
mary ferrous materials, i.e., sinter and pellets. Vale is actively investing in their “green
briquette” technology, which is reported to be suitable for use in a blast furnace at rates
up to 200 kg/thm [8]. Champion Iron Limited is developing a cold-bonded pelletizing
technology in partnership with Binding Solutions Limited with a similar use case [9].
Primary iron products based on both technologies are likely to become commercially
available in the near term and can be used as partial replacements for sinter. The remain-
der of the blast furnace burden will continue to be sinter or fired pellets.
Complete replacement of sinter and fired pellets with cold-bonded agglomerates is
technically challenging and unlikely to be achieved in the near or medium term. Partial
replacement of sinter and fired pellets, however, has potential as a decarbonization
strategy. A cold-bonded agglomerate does not require the significant fossil fuel-based
energy input that sinter and fired pellets require, so could contribute to the incremental
decarbonization of existing ironmaking assets. Replacing up to 200 kg sinter/thm, as
envisioned by Vale, could reduce BF-BOF steelmaking emissions by up 60 kg CO2e/t
cast steel (~3%).
Adopting this cold bonding technology will still need to consider the full supply
chain emissions impacts of the technology. For example, cement-based cold bonding
systems still require significant use of fossil fuels to produce cement. Although the total
emissions associated with cold bonding should be lower than sintering or pelletizing,
and the direct emissions may be very low, cold bonding will not be an entirely carbon-
free process.
5 Technical Challenges for Cold Bonding
Although cold bonding is already routinely employed in some steel plants, there are
significant technical challenges limiting its widespread adoption as waste management
and decarbonization strategy. First, the production of high-quality (i.e., high strength)
The Role of Cold Bonding in Sustainable Steelmaking 873
cold-bonded agglomerates relies on control of the particle size distribution and water
content of the feed materials. This is possible with good management of new arisings
but presents a significant challenge when considering the processing of historic waste
dumps or when segregation of different types of waste is not possible. Maintaining a
consistent feed to the cold bonding process is critical for producing consistent quality
agglomerates.
Second, many of the waste solids generated in a steel plant are very fine dusts or
sludges. These fine materials are more challenging to agglomerate, requiring higher
additions of binder, and in some cases producing systematically weaker agglomerates.
Third, the performance requirements for cold-bonded agglomerates to be recycled in
an iron-making process are high. It is very technically challenging to simultaneously
achieve both the physical (strength) and metallurgical (reducibility) requirements for
use in a blast furnace, and particularly for use in a direct reduction shaft furnace. The
current state-of-the-art performance from a cold-bonded briquette is sufficient for a
partial replacement of pellets or sinter in a blast furnace but inadequate for use on a
meaningful scale in the direct reduction plants.
Development of new cold bonding technologies that offer the required combination
of physical and metallurgical performance is required for widespread adoption of cold
bonding, particularly for primary iron ore applications. Reduced performance may be
acceptable for waste recycling applications where the fraction of the burden is naturally
limited by the limited volume of solid wastes available. Reduced performance will not
be accepted if these cold-bonded agglomerates are to be produced from primary iron
ore and form a major fraction of the burden.
Dilution of the ferrous material is also a concern. Many cold bonding technologies,
particularly those relying on cement or lime as a binder, significantly dilute the Fe con-
tent of the agglomerate, increasing the gangue content of the final material. This is
counter-productive from a decarbonization perspective, as it increases fuel rates and
slag rates.
Zinc accumulation is also a challenge. Zinc will report to the dust fractions of the
waste streams; if these zinc-containing dusts are entirely recycled to the process to
maximize the recovery of iron units, their outlet for zinc from the process is eliminated,
driving accumulation within the process. Because of this, there is a sensible upper limit
for the recycling of dusts within the process through cold bonding, and an alternate
disposal strategy for these dusts will still be required.
6 Conclusions
Cold bonding has significant appeal as a waste management strategy for integrated steel
plants and direct reduction plants, and as a partial replacement of indurated agglomer-
ates (sinter and pellets) produced from primary iron ore. Cold bonding also has signifi-
cantly lower energy requirements than conventional sintering and pelletizing operations,
making it attractive as a decarbonization strategy for the industry: a partial transition to
cold-bonded agglomerates, replacing 200 kg sinter/t hot metal, could reduce net steel-
making emissions intensity by 60 kg CO2e/t cast steel (approximately 3%). This level
of sinter replacement is likely to be possible for most blast furnace operations in the
near term.
Although cold bonding has been deployed in a limited number of industrial-scale
applications for several decades, significant technical challenges remain, particularly
for the use of cold-bonded agglomerates in a direct reduction plant. As direct reduction
plants proliferate as part of an industry-wide decarbonization strategy, the need for a
technically robust and climate-friendly waste management strategy will increase.
Additional research and development work in this area is needed.
Acknowledgments The authors would like to acknowledge the support of their colleagues, especially
Manuel Huerta, in completing this work.
References
1. Dawidowski L, Kolmogortseva O, Person Rocha e Pinho I, Okazaki T. 2019 Refinement to the

2006 IPCC Guidelines for National Greenhouse Gas Inventories. 4.2.worldsteel (2021). Steel
industry co-products. 2019.
2. Cameron I, Sukhram M, Lefebvre K, Davenport W. Blast furnace ironmaking. Amsterdam:
Elsevier; 2020.
3. SSAB EMEA AB: Miljörapport 2012. Luleå. 2012. p. 41.
4. SSAB EMEA AB & SSAB Merox AB Oxelösund: Miljörapport 2012. Oxelösund. 2012. p. 48.
5. SSAB Material efficiency Material efficiency – SSAB read on 2022/04/29.
6. Qatar Steel: Sustainability report 2019, Mesaieed, State of Qatar. 2020.
7. Vale: Vale announces “green briquette” capable of reducing CO2 emissions of steelmaking clients
by up to 10%. 2021.
8. Caron J, Athayde M, Lavoie F, Joyce R. Development of a cold bonding technology to high-grade
iron ore concentrate. In: Iron ore 2021 conference proceedings. 2021.
Canada as the Trailblazers in Net-Zero Carbon
Mining
Susanna Zhang1, Hakan Mustafa Tunc2 (*), and Ben Steyn1

1
BBA Consulting Ltd., Toronto, ON, Canada
susanna.zhang@bba.ca; ben.steyn@bba.ca
2
BBA Consulting Ltd., Vancouver, BC, Canada
hakan.tunc@bba.ca
Abstract. In the midst of a climate crisis, we are running out of time to change course,
decarbonize our way of mining the earth, and preserve our planet. The mining industry
already contributes 2–3% of global CO2 emissions [1]. It’s clear that we must take imme-
diate action. This is why politicians worldwide have committed to net-zero carbon emis-
sions by 2050 to limit the global temperature increase of 1.5–2 °C. Leading miners are
committed to this goal, and the mining industry is working hard to reduce its emissions.
Keywords: Net-zero carbon · Optimization · Transformative technology ·

Decarbonization
1 Introduction
In the midst of a climate crisis, we are running out of time to change course, decarbon-
ize our way of mining the earth, and preserve our planet. The mining industry already
contributes 2–3% of global CO2 emissions [1]. It’s clear that we must take immediate
action. This is why politicians worldwide have committed to net-zero carbon emissions
by 2050 to limit the global temperature increase of 1.5–2 °C. Leading miners are com-
mitted to this goal, and the mining industry is working hard to reduce its emissions.
Clean energy technology is one of the new buzzwords we often hear in discussions
about decarbonization. The world needs more of it! However, these technologies ironi-
cally require more metals and therefore more mining. So, we have to increase mining if
we want to decarbonize.
What are the metals we need? Battery metals used for electric vehicles are in high
demand (i.e. the raw materials used in the production of batteries such as lithium, and
nickel). How are we currently doing? Currently gold dominates new projects in Canada
by making up 59% of the new exploration activities and mining studies, but it does not
have a major role in decarbonization. So, there is work to be done. Canada will have to
increase its 2020 copper production by 7%, iron by 1%, zinc by 8%, and nickel by 99%
to meet demands for these ores.
https://doi.org/10.1007/978-3-031-17425-4_97
876 S. Zhang et al.
BBA sponsored the Canadian Minerals and Metals Plan (CMMP) for Quebec where
the following ten critical metals were declared essential for technological advances:
lithium, niobium, vanadium, antimony, tantalum, scandium, barite, fluorspar, manga-
nese, and tungsten.
Looking ahead, we are committed to mine the metals we need but directly minimize
our climate footprint at the same time. This means setting science-based targets to solve
some big problems. Our challenges will be:
• Declining ore grades
• Depleting resources
• High energy input requirements for extraction and processing
As graduate engineers new to the industry, we are committed to being part of the
industry’s transition to a sustainable future. This chapter will show our perspective on
the net-zero carbon goal, give a few ideas and hopefully motivate Canadian engineers
to develop transformative engineering strategies and solutions to help achieve the net-
zero goal in a short period of time.
2 Canada Is Moving Toward Net-Zero Carbon Goals
The Canadian mining industry is working toward the net-zero carbon goal. Mines are
moving toward cleantech. Capital markets and funding have a stronger incentive for
“sustainability”. The cost of capital can be as much as 20–25% higher for mining com-
panies with the lowest Environmental, Social, and Governance (ESG) scores [1].
BBA engineers are continuously working on decarbonization ideas, building path-
ways to achieve the goals, and already implementing a decarbonization execution cycle
(analyze, plan, and implement). Our value lies in our diversification and capability of
targeting GHG emissions, from energy efficiency, low to zero emission power genera-
tion, mining and processing optimization, to carbon capturing and offsetting.
Three ideas resonated with us (for now):
• Optimizing the process plant (maximum energy efficiency)
• Energy replacement
• Upgrading equipment
Idea 1: Optimize your plant regularly The BBA minerals processing group optimizes
existing process plants by auditing plants, analyzing historical plant data and recom-
mending operating changes to increase throughput at lower power draws. In particular,
the largest power consumers, the mills in the comminution circuit, needs regular checks
by statistical experts to ensure efficient operation. Slurry pooling in a SAG mill can
reduce energy efficiency by 10%, but we cannot measure it in real-time. So, we need to
continuously (or regularly) analyze our plant data to identify and solve problems.
Idea 2: Replace fossil-fuel-based energy sources One of the first steps toward electri-
fication in mining and processing requires strategies to replace fossil fuel-based energy
sources with cleaner energy sources. BBA integrated solar energy equivalent to
Canada as the Trailblazers in Net-Zero Carbon Mining 877
6,000,000 liters in annual diesel fuel at IAMGOLD in Burkina Faso as an energy alter-
native to a 225 kV transmission line. The mine utilized 15 MW of installed solar power
capacity and up to 10% renewable energy penetration on the system.
BBA assisted Tata Steel in reaching the ambitious GHG reduction targets set by
Newfoundland’s carbon pricing program in 2018. They achieved a 12% CO2 emission
reduction over 5 years. The transportation cycle time among the mineral deposits, waste
piles, and concentrator was analyzed and optimized to achieve the reported GHG reduc-
tion. Furthermore, transitioning from fossil fuels to renewable energy sources also con-
tributed to GHG reduction.
Idea 3: Upgrade equipment Governments are joining clean energy initiatives. Ontario
and ArcelorMittal Dofasco recently agreed on a $500 million investment to reduce the
GHG emissions from transformed steel production by replacing old ovens and furnaces
with new low-emission technologies.
Success story Newmont’s Goldcorp Borden Gold Project in Northern Ontario is a pio-
neer in transitioning to cleantech. The mine features state-of-the-art health and safety
controls, digital mining technologies and processes, and low-carbon energy vehicles.
3 The Vision for a Net-Zero Carbon Mine
We built a hypothetical process plant flowsheet and identified the areas of the flowsheet
where we could implement three ideas: (1) Optimize the process plant, (2) Include
alternative energy sources, and (3) Replace high-carbon emission equipment.
We considered new technologies and conceptualized future technologies (radical
ideas lead to game-changing inventions) to reduce Carbon emissions. We chose areas
that are known to contribute the most to high energy emissions and quantified the effect
of the proposed change. The base energy expenditure on this plant was high. The pre-
sentation will show the effect of the following changes:
• Pre-treat ore before grinding to reduce the energy requirement.
• Grinding plant without water or grinding media using high compression
equipment.
• Install solar and wind energy to supplement energy requirements.
• Replace all mining trucks with electrical trucks.
• Replace ovens and furnaces with low-emission equipment.
These small changes had a big effect on Carbon emissions. Imagine applying a full
cycle of innovative decarbonization ideas. We hope our follow-up papers in the coming
years will show the results of our starting efforts.
Acknowledgments Thank you to Jim Barrett, and Derek Blais for your constructive feedback and
ideas for this paper. Thank you to Jean L’Heureux, Murray Brown, and Colin Hardie for the Canadian
and global market analysis to better understand the state of the market. Finally, we would like to thank
BBA for sponsoring the co-authors to go to the COM2022 conference.
878 S. Zhang et al.
Reference
1. Legge H, Müller-Falcke C, Nauclér T, Östgren E. Creating the zero-carbon mine. 2021. https://
www.mckinsey.com/industries/metals-and-mining/our-insights/creating-the-zero-carbon-mine.
Accessed 10 Mar 2022.
Development of a Smart Database for Wastes
from the Canadian Mining and Metallurgical
Industry
Elhocine Derardja1 (*), Arash Rafiei2, and Elmira Moosavi-Khoonsari1

1
École de Technologie Supérieure (ÉTS), Montreal, QC, Canada
elhocine.derardja.1@ens.etsmtl.ca
2
Abstract. Managing wastes from the Canadian mining and metallurgical industry is a
multidisciplinary socioeconomic and environmental task. On the one hand, many indus-
trial wastes contain metals and materials, which are considered to be strategic and critical
for transformation to green technologies and the techno-economic development of soci-
ety. On the other hand, industrial wastes may contain toxic and harmful components to
the ecosystem. In addition, the diversity of these wastes in terms of physicochemical
properties creates a need for different valorization routes; which is, indeed, a scientific
and technological challenge. In this work, we present a systematic and knowledge-based
approach to generate a smart database for the wastes from the mining and metallurgical
industry with respect to circular economy policy, strategic mineral plan, and waste man-
agement regulations of Canada. The generated database and knowledge will form the
groundwork for the development of potential metallurgical routes for the valorization of
industrial wastes into in-demand products and recuperation of valuable metals and mate-
rials in a responsible and sustainable manner.
Keywords: Solid wastes · Mining and metallurgy · Circular recycling
1 Waste Classification Concepts
The concept of waste classification in the mining and metallurgical industry is usually
defined by waste management policies and environmental regulations, specific to a geo-
graphical area. Therefore, it is no surprise to observe minimum coherency among the
concepts of those classifications in different countries. For example:
• Mining wastes are classified by United Nations Environmental Programme into
acids and alkali, cyanide, heavy metals, asbestos, and uncertain solid remains [1].
• In the European Union, mining and metallurgical wastes are mainly classified into
mining excavation, drilling, metalliferous processing, and non-metalliferous pro-
cessing [2]. In addition, each waste is also categorized as “absolute entries” or
“mirror entries”. The absolute entries are defined as wastes that have environmental
hazards regardless of any threshold concentration, and mirror entries are waste with
a specific threshold concentration.
https://doi.org/10.1007/978-3-031-17425-4_98
880 E. Derardja et al.
• In Canada, the mining and metallurgical wastes belong to the “hazardous waste and
hazardous recyclable material classification”. The mining and metallurgical wastes
are divided into sub-groups with respect type of target material for processing [3].
• According to the United States, Environmental Protection Agency, mining and met-
allurgical wastes are considered “special wastes”, classified into mining, phosphate,
uranium, oil and gas, cement kiln dust, and utility wastes. The waste classification
“mining and mineral processing” is based on the type of target material for process-
ing [4].
• In Australia, the waste classification for data reporting and waste management pur-
poses is different. For the first purpose, wastes are classified into landfill, hazardous,
recyclable, and others [5].
Moreover, the mining and metallurgical waste classifications which have been pre-
sented in several academic and scientific publications mostly focus on environmental
hazard potentials (e.g., acid mine drainage) and waste management systems [6]. For
example, Lottermoser [7] classified the mine wastes into mining, processing and metal-
lurgical. According to our knowledge, few publications suggest additional global clas-
sification elements such as technical nature, hazard potential, and the best practice of
managing waste.
Consequently, the majority of current classification concepts for mining and metal-
lurgical wastes are based on environmental hazard potential, type of target material or
processing, or a combination of them. However, today, with respect to increasing global
attention to the circular economy, sustainability, and valorization; recycling of wastes,
recovery of their full value (as main products and/or by-products) and their reuse (i.e.
potential applications) must be considered. The most important advantage of this new
approach is the transition from the current reflective and preventing approach to a pro-
active and profitable one.
2 Objective
The objective of this work is to develop a waste database that is smart, proactive, mul-
tidisciplinary and knowledge-based. The Smart Waste Database will facilitate easy
access to classified, multidisciplinary, actual, and scientific information about solid
wastes from the Canadian Mining and Metallurgical industry, which are crucial for
proper waste management within the context of a circular economy. The collected,
relevant data about wastes will be classified under certain groups, e.g. chemical proper-
ties, physical properties, environmental impact index, etc. On the one hand, such an
approach to waste treatment is necessary to optimize the share of primary resources in
the production of raw materials and secure Canada’s global position in the field of criti-
cal and strategic metals and materials. On the other hand, it will contribute to minimiz-
ing the negative effect of waste disposal and landfilling on the ecosystem.
Development of a Smart Database for Wastes from the Canadian Mining… 881
3 Smart Methodology
The four pillars of the Smart Waste Database are (a) smart waste selection, (b) smart
waste characterization, (c) smart technical analysis, and (d) smart waste applications.
Figure 1 summarizes the structure of the database and different aspects of it, which will
be briefly explained below.
3.1 Smart Selection
There is a wide variety of wastes from the mining and metallurgical industry in Quebec
and Canada. The wastes will be selected based on their technical and economic impact
consulting the strategic mineral plans of the province of Quebec [8], Canada [9], and
their allies [10] as well as the growing demand for green technologies. In parallel, the
environmental impact (toxicity, reactivity, corrosivity, ignitability, leachability, etc.) of
wastes will be monitored and recorded in the database. For example, different priority
indexes from 0.0 to 3.0 can be assigned to the wastes based on their environmental
impact [6]. The order of priority for waste selection will be from wastes having both
high techno-economic and high environmental impact to wastes having either high
techno-economic or high environmental impact.
3.2 Smart Characterization
The objective of this step is to characterize the wastes to obtain maximum crucial physi-
cochemical data with minimum cost, necessary for their circular valorization. In fact,
accurate and quantitative characterization of any waste is essential before arranging any
treatment steps.
Chemical properties can be divided into two main groups bulk chemical composi-
tion and mineralogy/phase constituents. Physical properties include particle size distri-
bution, density, moisture content, calorific value, dielectric constant, and magnetism,
which can be also important for waste treatment in a circular manner.
3.3 Smart Technical Analysis
Multidisciplinary technical analysis (i.e., from different perspectives of mineral pro-

cessing, pyrometallurgy, and hydrometallurgy) of the waste, utilizing the collected
physicochemical data, is necessary for the development of recycling processes for the
recuperation of valuable metals and materials. Simulation and modeling techniques can
be also very useful in this setting, reducing hits and trials. Nowadays, with the aid of
thermochemical software [11–13], the full power of thermodynamics can be employed
to predict the best possible practice for a wide variety of wastes.
Fig. 1. Structure of Smart Waste Database and its different aspects.
3.4 Smart Applications
Data generated from the technical analysis of waste will be placed in line with the other
collected and/or available information including (a) inventory data (abundance, loca-
tion), (b) contents of critical and strategic metals and materials, (c) potential demands
from local industries, and (d) environmental hazards of wastes. Subsequently, potential
applications and end-users will be identified. Small-to-medium size plants will be able
to select a proper feed for their recycling operation. Large-scale ferrous and non-ferrous
producers can identify secondary resources which can be treated within their existing
flowsheets.
4 Potential Impact
The Smart Waste Database proposes a list of classified wastes including their charac-
teristics from the Canadian Mining and Metallurgical sector based on user concerns. In
addition, it helps to drive insight on potential waste recycling routes, applications, and
end-users according to accurate, scientific data and a multidisciplinary knowledge-
based approach. The possible impact of the database can be summarized from techni-
cal, economic, and user and market perspectives, as illustrated in Fig. 2.
Adapting a knowledge-based, multidisciplinary approach, different possible recy-
cling routes will be evaluated and the most promising ones considering concerns of the
user (i.e. both waste supplier and end-user) will be proposed. This scheme is superior
to a unidisciplinary method, solely based on either mineralogical, pyrometallurgical, or
hydrometallurgical treatment. In fact, the Smart Database can act as an automated
waste processing expert.
Certain wastes are important from an economic point of view because they are con-
sidered to be secondary sources of valuable metals and materials. Besides, the motiva-
tion of utilizing the full value of the waste and/or multi-metal recovery will enhance the
profitability of the potential recycling routes. With the aid of advanced experimental
and modeling methodology, it will become possible to develop resource (i.e., materials
and energy) efficient recycling processes with multiple products, minimize the number
of hits and trials, reduce costs, and save time.
Furthermore, it is necessary to proactively bring interdisciplinary viewpoints into
metallurgical plants and the mining sector to solve long-lasting waste problems via
finding possible treatments and end-users before they turn into environmental and
socioeconomic burdens. The results can be also adapted to certain plant needs. In addi-
tion, the generalization (interpolation and extrapolation) at different stages from waste
selection to finding potential applications will eventually become possible. For e xample,
Fig. 2. Possible impact of Smart Waste Database (“user” refers to both waste suppliers and
end-users).
some of the waste properties such as mineralogy and subsequently density, if unavail-
able, might be estimated from the waste bulk chemical composition. Or it could be that
the database would assign a similar recycling route to wastes with close physico-
chemical properties.
Other possible advantages are calculating more useful data (e.g., the carbon foot-
print, energy consumption, etc.) from the limited body of information, which can be
performed using simulation and modeling.
The database will have the capability to generate an integrated report about the waste
including the inventory level, location, physicochemical properties, possible recycling
routes, and potential applications. It enables intelligent selection of end-users for waste
in the context of the circular economy.
5 Future Collaboration
The Smart Waste Database is under development at École de technologie supérieure

(ÉTS), Montreal. The authors invite the mining and metallurgical industry, recycling
plants, governmental research and innovation centers, non-profit organizations, consult-
ing enterprises, and any other entities in Quebec and Canada, committed to a sustain-
able circular economy, to collaborate in this work. We believe establishing a symbiotic
relationship between ÉTS and different sectors of industry and specialists is pivotal for
the progression of this work. For more information about the most recent advancement
in waste database development, visit our website at https://www.researchgate.net/lab/
Sustainable-Waste-Valorization-and-Metal-Extraction-Lab-Elmira-moosavi-khoonsari.
References
1. Shahba S, Arjmandi R, Monavary M, Ghodusi J. Iron ore waste classification according to UNEP
guidelines (case study: Golgohar Mining and Industrial Complex in Iirjan, Iran). Appl Ecol
Environ Res. 2017;15:943–56.
2. DHI Water Environment Health, Swedish Geotechnical Institute, University of Science and
Technology-Krakow. Classification of mining waste facilities. Technical report, European
Commission-DG Environment. 2007.
3. Environment and Climate Change Canada Waste Reduction and Management Division. Guide to
hazardous waste and hazardous recyclable material classification. 2017.
4. EPA’s Web Archive. Special wastes. https://archive.epa.gov/epawaste/nonhaz/industrial/special/
web/html/index.html.
5. Department of Sustainability, Environment, Water, Population and Communities. Waste definitions
and classifications. Technical report, Hyder Consulting. 2012.
6. Josic L, Smith LD. It’s time to classify waste. CIM Magazine. 2018. https://magazine.cim.org/en/
voices/its-time-to-classify-waste-en/.
7. Bernd GL. Mine wastes: characterization, treatment, and environmental impacts. 3rd ed. New York:
Springer; 2010.
8. Ministère de l’Énergie et des Ressources Naturelles. Critical and strategic minerals: Québec plan
for the development of critical and strategic minerals 2020–2025. Technical report, Gouvernement
du Québec. 2020.
9. Natural Resources Canada. Canada’s critical minerals list. Technical report, Government of
Canada. 2021. https://www.nrcan.gc.ca/sites/nrcan/files/mineralsmetals/pdf/Critical_Minerals_
List_2021-EN.pdf.
10. U.S. Department of the Interior. Mineral commodity summaries. Technical report, U.S. Geological
Survey. 2020.
11. Bale CW, Bélisle E, Chartrand P, Decterov SA, Eriksson G, Gheribi AE, Hack K, Jung IH, Kang
YB, Melançon J, Pelton AD, Petersen S, Sangster CRJ, Spencer P, Van Ende MA. FactSage ther-
mochemical software and databases – 2010–2016. Calphad. 2016;54:35–53.
12. Andersson JO, Helander T, Höglund L, Shi PF, Sundman B. Thermo-Calc and DICTRA, compu-
tational tools for materials science. Calphad. 2002;26:273–312.
13. Roine A. HSC Chemistry® [Software], Metso Outotec, Pori. 2021. Software available at www.
mogroup.com/hsc.
Decanter Centrifuges and Dewatering Technology
Selection: A Case Study
Thomas Boundy1 (*) , Juan Alberto Juarez Garcia2, Moises Uriel Raya Sotelo2,
and Kathy Adams1
1
Paterson & Cooke, Golden, CO, USA
tom.boundy@patersoncooke.com
2
Koura Global, San Luis Potosi, Mexico
Abstract. Dewatering of mine tailings is an increasingly recognized value for mineral

processors enabling reduced fresh water consumption, lower risk tailings storage facili-
ties, reduced tailings storage volumes, and, in some cases, generate value-added products.
Decanter centrifuges represent a less commonly considered technology for dewatering
tailings due to their limited capacities and limited cake solids concentrations that can be
achieved compared to pressure filters. As part of a recent design of a desliming circuit in
a fluorspar concentrator, decanter centrifuges were evaluated for slimes dewatering suit-
ability in laboratory, mini-pilot, and pilot testing campaigns. Early stage laboratory cen-
trifuge tests indicated 52–66% cake solids by mass should be expected causing this
technology to be preferred to thickeners and pressure filtration technologies. When the
decanter centrifuge was advanced to mini-pilot and pilot scale tests using true decanter
centrifuges, solids concentrations by mass of 72% and 80% were obtained clearly outper-
forming early design expectations. Feed stream characteristics such as high clay content
and fine particle size distribution that may explain such behavior are discussed. These
results suggest that (1) caution should be taken when ruling out decanter centrifuges as a
dewatering technology based on insufficient cake solids concentrations in laboratory
scale test results and (2) relatively small streams of fine, especially compressible tailings
represent an attractive design scenario for decanter centrifuges.
Keywords: Decanter centrifuge · Dewatering · Tailings · Scale up · Pilot testing
1 Introduction
1.1 Fluorspar Flotation and Slimes Management
Las Cuevas mine and concentrator produce an acid-grade fluorspar product via froth
flotation. Fines or slimes removal is an important preparation step for the flotation of
many oxides in order to improve selectivity, improve final reagent grade, and reduce
reagent consumption. In 2020, a project was initiated to improve the efficiency of an
existing desliming system. The general block flow diagram of the proposed process is
shown in Fig. 1.
https://doi.org/10.1007/978-3-031-17425-4_99
888 T. Boundy et al.
Fig. 1. Block flow diagram showing desliming circuit to prepare feed for flotation.
1.2 Tailings Management Priorities
Las Cuevas has an interest in moving towards increased water recovery from tailings
and storage tailings with reduced ratios of contained water in order to manage water
availability in an arid climate and support the Global Industry Standard on Tailings
Management. In previous work, Las Cuevas studied filtered tailings approaches to tail-
ings management and found extremely high capital costs associated with very poor
tailings filterability. Therefore, instead of blending separated slimes with final flotation
tailings for treatment, a strategic decision was made to attempt to dewater slimes
separately.
1.3 Tailings Dewatering Technologies
A combination of thickeners and filters is most commonly employed to dewater mineral

processing slurries. However, due to the known poor thickening and filtration perfor-
mance of Las Cuevas tailings and the final particle size distribution of the fines, a trade-
off study considering alternative options was conducted.
One of the key alternative types of dewatering equipment considered was the
decanter centrifuge. Decanter centrifuges are occasionally employed in dewatering
mineral processing products and wastes, though much less frequently than various fil-
tration technology.
Decanter centrifuges operate by injecting slurries into a cylindrical bowl with a coni-
cal section at one end rotating at high speed. Higher density phases are forced up against
the bowl wall. A scroll internal to the bowl rotates at a slightly higher speed to the bowl
causing the higher density phases to be conveyed up the cone and out of the centrifuge.
The lower density phase flows over a weir and is discharged at the opposite end of the
centrifuge. This behavior is diagrammed in Fig. 2.
Decanter Centrifuges and Dewatering Technology Selection: A Case Study 889
Fig. 2. Decanter centrifuge operation.
1.4 Decanter Centrifuge Scale-Up Estimates
To estimate the performance of decanter centrifuges, various centrifuge vendors offer

laboratory spin testing in benchtop centrifuges to imitate the forces experienced in a
full-scale decanter centrifuge [1]. Such laboratory tests are often used in engineering
study phases to support the comparison of decanter centrifuges against other dewater-
ing operations such as filtration. If selected for further study, pilot testing is often pur-
sued prior to an installation of a decanter centrifuge to provide more reliable performance
estimates.
A number of scale-up parameters have been presented in the academic literature
enabling comparison between centrifuges across various sizes. Of these, the Sigma fac-
tor is perhaps the most well known [2]. The Sigma factor has been developed to relate
decanter or disk centrifuges of one geometry and set of operating parameters to another
and across types of centrifuges including laboratory benchtop centrifuges [3]. In spite
of the relatively long history of such models, the confidence with which engineers are
able to estimate the performance of full-scale decanter centrifuges based on pilot scale
data, let alone laboratory spin test data, remains limited [4].
This chapter presents findings of Las Cuevas slimes dewatering test work program
through its different phases and highlights key challenges experienced with regard to
the estimation of decanter centrifuge performance across engineering phases.
2 Laboratory Campaign Materials and Methods
2.1 Slurry Characterization
Samples of the existing desliming cyclone overflow were collected from the plant and
characterized for particle size distribution, slurry solids concentration, chemical com-
position, and mineralogy.
• Slurry solids concentration was determined gravimetrically via oven drying of
samples.
• Particle size distribution was determined via laser diffraction using a Microtrac
Model S3500 adhering to ISO 13320-2009.
• X-ray diffraction (XRD) was used to determine minerals present in the cyclone feed.
• X-ray fluorescence (XRF) analysis was conducted by the technical services labora-
tory in San Luis Potosi, Mexico serving Las Cuevas to quantitatively determine the
chemical composition of the solids.
Table 1. Desliming cyclone feed quantitative mineralological analysis as determined by XRD.

Testing round Round 1 Round 2 Round 3
Vendor A B A
Centrifuge Benchtop Benchtop centrifuge and 150 mm 300 mm diameter decanter
used centrifuge diameter decanter with 2:1 length to with 3:1 length to diameter
diameter ratio ratio
Reference – Figure 3 Figure 4
pictures
2.2 Dewatering Test Work
Dewatering test work was conducted in a phased approach to evaluate the promise of
various technologies. Test work was conducted on thickening, pressure filtration, and
centrifugation. Performance estimates obtained by the thickening and pressure filtra-
tions tests were poor, and only the centrifuge tests will be reported here.
Centrifuge testing was conducted by various centrifuge equipment manufacturers.
Initial laboratory spin tests were conducted by Vendor A. These tests were performed in
a Rotina 380R benchtop centrifuge by subjecting existing desliming cyclone overflow
to various durations of various centrifugal forces.
Following successful results from the project perspective, second and third rounds of
centrifuge tests were conducted. Fresh samples of existing desliming cyclone overflow
were conducted prior to each round of testing, and slurry solids concentration and par-
ticle size distribution were measured for each fresh sample. A summary of the tests
conducted in each round of testing is provided in Table 1.

3.1 Slurry Characterization
The desliming cyclone overflow used in the study had a fine particle size distribution
with a D80 of approximately 20 μm. The complete particle size distribution is shown in
Fig. 5.
XRD results on the cyclone feed stream shown in Table 2 revealed four major min-
eralogical components of the ore: quartz, calcite, fluorite, and various clays with smec-
tite representing the dominant clay species. XRF analysis performed across five
different size intervals revealed quantitatively the partitioning of these minerals across
the size fractions as shown in Table 3. The size fraction-specific data highlights the
enrichment of gangue minerals in the −20 μm size fraction and the limited product loss
associated with desliming.
Fig. 3. Round two decanter centrifuge (Vendor B).
3.2 Laboratory Testing (Round One)
Key results from the first round of laboratory centrifuge spin tests are shown in Table 4.
Based on these tests Vendor A suggested a solids concentration by mass of around 65%
should be expected in the centrifuge solids product assuming fairly standard feeding
rates to a decanter centrifuge. This made the decanter centrifuge an attractive option
relative to alternative dewatering concepts and further test work was pursued.
3.3 Additional Spin Tests and Mini-Pilot (Round Two)
Round two of the centrifuge testing was performed on a desliming cyclone overflow
sample with similar characteristics to the cyclone overflow sample tested in the early
laboratory trials in terms of solids concentration and particle size distribution. Spin tests
in benchtop and a benchtop centrifuge were used to benchmark the slurry of round two
against round one. Round two spin test estimates of sediment solids mass concentration
(52–54%) were lower than the round one estimates (62–66%). However, despite the
less concentrated spin test solids estimate, the decanter centrifuge trial performed in
round two gave a much more solids dense product of 73% solids by mass. Round two
results are shown in Table 5. A picture of this solids product is shown in Fig. 6 illustrat-
ing the paste-like consistency of the decanter centrifuge solids product indicative of the
high fines and clay concentrations.
Based on these results, Vendor B suggested that they would expect the cake solids
concentration produced by a full-scale decanter centrifuge to be about 55% solids by
mass and that the small-scale decanter results were likely an anomaly. This led to some
Fig. 4. Round three decanter centrifuge (Vendor A).
100%
90%
80%
70%
Cumulative Passing
60%
50%
40%
30%
20%
10%
0%
0.01 0.1 1 10 100 1000 10000
Particle Size (µm)
Fig. 5. Round one desliming cyclone overflow particle size distribution.
concern about the water recovery estimates forcaseted for the project; however, the
additional larger round of piloting was scheduled and conducted for comparison.
3.4 Large Pilot (Round Three) and Centrifuge Testing Comparison
Large pilot scale testing at Vendor A was conducted with a sample of desliming cyclone
overflow. A wide array of processing conditions was tested, but in all tests, the solids
mass concentration of the solids product was between 77% and 80%. The optimal oper-
ating condition determined from Round 3 testing is presented in the far right column of
Table 6. Included in this comparison is Ambler’s sigma factor and the ratio of flow rate
Table 2. Desliming cyclone feed quantitative mineralological analysis as determined by XRD.
Mineral Quantity (%)

Quartz 4
Calcite 23
Fluorite 38
Total clay 35
Smectite 28
Illite 2
Kaolinite 5
Table 3. Desliming cyclone overflow mineral composition by size fraction determined by XRF.
Size fraction Mass fraction Composition by mass
Top size (μm) Bottom size (μm) (%) SiO2 (%) CaCO3 (%) CaF2 (%) Other (%)
– 300 0 7 14 43 37
300 150 0 12 20 62 7
150 75 4 6 10 82 3
75 20 19 5 9 84 2
20 – 76 27 21 37 15
to sigma factor which is a commonly employed scale-up parameter for centrifuges

where more flow per sigma represents less intense centrifugation.
Changing operating conditions at the plant caused a lower solids concentration and
slightly finer particle size distribution to be produced by the desliming cyclone and used
for testing. Nonetheless, the large pilot test showed the same trend of higher sediment
solids concentrations associated with decanter centrifuges than with bench top centri-
fuges as round two had despite a lower flow to sigma ratio.
3.5 Phenomenological Understanding
In addition to the classic Sigma factor, other models including the Leung number [5],
and much more sophisticated models relying upon computational fluid dynamics have
been developed [6]. While the increasing sophistication of these models has led to
improved estimates of full-scale performance in some cases, these models rely on cali-
bration to some form of measured test data. This makes the comparison of centrifuges
of the same type (e.g. small decanter centrifuge versus a larger decanter centrifuge)
increasingly valuable. However, benchtop centrifuges do not test the impacts of scroll
compaction or drainage on collected sediment which fundamentally limits the insight
that these tests can provide to scale up efforts. To appreciate the differences between a
laboratory benchtop centrifuge and a decanter centrifuge, a depiction of the decanter
centrifuge is produced in Fig. 7 with key zones related to solids dewatering behavior
highlighted.
When solids first enter the decanter centrifuge, they are fed to the settling zone of the
decanter in which more dense materials are propelled via centripetal force towards the
Table 4. Centrifuge spin test solids concentration estimates.

Parameter Units Test 1 Test 2
Relative centrifugal force G/g 1000 3000
Spin time min 5 5
Temperature °C 20 20
Sediment texture – Semicompact Compact
Sediment solids concentration by mass % 66 62
Table 5. Round 2 centrifuge testing summary.

Test Spin test Small decanter
Feed slurry solids mass concentration 23% 23%
Feed slurry D80 (μm) 11.6% 11.6%
Residence time 5 min <10 s
Relative centrifugal force (G/g) 3000 2000
Type of centrifuge Bench top 150 mm diameter decanter
Discharge solids mass concentration 52–54% 73%
Fig. 6. Pictures of collected decanter solids product showing paste-like consistency.
bowl wall leaving a zone of the clarified solution near the center of the centrifuge. This
zone represents the initial separation of high and low-density phases and is well repre-
sented by the laboratory benchtop centrifuge.
Once solids have compacted against the bowl, they are propelled towards the solids
discharge end of the decanter centrifuge via the internal scroll. Upon contact of this
solids bed with the rotating scroll, shear is imparted to the bed causing some compac-
tion. The intensity of this compaction is intensified as compressive force is added to the
bed of solids when they encounter the conical section of the decanter. This is referred
to as the “compaction zone” in Fig. 7. It has been recognized that compaction of solids
occurs in the conical section of the decanter, especially for “compressible” materials
[1]; however, this behavior is not well captured by laboratory spin tests.
Fig. 7. A diagram illustrating the movement of solids and liquids in a typical decanter centrifuge in
which different separation zones are dominant.
Table 6. Comparative summary of centrifuge testing results.

Parameter Unit Value
Test work vendor – A A B B B A
Testing round – 1 1 2 2 2 3
Solids concentration by mass % 11.7 11.7 11.6 11.6 11.6 5.5
D80 μm 20 20 23 23 23 15
Mass passing 5 microns % 37 37 25 25 25 50
Centrifuge type – Lab Lab Lab Lab Decanter Decanter
Relative centrifugal force G/g 1000 3000 1000 3000 2000 2610
Flow rate or equivalent m3/h 2E- 2E-04 6E-04 6E-04 0.056 6.1
04
Sigma factor m2 1.14 3.42 3.80 11.4 122 1017
Flow rate/sigma m3/h/m2 2E- 5E-05 2E-04 5E-05 5E-04 6E-03
04
Sediment solids mass concentration % 66 62 55 53 73 78
Finally, as solids are conveyed up the conical section of the decanter and out of the
liquid pool, the centripetal force draws liquid out of the formed solids bed in a drainage
mechanism. This area is called as the “drainage zone” in Fig. 7 and has been recognized
as a more active phenomenon in “incompressible materials” [1]. However, as with the
activity of the compaction zone, this behavior is not well captured by the laboratory
spin test.
Some centrifuge equipment vendors are known to be developing additional bench-
top centrifuge-based tests to provide additional insight into the behavior of centrifuge
solids under compaction and drainage conditions. A large history of testing these new
laboratory protocols with benchmarking against industrial centrifuge performance will
be necessary to improve the reliability of laboratory-scale centrifuge testing, but these
are important efforts to reduce the likelihood that centrifuge technology is ruled out
prematurely due to conservative laboratory estimates.
4 Conclusions
These results suggest that:

1. While it has been reported that the classic laboratory spin test typically produces a
conservative estimate of solids product solids concentration, Las Cuevas testing
demonstrates that spin test results may on occasion be very conservative.
2. Caution should be taken when ruling out decanter centrifuges as a dewatering tech-
nology based on insufficient cake solids concentrations in laboratory scale test
results
3. Relatively small streams of fine, especially compressible tailings represent an attrac-
tive design scenario for decanter centrifuges that may be prone to high sediment
product solids concentration improvements between laboratory and decanter
centrifuges.
Consideration of the separation mechanisms within a decanter centrifuge suggests
that there may be room for the development of improved laboratory tests capable of
giving insight into the compaction and drainage characteristics of solids to be collected
via centrifuge. Confidence in the results of such a test would allow improved ability to
evaluate the dewatering performance of centrifuges in early-stage trade-off studies.
References
1. Leung WW-F. Industrial centrifugation technology. McGraw-Hill; 1998.

2. Ambler CM. The theory of scaling up laboratory data for the sedimentation type centrifuge. J
Biochem Microbiol Technol Eng. 1959;1:185. https://doi.org/10.1002/jbmte.390010206.
3. Ambler C. Centrifugation equipment theory. Ind Eng Chem. 1961;53:430.
4. Menesklou P, Sinn T, Nirschl H, Gleiss M. Scale-up of decanter centrifuges for the particle sepa-
ration and mechanical dewatering in the minerals processing industry by means of a numerical
process model. Fortschr Mineral. 2021;11:229. https://doi.org/10.3390/min11020229.
5. Leung WWF. Inferring in-situ floc size, predicting solids recovery, and scaling-up using the
Leung number in separating flocculated suspension in decanter centrifuges. Sep Purif Technol.
2016;171:69. https://doi.org/10.1016/j.seppur.2016.07.010.
6. Zhu G, Tan W, Yu Y, Liu L. Experimental and numerical study of the solid concentration distribu-
tion in a horizontal screw decanter centrifuge. Ind Eng Chem Res. 2013;52:17249. https://doi.
org/10.1021/ie401902m.
Slag Resource Recovery Via Air Granulation
Santiago Faucher1, 2, 4 (*), Daniel Park1, and Amit Shrivastava3

1
Ecomaterials Inc., Hamilton, Canada
santi@ecomaterials.net
2
Ecomaister Co. Ltd., Seo-gu, Incheon, Republic of Korea
3
Ecomaister Beads India Private Ltd., Raigarh, CT, India
4
Insight R&D Inc., Hamilton, Canada
Abstract. Slag remains one of the most important undervalued resources produced by
the pyrometallurgical industry, where it accounts for approximately 10–98% of the mass
and energy flow from a pyrometallurgical process. That is, the slag is commensurate in its
material and energy footprint to the metals, mattes, and alloys produced, yet it is treated
not as a resource but as a waste. Advances in the minerals sector have allowed meager use
of this resource in certain applications, but substantial work remains by industry and
government to truly valorize the volumes of slag produced. Over the last 30 years,
Ecomaister has been contributing to this valorization through the development of its Slag
Atomization Technology to both granulate slag into products with air and recover its
energy content. These efforts have led to the construction and operation of over 30 com-
mercial Slag Atomization Plants, two of which have been for the last decade successfully
recovering energy from slag. This air-driven slag granulation technology has been applied
to the steel and non-ferrous sectors at full-scale, granulating up to 200,000 tonnes per year
of slag per unit. Recently, we have demonstrated further scale-up of the technology to
granulate high slag rates from blast furnaces and have implemented new adaptations of
the technology to produce a wider array of slag products. Both historic and arising devel-
opments in Slag Atomization Technology and slag applications are presented and dis-
cussed with supporting data. Challenges for further adoption of slag as a resource and air
granulation as a technology will be discussed and outlined.
Keywords: Slag · Cement · Energy · Granulation · Atomization · Circular

Economy · Sustainability · Resources · Waste · Valorisation
1 Introduction
Ecomaister began to study the granulation of molten materials in 1988 when it acquired
water granulation technology to produce steel shots for surface preparation (steel-shot
blasting). Around this time, and in relation to steel shot manufacturing, it started to
explore methods of metal recovery from steel slags leading in 1993 to its first patent on
the subject [1]. This first process was based on water granulation of steel-laden slag,
followed by separation of the metal-rich and poor (slag) streams. Safety improvements
led to the use of air as a granulation medium and the elimination of water. This created
a completely dry granulation (atomization) process and novel slag materials with
https://doi.org/10.1007/978-3-031-17425-4_100
898 S. Faucher et al.
Fig. 1. Ecomaister Slag Atomization Plant (full-scale pilot-plant) Posco Pohang.
interesting properties. Building on these discoveries, Ecomaister began to further

develop its technology and applications for atomized slag products.
2 Slag Atomization Plants: From Pilot to Production
While preliminary works had been undertaken by Ecomaister at the laboratory scale,
the first full-scale pilot-plant designed by Ecomaister to granulate slags with air was
established at POSCO-Pohang in 1999 (Fig. 1). This full-scale pilot-plant allowed for
the exploration of a wide range of process conditions of importance to the operability
of a Slag (air) Atomization Plant (SAP) and to the properties of the products produced
therefrom. Learnings from this first plant were then built upon to create a first commer-
cial SAP at Hyundai Steel in Dangjin (Fig. 2), and from there further improvements
were made to establish a first production plant at SeAH Steel – Gunsan (Fig. 3).
Generation by generation, the technology has been improved and optimized by
Ecomaister’s researchers, engineers and partners, as they progressed through the design
and supply of 30 SAPs at 15 locations (Fig. 4).
3 Slag Energy Recovery
With the development of its well-established slag atomization technology, Ecomaister

turned its attention to slag energy recovery and implemented three slag-energy recovery
plants. Two of these plants are now in their second decade of operation (Fig. 5).
Slag Resource Recovery Via Air Granulation 899
Fig. 2. Ecomaister Slag Atomization Plant at Hyundai Steel Dangjin.
Fig. 3. Ecomaister’s Slag Atomization Plant at SeAH Steel Gunsan.
4 Valorisation of Slag Resources
While Slag Atomization Plants (SAPs) lead to economies in slag handling OPEX and
CAPEX [2], the primary economic driver for their adoption is the conversion of under-
utilized slag resources into stabilized products. Air atomization of slags provides a slag
quenching rate equal to or superior to that achieved with water granulation. Thus, like
in water granulation, high glass contents are observed in atomized slags. However,
unlike water granulation, air atomization produces particles of smooth surface, low
porosity, and high density [3]. An example of this is shown in Fig. 6. These particles are
found to have substantially lower leaching rates and greater environmental stability
than non-atomized slags on account of reduced surface defects and the fast quenching
rates that limit free-lime formation, [4–6].
Fig. 4. Ecomaister’s Slag Atomization Plants and Locations (multiple plants per site).
Fig. 5. Ecomaister Slag Atomization and Energy Recovery Plant.
For ladle-furnace slags, air atomization produces slag granules that when ground
demonstrate cementitious properties [7]. Overall, these reduced leaching properties,
higher densities, greater hardness, kinetically stabilized phases, etc… allow what would
otherwise be considered waste to be valorized into a number of products. Examples of
such applications that have been commercialized by Ecomaister from air atomized
slags are shown in Fig. 7.
Fig. 6. Air atomized steel slag (PS Ball™) produced by Ecomaister.
Fig. 7. Air atomized slag applications developed and commercialized by Ecomaister.
5 Recent Advances in Slag Atomization
5.1 Part I: Blast Furnace Slag Atomization at Full-Scale
Ecomaister began studying the slag atomization of iron-making blast furnace (BF)
slags in 2013 and demonstrated the production of atomized BF slags at a laboratory
scale (Table 1, POSCO-2013 and SSAB-2015). Early laboratory scale experiments that
remelted BF slags and then air atomized them to produce slag granules gave high glass
content [8]. For POSCO slag, air atomization produced a glass content of 92.5% as
compared to the 97.5% achieved through water granulation. Later trials with SSAB slag
produced air atomized slag granules with 99.9% glass content as compared to the origi-
nal glass content of the water granulated slag at 98%. Thus, laboratory scale work
Table 1. Glass content of water and air atomized blast furnace slags.
Slag Source Scale Water Gran. Slag Air Atom. Slag
POSCO Lab 97.5% 92.5%
SSAB Lab 98.0% 99.9%
demonstrated some first positive proofs of concept with the air atomization of blast
furnace slags.
Scale-up risks with air atomization however remained. Firstly, it was not known if
the high glass contents could be maintained at a larger scale. As atomization is scaled-
up, the surface area of the SAPs decreases in relation to the slag amount processed and
this decreases the slag cooling rate. Lower slag cooling rates might lower the glass
content of the atomized slag product, and thus reduce its viability as a cement substi-
tute. Tests, therefore, had to be carried out to address this concern and determine if high
glass content could be achieved at full-scale.
The second risk is related to the ability of Ecomaister’s SAP to process slag at rates
required for blast-furnaces (peak rates of 10 tonnes of slag per minute). Ecomaister’s
SAPs had been historically designed to treat up to 3 tonnes per minute of slag, and it
was unknown if the technology could be translated to higher slag flowrates. Such high
flowrates could potentially further negatively affect the cooling rate and glass content
on account of higher heat loads.
In order to address these two risks, and determine the feasibility of full-scale blast
furnace slag granulation, trials were planned and undertaken with Jindal-Steel-Works
Vijayanagar (JSW) [8]. At this site, Ecomaister already had three slag atomization
plants servicing steel slags, and one could be spared to undertake the necessary blast
furnace slag test work over a two-week period.
The test work consisted of collecting slag by slag-pot-carrier from JSW’s No. 2 blast
furnace, carrying it to Ecomaister’s full-scale SAP, and atomizing the BF slag in the
SAP. Figure 8 shows how the 40-tonne slag sample was collected, transported, and
atomized. Also shown is the resulting air atomized slag deposited in the slag yard. In
total, the tests processed seven slag pots under a wide range of atomization conditions,
at a total of 280 tonnes of slag.
Varying slag flowrates were explored during the trials, ranging from 1.4 to 6 tonnes
of slag per minute, by controlling the flow of slag using the slag-pot carrier. These rates
are similar to those produced from a blast furnace. For each trial, samples were col-
lected and segregated for later assay and cement work. Figure 9 shows the amorphous
content of the atomized slag samples produced, at their corresponding slag atomization
rate. Amorphous contents ranged between 93.7% and 98.8%, which compared favor-
ably against BF slag that was water granulated by JSW (glass content of 94%). This
work thus demonstrated for the first time that high glass contents are achievable through
full-scale air atomization for BF slags; without the use of water.
In order to validate that the air atomized granules were a viable alternative to water
granulated granules, cement mortar tests were undertaken where ground-atomized blast
furnace slag (GABFS) substituted Ordinary Portland Cement in mortar mixes. Figure 10
summarizes the results of the mortar compressive strength test work completed.
Fig. 8. (a) Collection of blast furnace slag from a blast furnace in slag pot, (b) transport of pot and slag
by slag-pot-carrier to the SAP, (c) full-scale atomization trials at the SAP, (d) atomized blast-furnace
slag granules and traces of wool on atomization yard floor.
Fig. 9. Amorphous ratio (glass fraction in %) for atomized blast furnace slag samples at full-scale.
Rates in tonnes per minute indicate the rate at which samples were atomized.
Fig. 10. Compressive strength of OPC mortars as a function of substitution by ground-atomized BF

slag (25% and 50% substitution) of varying glass content (93.8% vs. 98.8%).
Mortar compressive strengths were found to be slightly improved with a 25% sub-
stitution of OPC with GABFS. A further substitution to 50% resulted in 56-day mortar
strengths which were substantially higher than those of the OPC baseline (35.4 vs.
42.8 MPa). Thus, this work showed conclusively that SAT could be utilized to produce
high-quality GABFS at full-scale.
5.2 Part II: Direct Slag Atomization from Steel EAFs at Full-Scale
The steel industry is shifting to the use of blast furnace iron and direct-reduced iron as
feed-stocks for EAFs. This has resulted, particularly in India, in an increase of slag
generation rates that make slag handling by slag-pot-carrier difficult. Slag generation
now exceeds the capacity of historical slag-pot-carrier (SPC) equipment. Given these
difficulties, Ecomaister explored and demonstrated with Jindal Steel and Power Limited
(JSPL) the feasibility of replacing slag-pots and slag-pot-carriers with direct slag
atomization.
This innovation consists in granulating the slag directly into the slag alley via slag
atomization equipment placed below the EAF. Figure 11 shows the results of the trials,
which have been successful to date. This work demonstrates the potential for process-
ing large slag volumes safely from EAFs without the use of SPCs.
Fig. 11. Atomization of steel slag flowing directly from an EAF at JSPL Angul.
6 Challenges Facing Slag Resource Utilization
Although blast furnace slags are utilized to a greater degree today to offset OPC utiliza-
tion, it has taken decades for them to reach partial adoption. This is despite the material
providing several technical, environmental, and commercial benefits over OPC. Lack of
market competition, perceived technical risk, lack of industry knowledge, and the
absence of strong environmental legislation that forces adoption are but some of the
factors causing this slow uptake rate.
In comparison to blast-furnace slag, the use of steel and non-ferrous slags is rare. As
a result, they are confined to stock piles that create long-term liabilities. However, these
under-utilized resources, if processed adequately, can find use in society.
Air atomization has for example opened up the market for the use of steel and non-
ferrous slags in Korea and India. The environmental stability provided by slag-
atomization, and the scarcity of sand, have made atomized slag a desirable sand
substitute. However, it has taken nearly 30 years of effort to get Korea to partially adopt
atomized slag products.
For slags (atomized or otherwise processed) to reach further widespread use, signifi-
cant advances in research, development, engineering, marketing and legislature will be
required to shift society to these more sustainable materials. Carbon tax credits are a
good way to accelerate this change, and it is hoped that this will catalyse a shift towards
slag utilization.
7 Conclusions
Ecomaister has dedicated 30 years to the conversion of slags into products. This jour-
ney has been challenging, interesting, and rewarding. The results to date have yielded
over 30 atomization plants and a wide range of new atomized slag materials that re-
enter the circular economy. Expanding the adoption of this technology beyond Korea
and India will require a concerted effort and shift in our society’s perspective on slag,
and wastes in general. Such materials should no longer be perceived as waste, but rather
as under-utilized resources. The ramp-up of carbon tax credits will surely help us in this
ideological transition.
References
1. Oh OS, Cho WD, Oh ES. Korean Patent Application 9965/1993.

2. Mostaghel S, So L, Cramer M, Faucher S, Lee SK, Oh S-Y. The fourteenth international ferroal-
loys congress, 2015.
3. Oh SY, Moon YB. Korean Patent No. 1012300330000.
4. Oh SY. US Patent Application 2009/0193849 A1.
5. Mostaghel S, So LC, Faucher S, Gour J, Lee S, Oh S-Y. 5th International conference on process
development in iron and steelmaking, Lulea, 12–15 June 2016.
6. Bolen J, Mostaghel S, So L, Faucher S. AISTECH 2017, Nashville, May 2017.
7. Kim JM, Kwak EG, Oh SY, Kim CH, Kang KW, Heo DC. US Patent No. US20130269573A1.
8. Faucher S, Oh SY, Shim SI, Im SJ, Kumar DS, Kumar P, Lindvall M, Orrling D, Nordberg LO. The
11th Molten Slags, Fluxes and Salts [MOLTEN 2020], Seoul, 25–29 May 2020.
Using Bio-Based Surfactants as Frothers in Froth
Flotation to Improve Renewable Carbon Index
Laura Benavides1 (*)

1
Integrity Mining and Industrial, Tucson, AZ, USA
LBenavides@IntegrityMiningandIndustrial.com
Abstract. Developing quantifiable environmental social and governance (ESG) criteria

is an opportunity for mining companies to differentiate themselves. With a large focus
recently on GHG emissions and clean energy, there is a large opportunity for companies
to demonstrate “environmental friendliness” and “sustainable supply”. There are several
reliable tools that can be adopted by the industry to set transparent standards for socially
responsible products. Integrity Mining and Industrial (IMI) bases the carbon signature of
its biopolymer-based products on its renewable carbon index. The Renewable Carbon
Index (RCI) readily classifies the origin of carbon in a molecule, i.e., modern vegetative
sources or extracted petroleum derivatives. Specifically, it measures the number of car-
bons derived from renewable sources compared to the total number of carbons in the
product. The more the renewable ingredients in a product, the higher the RCI. A typical
“green” product will have an RCI higher than 75%, with 100% possible for products with
fully bio-based ingredients.
Keywords: Frother Flotation · Surfactants · Bio-Surfactants
1 Introduction
1.1 Renewable Carbon Index
Developing quantifiable environmental social and governance (ESG) criteria is an

opportunity for mining companies to differentiate themselves. With a large focus
recently on GHG emissions and clean energy, there is a large opportunity for companies
to demonstrate “environmental friendliness” and “sustainable supply”. There are sev-
eral reliable tools that can be adopted by the industry to set transparent standards for
socially responsible products. Integrity Mining and Industrial (IMI) bases the carbon
signature of its biopolymer-based products on its renewable carbon index. The
Renewable Carbon Index (RCI) readily classifies the origin of carbon in a molecule,
i.e., modern vegetative sources or extracted petroleum derivatives. Specifically, it mea-
sures the number of carbons derived from renewable sources compared to the total
number of carbons in the product. The more the renewable ingredients in a product, the
higher the RCI. A typical “green” product will have an RCI higher than 75%, with
100% possible for products with fully bio-based ingredients.
https://doi.org/10.1007/978-3-031-17425-4_101
908 L. Benavides
1.2 Frothers
Integrity Mining and Industrial has developed a bio-based surfactant technology with a
tunable HLB range between 7 and 19. These bio-based surfactants can be used to
replace leading synthetic polymers and/or can be blended with other surfactants to meet
sustainability and environmental regulations. Currently, IMI bio-based surfactants con-
tain a renewable carbon index (RCI) between 60% and 95% depending on the applica-
tion. In this preliminary testing, IMI bio-based surfactants were evaluated and compared
to commodity frothers used in the industry during froth flotation in a standard Cu sul-
fide lab-scale rougher circuit.
RCI classifies the origin of carbon in a molecule and determines if it is from bio-
based sources or petroleum-based derivatives. Specifically, it is a percentage calculated
by dividing the number of carbons derived from renewable sources by the total number
of carbons in the product – the higher the percentage, the more renewable carbon is
present. A typical “green” product will have an RCI higher than 75%, with 100% pos-
sible for products with fully bio-based ingredients. Bio-based ingredients are derived
from naturally occurring sources such as plants and other renewable materials.
2 Discussion
The focus of this paper will be to document ongoing evaluations of IMI bio-based sur-
factants when compared to alcohol-based frothers, propylene glycol, and other glycol
ethers. By replacing synthetic frothers with bio-based surfactants, mining companies
can take control of the renewable carbon being used in their products.
Prior to comparing Tegrasurf to other frothers, we evaluated four different iterations
of TegraSurf on its own in order to determine the best “frother” for use in our compari-
son testing (Table 1). Table 1 compares the copper recovery when using four different
TegraSurf products (with different HLB ranges), and also lists their contact angles. In
the flotation of minerals, contact angle plays a major role in hydrophobicity. The higher
the contact angle, the greater the hydrophobicity and ideally, the greater the floatability
[1]. Since most of these products have a similar contact angle, it was decided to use
TegraSurf 160 for the initial round of comparison testing which had the highest Cu
recovery. Further testing, will expand upon TS90, which also had a promising
recovery.
TS160 was compared to five common frothers used in the industry, labeled here as
A-E. Testing was conducted in standard froth flotation cells, with 1000kg charges of
Table 1. TegraSurf evaluations.

Cu Recovery, % HLB Contact Angle
TS 90 85.49 9 37.2
TS 126 80.21 12 –
TS 160 85.57 16 37.1
TS 190 82.37 19 37.9
Using Bio-Based Surfactants as Frothers in Froth Flotation to Improve Renewable… 909
89 88.61 88.39
88.3
88.06
88
87
86.04
86 85.57
84.8
85
84
83
82
Product A Product B Product C Product D Product E TS160 Baseline
Fig. 1. TegraSurf 160 bio-based surfactant versus common synthetic frothers.
copper ore provided by a minesite in Nevada. The floats were conducted for 11 minutes
each. The baseline in Fig. 1 refers to the currently used frother at the mine site. Based
on this data, TS160 performs better than the baseline, and similarly (within 5% Cu
Recovery) of the major synthetic frothers commonly used in the industry. This data is
promising, and additional testing will be done on different ores as well as potential
blends between TegraSurf and the other frothers in order to increase the metal recovery.
The TS160 was tested in triplicate, and the average Cu recovery was used. The other
frothers were based on one test.
3 Conclusions
By implementing natural products composed of renewable carbon, mines, distributors

and chemical manufacturers can begin to assume ownership over a product’s carbon
lifecycle. Controlling which products are used, allows for mining companies to have a
direct effect on the processes that can be improved down the line.
Increased focus on GHG emissions and clean energy allows mining companies to
demonstrate environmental friendliness. It may be impossible to switch to 100% bio-
based reagents in our lifetime, but by taking small steps to improve the renewable car-
bon index of the products we consume, we can achieve large changes in the industry.
By evaluating a “frother” made from bio-based ingredients, with a high renewable car-
bon index, IMI hopes to begin to shift the conversation towards commodity products
that are still high in technical value and also considered environmentally friendly.
Reference
1. Pattanaik A, Venugopal R. Role of surfactants in mineral processing: an overview. In: Dutta AK,
editor. Sufactants and detergents; 2019. https://doi.org/10.5772/intechopen.85947.
Microwave Impacts on More Sustainable Kimberlite
Processing
Muhammad A. Rasyid1, Azlan Aslam1, Arash Rafiei1 (*), Agus P. Sasmito1,

and Ferri Hassani1
1
Mining and Materials Engineering Department, McGill University, Montreal, QC, Canada
arash.rafiei@mail.mcgill.ca
Abstract. Diamond mining from kimberlite ore has some challenges related to sustain-
ability: (1) low recovery of diamond, (2) high waste management cost, and (3) high pro-
cess water consumption. These problems are closely related to the liberation of diamonds
in the crushing and separation process. Low diamond recovery comes from unlocked
diamond particles from their gangue mineral due to inefficient crushing. In kimberlite ore
processing, more than 99% of the feed will become tailings (waste). Therefore, the mine
needs bigger tailings storage facilities (TSF) which consume higher costs to maintain
relatively to any regular TSF. Typically, 70% of these tailings materials are fine fractions
(<1 mm) produced directly from the crushing circuit [1]. These fine fractions are counted
as ore loss because any diamonds inside cannot be recovered by the dense media separa-
tion process. Ultimately, handling fine fractions requires more water consumption in the
thickening and slurry pumping to the final TSF [1, 2]. The sedimentation rate is also sig-
nificantly slower, and thus coagulant and flocculant reagents consumption becomes nec-
essary [3]. Selective liberation by producing coarser crushing products can potentially
minimize these issues.
Keywords: Microwave · Comminution · Kimberlite · Size distribution · Heavy

liquid separation
1 Introduction
Diamond mining from kimberlite ore has some challenges related to sustainability: (1)
low recovery of diamond, (2) high waste management cost, and (3) high process water
consumption. These problems are closely related to the liberation of diamonds in the
crushing and separation process. Low diamond recovery comes from unlocked dia-
mond particles from their gangue mineral due to inefficient crushing. In kimberlite ore
processing, more than 99% of the feed will become tailings (waste). Therefore, the
mine needs bigger tailings storage facilities (TSF) which consume higher costs to main-
tain relatively to any regular TSF. Typically, 70% of these tailings materials are fine
fractions (<1 mm) produced directly from the crushing circuit [1]. These fine fractions
are counted as ore loss because any diamonds inside cannot be recovered by the dense
https://doi.org/10.1007/978-3-031-17425-4_102
912 M. A. Rasyid et al.
media separation process. Ultimately, handling fine fractions requires more water con-
sumption in the thickening and slurry pumping to the final TSF [1, 2]. The sedimenta-
tion rate is also significantly slower, and thus coagulant and flocculant reagents
consumption becomes necessary [3]. Selective liberation by producing coarser crush-
ing products can potentially minimize these issues.
Microwave pre-treatment has gained interest from the mining industry for its ability
to selectively heat minerals with different dielectric properties. It has been well estab-
lished that microwave rapid heating generates micro-cracks between mineral grains,
popularly aimed to lower specific comminution energy [4–6]. However, only a few
publications have highlighted the impacts of microwave pre-treatment on product size
distribution [7], particularly for kimberlite ore. The unique process requirement of kim-
berlite processing can take advantage of it. Ideally, the coarser crushing product with
minimum over-crushing (minimum diamond breakage) will improve the sustainability
of kimberlite processing. Therefore, the primary objective of this paper is to investigate
the application of microwave pre-treatment to produce more liberated heavy minerals
with fewer fine fractions.
The experiment was performed at Geomechanics Laboratory at McGill University.

Hypabyssal Kimberlite (HK) ore from Northwest Territories is used in this work. The
x-ray diffraction (XRD) result of the HK sample has shown a minor amount of magne-
tite, which is a decent microwave-absorbent mineral. An abundance of silicates and
carbonate class minerals as the non-absorbent matrix are also present. The mineralogy
is preferable for selective heating by microwave because it follows ideal ore properties
for microwave treatment [4, 8]. The whole bulk sample was jaw crushed and sieved to
−31.5 mm and divided into 500-g individual samples. Microwave heating was per-
formed in an industrial single-mode cavity. Microwave irradiation between 5 to 15 kW
power and 7 to 140 kWh/t energy input were applied to cover a wide range of micro-
wave pre-treatment. Each microwave condition was repeated three times. Some HK
samples were kept untreated for comparison references. All the samples were then
crushed in a laboratory single-roll crusher and were sieved using standard Tyler’s sieve
series (6.7, 4.75, 3.35, 2.76, 1.4, 1, and 0.6 mm). The particle size distribution (PSD)
data were analyzed using two well-known PSD models, the Gates-Gaudin-Schuhmann
(GGS) and the Rosin-Rammler (RR) models. The models calculated the D50 and
<1 mm fraction parameters for comparison. Selected samples were further analyzed
using the standard heavy liquid test to determine their density distribution to calculate
the ratio of coarse (>1 mm) fraction over fine (<1 mm) fraction. According to the pro-
cess flow diagram of kimberlite, the coarse fraction (C) is considered the feed of the
separation process, and the fine fraction (F) goes to the tailings.
Microwave Impacts on More Sustainable Kimberlite Processing 913
3 Results
3.1 PSD Models Accuracy
The data shows both models have very high R2 values, but GGS model is slightly has
better accuracy (non-treated samples R2 = 0.993 ± 0.001, microwave-treated samples
R2 = 0.994 ± 0.001) than RR model (non-treated samples R2 = 0.944 ± 0.006, microwave-
treated samples R2 = 0.933 ± 0.002).
3.2 D50 and <1 mm Fraction
The kimberlite is heterogenous ore and the moving average method was applied to
minimize its negative effect on the data reliability. Therefore, each point in the graph is
an average of 4–6 tests. Graph (a) of Fig. 1 illustrates the moving average D50 of crushed
kimberlite PSD. The D50 gets coarser with lower microwave energy input for all the
applied powers. The effect of microwave power intensity is less significant except at
very low microwave energy input, where the maximum improvement is reached with
9% coarser D50. In graph (b) of Fig. 1, at 10 kW, the lowest energy input produces the
maximum improvement at 12% less generation of <1 mm fraction. Higher microwave
energy input induces microcracks within the main body of mineral grains and thus pro-
duces finer D50 and higher <1 mm fractions, which are not suitable for kimberlite
processing.
Reduction of <1 mm fraction by microwave pre-treatment affects the fines removal
process, including the desliming hydrocyclone and the fines removal wet screen.
According to JKMRC vibrating screen model, the increase in the D50 size of the
crushed product will increase the size classification efficiency [9]. The coarser feed
14 14
12 5 kW 12 5 kW
10 kW 10 kW
[% improvement]
[% improvement]
10 10
< 1 mm fraction
15 kW 15 kW
8 8
D50
6 6
4 4
2 2
0 0
0 25 50 75 100 125 150 0 25 50 75 100 125 150
Microwave energy input Microwave energy input
[kWh/t] [kWh/t]
(a) (b)
Fig. 1. Moving average percentage improvement in (a) D50 and (b) <1 mm fraction plotted against
microwave energy input at three different powers.
914 M. A. Rasyid et al.
9.08
8.62
7.72
7.29
6.93
6.18
C/F ratio
Non-treated
Treated
+2.7 2.5-2.7 -2.5
density fraction (g/cc)
Fig. 2. Coarse to fine weight ratio (C/F) for different density fractions.
produced by microwave pre-treatment may provide the opportunity to increase the sol-
ids percentage of crushed product slurry. It means the process may require lesser pro-
cess water. The Plitt model for hydrocyclone agrees with the increase in solids
percentage when the ratio of the underflow rate over the overflow rate increases, assum-
ing constant hydrocyclone geometry [9].
3.3 Density Distribution
As illustrated in Fig. 2, microwave-treated crushed kimberlite has improved coarse over

fines ratio (C/F) at dense and mid-range density fractions (+2.7 g/cc and 2.5–2.7 g/cc).
It indicates better liberation of heavy and moderate dense minerals by microwave,
which also means improved diamond liberation (density = 3.5 g/cc). Lower C/F of
microwave-treated kimberlite at −2.5 g/cc fraction suggests preferential breakage on
less-dense minerals. Microwave impacts transfer a fraction of high-density particles
that ended up in −2.5 g/cc (locked or unliberated particles) to higher density fractions
as more liberated particles with higher potential for separation.
4 Conclusion
The kimberlite processing plant can be more sustainable by implementing microwave-

assisted comminution. It is based on the following experimental work outcomes:
• A higher C/F ratio at a high-density fraction, combined with overall D50 improve-
ment and <1 mm reductions, is an ideal output in sustainable kimberlite processing.
Also, it will potentially reduce the circulating load and broken diamonds by
over-crushing.
• A lesser amount of <1 mm fraction will reduce the tailings treatment cost and related
operational aspects.
Microwave Impacts on More Sustainable Kimberlite Processing 915
• According to available size classification models, an increase in the median size of

the crushed product may provide an opportunity of increasing the solids percentage
of slurry, reducing the amount of process water.
• Observing maximum improvements at low energy input and high power microwav-
ing agrees with some published research works trying to make the profitable applica-
tion of microwave technology in the mining and mineral processing industry [10,
11].
Acknowledgments This research was financially supported by the Natural Sciences and Engineering
Research Council of Canada (NSERC) and our industrial partners DeBeers company and Metso as
well as Ph.D. scholarship by the Indonesia Endowment Funds for Education (LPDP) awarded to Mr.
Muhammad A. Rasyid. All of them are gratefully acknowledged.
References
1. Rollo HA, Jamieson HE. Interaction of diamond mine waste and surface water in the Canadian
Arctic. Appl Geochem. 2006;21(9):1522–38. https://doi.org/10.1016/j.apgeochem.2006.05.008.
2. Rylatt MG, Popplewell GM. Diamond processing at Ekati in Canada. Min Eng. 1999;51(2):19–25.
3. Vietti AJ. A strategy for improving water recovery in kimberlitic diamond mines. J South Afr Inst
Min Metall. 2019;119(2):165–71. https://doi.org/10.17159/2411-9717/2019/v119n2a9.
4. Kingman SW, Vorster W, Rowson NA. The influence of mineralogy on microwave-assisted grind-
ing. Miner Eng. 2000;13(3):313–27. https://doi.org/10.1016/S0892-6875(00)00010-8.
5. Kingman SW, Jackson K, Bradshaw SM, Rowson NA, Greenwood R. An investigation into the
influence of microwave treatment on mineral ore comminution. Powder Technol. 2004;146:176–84.
https://doi.org/10.1016/j.powtec.2004.08.006.
6. Walkiewicz JW, Clark AE, McGill SL. Microwave-assisted grinding. IEEE Trans Ind Appl.
1991;27(2):239–43. https://doi.org/10.1109/28.73604.
7. Bayat O, Altıner M, Altınçelep Z, Kaymakoğlu B. Effects of microwave pre-heating on grind-
ability and liberation degree of chromite ores. XV Balk Miner Process Congr. 2013. p. 177–82.
8. Batchelor AR, Jones DA, Plint S, Kingman SW. Deriving the ideal ore texture for micro-
wave treatment of metalliferous ores. Miner Eng. 2015;84:116–29. https://doi.org/10.1016/j.
mineng.2015.10.007.
9. Napier-Munn TJ; J. K. M. R. Centre. Mineral comminution circuits: their operation and optimisa-
tion. Julius Kruttschnitt Mineral Research Centre; 1996.
10. Hassani F, Shadi A, Rafezi H, Sasmito AP, Ghoreishi-Madiseh SA. Energy analysis of the
effectiveness of microwave-assisted fragmentation. Miner Eng. 2020;159:106642. https://doi.
org/10.1016/j.mineng.2020.106642.
11. Batchelor AR, et al. Towards large scale microwave treatment of ores: part 2 – metallurgical test-
ing. Miner Eng. 2017;111:5–24. https://doi.org/10.1016/j.mineng.2017.05.003.
Decarbonization of Mining and Metals Industry.
A Critical Overview
Ashok D. Dalvi1 (*)

1
Dalvi Associates, Oakville, ON, Canada
Abstract. This is a follow-up to the discussion on sustainability challenges presented

during COM2020. Decarbonization of mining and metals industry is a major challenge of
the twenty-first century. In this paper, the targets related to the decarbonization of mining
and metals industry are quantified – specifically for steel, aluminium, copper and nickel.
Corresponding technologies are identified based on a literature review and the author’s
experience. Implementation of these technologies is discussed based on the author’s 40+
years of hands-on experience in the industry involving all stages of projects from concep-
tual design to commercial-scale implementation. Potential timelines for implementation
as well as technological, regulatory and other constraints are discussed. Metals have been
an important part of the circular economy; the end-of-life recycling rates for most com-
mon metals are greater than 50%. Recycling is an important part of decarbonization and
there is scope for increasing it.
Keywords: Decarbonization · Energy · Electricity Generation · Net-Zero · GHG

Emissions · Carbon Pricing · Mining · Steel · Aluminium · Copper · Nickel ·
Hydrogen Economy · Circular Economy · CCS/CCUS
1 Introduction
CO2 emissions from fossil fuels for energy generation and industrial use constitute 65%
of the total GHG emissions. While many ambitious targets have been declared by gov-
ernments and businesses and have been analysed in many prestigious studies such as
those by IEA and EIA, they focus on what needs to happen and what needs to be done
to achieve net-zero. The timeline for this has been adopted to be 2050. There is a lack
of analysis of what that entails, industry by industry, and how to implement the action
plan.
In this paper, we attempt to do a high-level analysis of what is required and what is
possible. Our primary focus is on the mining and metals industry, and specifically, we
will focus on mining and on metals such as iron and steel, aluminium, copper and
nickel. We will review energy production and transmission as energy underpins every
industrial and economic activity. Ironically, the usage of all these metals will be
https://doi.org/10.1007/978-3-031-17425-4_103
918 A. D. Dalvi
accelerated due to projects related to decarbonization since they are all vital to
decarbonization!
We will review the role of enabling factors, namely carbon pricing, hydrogen econ-
omy and carbon capture, utilization and storage (CCUS).
We take a pragmatic-optimistic attitude to the endeavour. This would recognise vari-
ous challenges faced but would focus on finding solutions to the challenges. We recog-
nize that the net-zero target will not be achieved in a single giant step, but will require
continuous reduction in carbon emissions. We need to be agnostic about the routes
given the urgency of the task. We recommend ALL possible options which would
reduce carbon and GHG emissions compared to today. That is: “All hands on deck!”
Timeline to Net-Zero Existing energy, infrastructure and industrial facilities have life
in the range 25–50 years and will not be discounted fully and replaced for a long time.
Premature discounting of existing facilities would be financially not possible without
serious economic consequences. Replacing the existing infrastructure and facilities
with new infrastructure and facilities needed for net-zero targets would require many
new developments requiring huge amounts of capital, new technology and skilled man-
power, as well as regulatory changes. Overhauling and rebuilding infrastructures in
various areas of economies will be required. Major developments are required in mate-
rials science and technology and extractive metallurgy and in manufacturing
engineering.
The timeline for implementation for each of the major projects involved, including
financing, engineering studies, design, construction and commissioning, is in the order
of 10 years. It will also require new legislation and government funding on a large scale,
as well as regulatory overhaul.
Given the challenges and constraints, the timeline to net-zero is closer to the year
2100 rather than 2050.
• Decarbonization of the energy sector is critical for decarbonization since energy is
fundamental. We need zero-carbon energy for electrification of the economic sec-
tors, as well as to facilitate hydrogen economy.
• Decarbonization entails hydrogen economy since hydrogen is likely required to sub-
stitute for liquid fuels, for energy storage, as a reducing agent or feed material in
metallurgical and other processes.
• Decarbonization entails circular economy, i.e. reduce, reuse (refurbish), recycle.
Metals can be endlessly recycled without loss of properties. There is an infrastruc-
ture for gathering and processing them. Metals have been historically amenable to
recycling. This is because processing recycled metal is much cheaper than process-
ing raw ores to produce virgin metal. Recycling rates for most metals are very high.
However, to get to net-zero more recycling needs to be done.
Decarbonization of Mining and Metals Industry. A Critical Overview 919
2 Energy
The world currently consumes 17.7 terawatts of energy from all sources [1]. This is
expected to increase by about 50% by 2050 [2, 3]. The challenge is to provide carbon-
free energy sources for new capacity online and to decarbonize existing capacity at the
same time!
Decarbonization of mining and metals industry requires decarbonization of the
energy sector. However, it is critical to recognize that there is a transition period to the
net-zero goal. The transition period is going to be long and needs to be well planned and
handled carefully. Oil and natural gas will play a critical role in this transition
It is important to recognize that oil and natural gas will be needed for a long time as
the renewable energy capacity builds up and the power grid is made robust and “dis-
patchable”. Currently, the transition is not handled well leading to under-investment in
fossil fuels and nuclear energy; this is leading to energy shortages and very high energy
prices. The role of nuclear energy as a carbon-free source of energy should be recog-
nized. It is important to assess risks associated with nuclear power generation realisti-
cally in light of actual history. We need to take into account advances in nuclear power
plant design in relation to safety and cost as well as technically viable waste-disposal
solutions.
A breakdown of current sources for electricity and total energy is provided in
Table 1. Figure 1 from EIA report [2] shows historical and projected energy consump-
tion by energy source in quadrillion BTU (1 quadrillion BTU is approximately 1.055
exajoules). It also provides share of primary energy and electricity by source in 2020
and 2050.
• Currently, fossil fuels account for almost 2/3rd of source for electricity and almost
85% of source for the total energy.
• Renewables will be the fastest growing source of energy between now and 2050.
Most of the new electricity generation capacity will come from renewables.
Renewable energy will be the biggest source of energy by 2050.
Table 1. Current sources for electricity and total energy.
Source Electricity Total energy

Oil 3.1 33.1
Coal 36.7 27.0
Gas 23.5 24.2
Nuclear 10.4 4.3
Hydro 15.8 6.4
Wind 5.3 2.2
Solar 2.7 1.2
Other Ren 2.5 1.6
Total 100 100
Fossil fuels 63.3 84.3
920 A. D. Dalvi
Fig. 1. Primary Energy Consumption and Share by Source (www.eia.goc/ieo).
t CO2/GWh
1500
1000
500
0
Lignite Coal Oil N.G. Solar PVBiomass Nuclear Hydro Wind
Fig. 2. Lifecycle Greenhouse gas emissions, t CO2/GWh.
• Thus, EIA projects that the use of fossil fuels as energy source will continue to grow
to satisfy growing energy demand. Electricity generation from fossil fuels will
increase from 15 to 19 tera kWh but their share of total electricity generation will fall
from 61% to 43%.
• Projected global electricity demand will almost double from about 7300 to 14,000
GW.
• The share of coal as energy source will diminish but the shares of natural gas and oil
will decrease only slightly.
• Installed nuclear power generation capacity will increase only marginally due to
current regulatory environment, public perception and the slow progress of the tech-
nology development for the next generation power plants.
Figure 2 shows Life Cycle greenhouse gas emissions for various energy sources for
electricity generation. The emissions for fossil fuels are in the range 500–1000 t CO2-e/
GWh. Emissions for natural gas are about half of those for lignite. Emissions for solar
PV, biomass, nuclear, hydro and wind power are less than 100 CO2-e/GWh. The capac-
ity factors for various sources are as follows:
Nuclear: 94%; Natural Gas: 57%; Coal: 48%; Hydro: 39%; Wind: 35%; Solar: 25%
Of the various sources, nuclear, coal and hydro are well suited for continuous power
generation and therefore act as base load. Nuclear power generation has the highest
capacity factor of >90%. Power generation with natural gas is very flexible since the
gas turbines can be switched on and off and can adjust to loads quickly. Hence gas
turbines are used as Peaker Plants. Wind and solar power generation is intermittent with
capacity factor less than 35%. They need augmentation by the other sources to provide
“dispatchable power” (that is power on demand).
Global Power Generation Capacity and Cost The global power generation capacity
in 2020 was 7300 GW and is expected to nearly double to 14,000 GW by 2050 from
data provided by Statistica [4].
The scenario presented in Table 2 is constructed based on various sources, but gener-
ally consistent with the projections of Statistica and EIA. Many of the sources provide
unrealistic growth rates for Wind, Solar and Carbon Capture and Storage (CCS). The
growth of Solar and Wind assumed is still substantial and is 4 times the capacity in 2020
consistent with a pragmatic-optimistic approach. The growth in Natural gas, coal and
nuclear sources is assumed to be moderate and consistent with EIA data. It is assumed
that CCS will be applied to incremental capacity in natural gas and coal power genera-
tion in moderate amounts. CCS is not yet fully proven. It is dependent on high carbon
price (>$50/t CO2) and local acceptability. CCS has been applied for Enhanced Oil
Recovery (EOR) where there is ready place for sequestration that has payback in terms
of oil recovery. However, it has proved uneconomical elsewhere.
Gonsales et al. [5] have provided a general review of CCS. Rubin et al. [6] have
estimated incremental capital cost for CCS applied to power plant as ~$2000/MW or
75% increase in the capital cost for power generation plant and 62% increase in
Levelized Cost of Electricity (LCOE). Also, CCS decreases efficiency of the power
plant BY ~10% due to energy required for carbon capture and transport. The cost of
Table 2 Current and Projected Electricity Generation Capacity and Cost.

Projected Capacity
Source Installed Capacity, GW GW Net Increase GW Unit cost Total Cost
2020 2050 $/kW $billion
Wind 743 2700 1957 2447 4789
Solar 773 3350 2577 1248 3216
Nat Gas 1839 2400 561 2481 1392
with CCSa
Coal with 2201 2300 99 5084 503
CCSa
Adv. 415 650 235 6000 1410
Nuclear
Hydro 1330 2600 1270 2769 3517
Total 7301 14000 6699 14827
a
CCS assumed only for the incremental capacity
922 A. D. Dalvi
Table 3. Power Generation and CO2 Emissions.

Year 2020 2050
Power Generation Capacity GW 7300 14,000
Electricity Generation TWh 30,187 51,052
CO2 Emission Gt/year 13.09 13.72
CO2 Emission t/GWh 434 268
CO2 avoided was $63/t. Current installed capacity (mostly for EOR) is 40 Mt/year. This
is 0.1% of the global CO2 emission rate.
As seen from Table 2, the total estimated capital cost for incremental capacity is
close to US$15 trillion in today’s dollars! This is a huge amount of capital to be invested
in a very short time ($500 billion per year for 30 years). This capital doubles the total
power generating capacity and increases the share of carbon-free power generation
from 45% to 63%. Table 3 provides a summary of power generation and CO2
emission.
While the power generation capacity doubles from 2020 to 2050, the actual amount
of electricity generated increases by 69% (Table 2). This is because the various capacity
factors are less than 1 and the proportion of different sources has changed. The increase
in CO2 generation is about 5% and the CO2 generation intensity decreases by 40%.
Thus, CO2 emission avoided is 9 Gt/year.
In this scenario, we have assumed:
• Current technology and costs for the various power generation sources
• Significant increase in wind and solar power generation
• Modest increase in hydro and nuclear power generation capacity
• CCS only applied to incremental natural gas and coal generation capacity. This
requires 170 Mt CO2/year CCS capacity, compared to current global capacity of 40
Mt/year.
• A doubling of power generation capacity to meet demand.
However, even after employing $15 trillion capital there is net INCREASE in CO2
generation. We will not be close to net-zero by 2050!
3 Decarbonization of Mining
The mining industry contributes 2–3% of global CO2 emissions. Additional GHG emis-
sion is attributed to methane evolution from coal mines. The top 10 mining companies
account for >210 Mt CO2 in 2017, according to Statistica [7]. Iron ore and copper mines
each contribute 74 Mt CO2 emission/year, while gold mines contribute about 22 Mt
CO2/year.
Coal mine methane emissions contribute the largest share of emissions. These per-
sist even after the mines are closed. Kholod et al. [8] have analysed global coal mine
methane emissions and have estimated that they will increase by 85% from 2020 to
2050 unless mitigation is in place. Global Methane Initiative (GMI) estimated global
methane emissions in 2010 as 584 Mt CO2-e. [8]. This is huge since 1 t of methane is
equivalent to 84 t CO2. GMI has discussed various means of capture and utilization of
methane which have a payback. Environmental Defence Fund (EDF) using McKinsey’s
MAC Curve has estimated a payback of ~$10/t CO2 [9, 10]
The majority of the (non-methane) emissions arise from diesel combustion in mobile
and other equipment. Most of the rest is due to electricity usage. For example, McKinsey
provides an example of an Australian Open Pit Mine where 55% of the emission is due
to diesel, 37% is due to electricity, and 7–8% due to other sources.
When compared to other industries, the mining sector has been reluctant to adopt
electrification due to its historical reliance on diesel for the operation and powering of
heavy machinery. Mining has a clear pathway to decarbonization as delineated by
McKinsey, Warren Centre [11, 12] and others. Emissions within mining can be broken
down into three broad types: Emissions from diesel: 40% to 50% of CO2 emissions
come from diesel used in mobile equipment; Emissions from electricity generation:
30% to 35% from non-renewable electricity; and Emissions from the supply chain and
transport. The emissions intensity varies widely across mines. Copper has 20 times
higher emission intensity per tonne of ore, compared to iron ore. However, average cop-
per ore grade is below 1% while most iron ores contain iron in the range 50–60%. Thus,
the run-of-mine ore is much higher for copper for the same amount of contained metal.
In a report by Warren Centre [12], an extensive list of possible technologies and
options has been presented. Pathway to decarbonization include:
• Short term:
–– Improve operational efficiency
–– Replace diesel with sustainable fuels such as biodiesel
• Mid-term: Introduce hybrid diesel-electric as a transitional step before introducing
BEV (Battery Electric Vehicles) or FCEV (Fuel Cell Electric Vehicles). The hybrid
diesel-electric trucks could be augmented by trolly assist.
• Sustainable sourcing
• Longer term:
–– Fully electric mine
–– BEV or FCEV equipment. BEV may be preferred for underground mining as
well as for smaller equipment. FCEV may be more suited for large equipment.
Both of these technologies are being developed.
–– Carbon-free electricity either with local microgrid using solar PV or wind gen-
eration. Falconbridge (now Glencore) have reduced carbon intensity of their
Raglan Nickel mine in Quebec by installing wind turbines in combination with
energy storage and management [13].
–– For large, long-life mines coupling the mine with SMR is a possibility.
These will require significant mine-level investments—typically $100 million to
$130 million for a 25 million metric ton run-of-mine (ROM) facility.
An all-electric mine using on-site renewable power with battery storage and a
microgrid has potentially very low operating costs and low sensitivity to energy prices.
All-electric mines are more amenable to digitization and automation leading to higher
productivity. All these advantages can provide cost savings in the order of 5–10% and
potential savings in infrastructure amounting to tens of millions of dollars, and thus a
924 A. D. Dalvi
return on the additional investment required [14, 15]. This is in addition to reduction in
CO2 emission progressing towards a net-zero mine.
Compared to other sectors, the technology required for this is well advanced. OEM’s
such as Sandvik, ABB, Siemens, Volvo, JCB, as well as large mine operators in collabo-
ration with local industries are developing various components required for an all-
electric mine. ABB has an e-Mine solution that provides integrated solution to all aspect
of mining, including mobile equipment, power supply, crushing and grinding equip-
ment, power generation, micro-grid and automation.
Currently, mine electrification is only 1% of the total, but it is rapidly advancing.
Newmont has built the world’s first all-electric underground gold mine near Borden
Lake in Northern Ontario, which started operation in November 2019. The Matawinie
graphite project being developed by Nouveau Monde in the Québec Province of Canada
is expected to be the world’s first all-electric open-pit mining operation.
4 Decarbonization of Steel Production
Global steel production was 1.878 billion tonnes in 2021 [16] and is growing at a rate
of 3% per annum and is expected to grow by 33% by 2050 to ~2.5 billion tonnes. Each
tonne of steel produced currently emits 1.85 tonnes of CO2 [17]. Production of steel is
responsible for about 8% of global GHG emissions (~4.3 Gt/year) and 20% of coal
consumption. In 2020 China produced 56.7% of the world’s crude steel with 91% man-
ufactured via the BF-BOF process, leaving just 9% produced via EAF [16, 18].
In 2019, 70% of the steel was produced by the conventional BF-BOF process; 7% of
the steel was produced by DRI-EF and most of the remaining (22%) was produced by
scrap-based EF [19].
Currently, 70% of steel comes from ore and the remaining 30% comes from recycled
scrap. China is an important player in decarbonization of steel due to its dominance in
the industry and its high carbon intensity.
CO2 intensity of the BF-BOF process varies between a low of 1500 kgCO2/t steel to
a high of 2350–2750 kg CO2/t in China and India. The energy intensity varies from a
low of 10 GJ/t (Italy and Spain) to a high of 18–19 GJ/t in India and China. The carbon
intensity of the EAF process varies from a low of 350–400 kg CO2/t in Canada and
France and a high of 1400–1600 kg CO2/t in China and India [20]. This is related to the
high use of scrap steel in the developed economies and available low-C power (nuclear,
hydro and renewables). These values already indicate a pathway to decarbonization,
namely:
• Making incremental improvements using best practices
• Switching to scrap melting in EAF
• Switching to DRI-EAF process
• Switching to low-C electricity generation.
The lowest value is for EAF scrap-based production in Canada which is 350 kg
CO2/t steel. Ontario, where major steel producers in Canada are located, has 95%
carbon-free power generation capacity. The low value of CO2 emission represents
>80% reduction in CO2 emission. In 2020, 71% of total crude steel production in the
U.S. was via EAFs. In Canada and the EU, the figure was 42%. This strongly hints at
the potential for more scrap use in economies such as China, Japan and South Korea
where steel recycling rates are much lower. It is to be understood that there is not
enough steel scrap to satisfy global steel demand. Scrap generation in developing econ-
omies is low since the infrastructure is new and is not at the end of life. As it becomes
older, more scrap is expected to come to the market. Global scrap availability is expected
to reach about 1 billion tonnes in 2030 and 1.3 billion tonnes in 2050. In other words,
we expect to see a growth of more than 500 Mt in one of the main steelmaking materials
in the next 30 years [21, 22].
Currently, there is excess production capacity in the steel industry. (2.452 billion
tonnes [23] vs production of 1.877 billion tonnes in 2019 (i. e. about 30% excess capac-
ity)). Therefore, the transition to net-zero could lead to a large portion of stranded
assets.
EPA studied available and emerging technologies for reducing GHG emissions from
the iron and steel industry. Their report [24] provides an extensive list of options for
incremental efficiency gains as well as a review of emerging technologies. Liliana
Guevara Opinska et al. did a study for the EU parliament which evaluated various tech-
nology routes and their relative readiness and costs [25]. Zhiyuan Fan and Friedmann
did a study for Centre on Global Energy Policy assessing options for low-C steel pro-
duction [26].
The pathway to decarbonization of the steel industry is expected to include follow-
ing options:
• Increasing the efficiency of current production methods- especially the BF-BOF
process. Hydrogen injection in BF, as well as OBF.
• Maximize scrap usage and thus maximize recycling of steel
• Switch to EAF smelting at an earliest opportunity
• Optimize DRI-EAF process by switching to natural gas in the early period.
Conventional DRI-EF technology produces 50–70% less CO2 per tonne of steel than
BOF process
• Gradually switch from natural gas to hydrogen in DRI production
• Switch to low-C or zero-C electrical power including renewable, hydro and nuclear
(the Ontario and French model).
• Use CCUS where feasible and economical.
Sara Hornby [27] has carried out a technical and cost analysis of various decarbon-
ization options including DRI with hydrogen. She estimates a capex increase of 25%
compared to BF-BOF and opex increase of ~65% for H2 DRI. There are no serious
issues related to H2 DRI production. It is possible to start with Natural Gas DRI and
gradually increase the proportion of hydrogen (as has been proposed by ArcelorMittal
and others). However, there are serious technical issues related to EAF steel operation
with carbon-free DRI. Also, infrastructure for the production of H2 and its delivery is
not in place. Green Steel for Europe [28] has carried out investment needs for green
steel limited to Europe and have estimated an investment in excess of 25 trillion Euros!
This does not include the investment required for carbon-free electricity generation to
produce green hydrogen.
926 A. D. Dalvi
Global steel market is very competitive with thin margins. Therefore, the ability of
producers to provide large investments in new facilities with novel (and therefore risky)
technologies and absorb losses due to stranded assets is limited. BOF plants are 13 years
old on average, with an expected total life of 40 years. It is therefore imperative to mini-
mize stranded assets and introduce new technologies gradually.
It is possible to do stepwise reduction in CO2 emission by a combination of strate-
gies while minimizing capital intensity and minimizing stranding existing assets. This
would also allow for adopting new technologies and develop necessary carbon-free
power generating capacity and green hydrogen production capacity.
A 2020 report by OECD provides the latest steelmaking capacity developments
[23]. Vogl, Olsson and Nykvist [29] have done an excellent analysis of transition strat-
egy for steel works, based on identifying assets which need not be replaced and sched-
uling asset replacement based on Blast Furnace reline schedule. BF has the longest life
(15–20 years) between relines, and replacement of old assets with newer assets with
low carbon footprint is best scheduled at the time of the next reline to minimize wasted
capital and stranded assets. A detailed techno-economic assessment by the IEAGHG
has found the value of retainable assets (1817 million USD) to be close to half of total
investment costs for a greenfield integrated steel mill.
There have been two waves of capacity addition: a global expansion period in the
1970s and the great Chinese acceleration between 2000 and 2018. This defines the
committed emission from existing blast furnaces between now and 2037. Replacement
of old assets has to be timed with the retirement of the existing assets. If this timing is
missed, then the next period will be 15–20 years later!
The average direct CO2 emission intensity of steel production must decline by 60%
by 2050, to 0.6 tonnes of CO2 per tonne of crude steel (t CO2/t), relative to today’s levels
(1.85 t CO2/t). The energy intensity of state-of-the-art blast furnaces is already approach-
ing the practical minimum energy requirement.
Improvements in operational efficiency, including enhanced process control and pre-
dictive maintenance strategies, together with the implementation of best available tech-
nologies would contribute around 20% of cumulative emissions savings in the
Sustainable Development Scenario.
In the Sustainable Development Scenario considered by The International Energy
Agency (IEA) [30], the deployment of one hydrogen-based DRI plant per month is
required globally following market introduction of the technology. This raises electric-
ity demand by 720 terawatt hours by 2050, equivalent to 60% of the sector’s total elec-
tricity consumption today (or, one 800 MW nuclear power plant per 3Mt/year integrated
steel plant). The concurrent deployment of CCUS-equipped plants requires around 0.4
Gt CO2 capture globally in 2050, equivalent to the deployment of a large CCUS instal-
lation (1 million tonnes CO2 capture per year) every 2–3 weeks from 2030!
IEA considers two possible scenarios for 2050 [31]. In both scenarios, scrap-based
EF production increases from 22% to 36–38%. However, the Sustainable scenario IEA
projects 8% new DRI-EF capacity with hydrogen and 5% capacity with CCS, thus
substantially decreasing CO2 emission per tonne of molten metal. IEA estimates the
increase in production cost to be between 10% and 50% compared with today. However,
the CCS infrastructure is currently not developed and the estimated costs are high.
Several producers are switching to DRI-EAF technology from BF-BOF

technology:
• Arcelor Mittal have announced a decarbonization project at its Hamilton Ontario
facility consisting of a 2.5 million tonne capacity DRI facility and an EAF facility
capable of producing 2.4 Mt steel at a cost of C$1.8 billion. The investment will
reduce annual CO2 emissions by approximately 3 million tonnes, representing
approximately 60% of emissions [32].
• Arcelor Mittal Europe also has announced CO2 reduction plan involving H2 DRI-
EAF and Smart Carbon technologies [33].
• Algoma Steel of Canada have announced construction of two new electric-arc-
furnaces (EAF) to replace existing BF-BOF facility. The transformation is expected
to reduce Algoma’s carbon emissions by approximately 70% [34].
• Thyssenkrupp has launched decarbonization project involving H2 injection in BF, H2
DRI and Smart Carbon technologies [35].
• Scandinavian companies SSAB, LKAB and Vattenfall have developed a carbon-free
process called Hybrit process and have constructed and operated a pilot plant to
produce carbon-free steel [36]
Table 4 illustrates a moderate scenario for decarbonization of steel that incorporates
most of the options discussed above, some of which are more optimistic than that pro-
posed by IEA. In this scenario, the global steel production capacity increases from
current 2452 to 2670 Mt/year. Currently, there is an excess production capacity of 574
Mt/year. This is assumed to decrease by 404 Mt/year – all for the BF-BOF process.
The steel production (as opposed to capacity) increases from a1878 Mt/year to 2500
Mt/year (an increase of 33%. Following other assumptions are made:
• Scrap usage increases from 413 Mt/year to 1000 Mt/year
• DRI-based production increases from 131 Mt/year to 350 Mt/year, of which 50 Mt/
year is natural gas based and the remaining 300 Mt/year is H2 based. The H2 based
production requires commercial development, as only a small production facility at
a pilot scale is currently operating (Arcelor Mittal). Currently, only 1% of the hydro-
gen produced is green hydrogen; the remaining 99% is grey hydrogen which results
in CO2 emissions.
• The emission intensities are assumed as follows: BF-BOF: 1.782 t CO2/t steel, i.e.
20% reduction in CO2 emission per tonne of steel by implementing various tech-
nologies; Scrap steel EAF melting: 0.30 t CO2/t steel; DRI with natural gas: 0.9 t
CO2/t steel; DRI-EAF (green hydrogen, green electricity):0;
• Investments required: For BF-BOF improvements: 20% of greenfield plant; For new
scrap melting facility: US$ 450/t of rolled steel; For a greenfield DRI-EAF facility
(with natural gas); US$ 690/t product; For DRI-EAF facility (with H2): 20% above
a facility with natural gas (but excluding H2 generation and related power generation
facilities). These values are used to provide an order-of-magnitude comparison and
need to be fine-tuned.
928 A. D. Dalvi
Table 4. Decarbonization of Steel Production: Moderate Scenarioa.

2020 2050
C C
Production Intensity Emission Production Intensity Emission Investment Capex
$/t
Mt/year Mt/year Mt/year Mt/year $ Billion Steel
BF-BOF 1314 2.227 2926 0 0
BF- BOF 1100 1.782 1959.76 99.0 90
with
improvements
Scrap EF 413 0.65 268 1000 0.30 300 264 450
DRI - 131 2.00 262 0 0 0
Conventional
DRI NG 0 50 0.90 45 34.5 690
DRI Green 0 300 0 0 300 1000
H2
Other 20 0.75 15 50 0.60 30 30 1000
Total 1878 1.85 3472 2500 0.934 2335 728
Capacity 2362 2670 4625(1)
Overcapacity, 26 6.8
%
CO2 1137
Reduction,
Mt
CO2 2290
avoidance, Mt
No CCUS is assumed; assumed used for power generation, Natural gas sweetening and cement produc-
tion
Investment excludes hydrogen generation and related power generation capacity
(1) Projected emission at current carbon intensity
a
Assumed 214 Mt of BF-BOF capacity is decommissioned
The final results are as follows:

• CO2 emission decreases by 1137 Mt/year (or 1/3rd) while increasing production by
1/3rd. This is equivalent of CO2 avoidance of 2287 Mt/year.
• CO2 intensity reduces almost by half.
• Total capital needed (over and above ongoing capital) is US$ 728 billion (excluding
capex needed for H2 generation and new zero-C electricity generation capacity).
This scenario requires a large investment of capital as well as the development of
new technologies and execution of numerous large construction projects requiring huge
resources and skilled manpower. Most of this investment is backend loaded later than
2030. This is due to BF relining schedule and requirement of zero-carbon electricity
and time required for installation of green and blue H2 generation capacity. According
to ReThink Energy [37] total hydrogen requirement for decarbonizing steel is 60 Mt by
2050. It would require 4000TWh of energy (~1.3TW of installed power capacity, com-
pared to 7.3TW current installed capacity)!
Yet, while it reduces the CO2 emission intensity of the steel industry by 50% by
2050, it does not reach net-zero. To get to net zero much larger capital outlay is required
as well as an expansion of hydrogen generation capacity and zero-carbon power gen-
eration capacity. It also possibly entails CCUS for sequestering residual emissions.
5 Decarbonization of Aluminium Production
Global aluminium production in 2021 is expected to be ~100 Mt. Of this 67.3 Mt would
be primary production and the remaining (~33%) would be secondary production from
scrap melting [38, 39]. China is responsible for 60% of the global production. The
global aluminium production has grown by 52% from 2010 to 2018, much of it in
China with energy supplied by coal-fired power plants. This has resulted in significant
growth in CO2 emissions by the aluminium industry. Therefore, China is an important
player in decarbonization of the aluminium industry. Aluminium is a strategic metal for
electrification and decarbonization and its production is expected to grow by 80% by
2050 spurred by demand for solar PV panels. However, the share of the scrap is expected
to grow to ~50% by 2050 from 33% today (i.e. ~90 Mt) [40]. Without significant reduc-
tion in the carbon intensity, the CO2 emissions from the aluminium industry will grow
by 50% by 2050.
Aluminium industry produces 1.16 Gt CO2-e/year, representing 2% of total global
GHG emissions, second only to steel among metals [41].
The average carbon intensity for aluminium production is 11.6 t CO2/t Al [41, 42]
(adjusted per Table 3). There is a large variation in the carbon footprint between various
operations. Power supply for aluminium smelting in North America and Russia is pre-
dominantly hydro with very small carbon footprint. While China and the rest of Asia
the footprint is very large due to use of coal in power generation. Natural gas is used for
power generation in the Middle East and therefore it has an intermediate carbon
footprint.
CO2 emission in primary aluminium production breaks down as follows:
• Energy*: 64%; Process related: 29%; Transportation & Ancillary raw materials: 7%
*CO2 emissions/kWh adjusted from [41] based on Table 3.
Of the process-related emissions, 14% are from CO2 emission from carbon anode
and 15% is from thermal energy required for the process.
The carbon intensity of recycled aluminium is 1/20th of primary production inten-
sity at 0.5–0.6 t CO2/t aluminium since processing requires only the melting energy.
Canada has the lowest carbon footprint for aluminium production at 2 t CO2-e/t alu-
minium due to use of hydro power and implementation of best practices. Rusal in
Russia has a carbon footprint of <4 t CO2-e/t aluminium, also due to hydro power gen-
eration. While China and the rest of Asia has the highest at 17.3 t CO2-e/t of Aluminium.
Thus, the scope for carbon intensity reduction is 80% on a global basis and 90% for the
930 A. D. Dalvi
biggest emitters, based on the best practice. The aluminium industry has made great
strides in reducing PFC’s. Further reduction is needed.
This defines the pathway to decarbonization:
• Decarbonization of the power supply will decrease the carbon intensity by 64%.
• An alternative to carbon electrodes in primary production will decrease emissions by
14%. This will also reduce PFC emission
• Decarbonization of thermal energy in primary production will decrease emissions
by further 15%.
• Increasing the proportion of recycled scrap from the current ratio of 33% to ~50%
targeted for 2050 will substantially reduce the average carbon intensity.
Thus, there is scope to reduce carbon intensity of aluminium below 2 t CO2/t alu-
minium. Various pathways for reducing the carbon footprint of the aluminium industry
have been extensively discussed [41–45]
Decarbonization of the power supply could be achieved by replacing thermal gen-
eration with zero-carbon generation including hydro, renewables and nuclear. Nuclear
power plants are well suited for aluminium smelters since the power requirement for
smelters are in the range of 1000 MW. As discussed under the Energy section, there are
several options available using 4th generation technology for nuclear power plant.
Another option is to develop thorium-based reactor which was experimentally tested at
a pilot plant scale at the Argonne National Laboratory in the USA in late 1960’s. China
is currently about to start operation of a pilot thorium-based reactor. India is also active
in the development of thorium reactor which is in an early stage.
The additional primary aluminium capacity needed between now and 2050 is about
22 Mt/year. At a capital cost US$6000/t Al annual capacity, the extra capacity would
require $132 billion new investment [46, 47]. Al production is highly energy intensive
requiring 17,000 kWh/t. Power requirement is about 1.4 kW/t. Thus, a plant producing
1.5 Mt aluminium per year would require power supply is about 2000 MW. Additional
power requirement for 22 Mt/year capacity would be ~30 GW. A blended capital cost
of $2200/kW is estimated for zero-carbon power generation (see Table 2). This means
the power generation alone would require $65 billion. These costs do not include the
capital required for the related infrastructure for transportation etc. Lead times for large
metallurgical facilities is in the range 5–8 years. Aluminium is currently in short supply.
As the demand is increasing rapidly, decisions for expanded capacity as well as the
power supply and infrastructure need to be taken soon. The new plants have to incorpo-
rate the latest technologies available for reducing CO2 emissions and energy require-
ments as discussed below.
Given the state of development of CCS and H2 infrastructure, as well costs, applica-
tion of these for aluminium related energy generation is considered remote in the near
future.
Inert Anodes Replacing carbon anodes with inert anodes has the potential to reduce
CO2 emissions. Then the anode gas would be oxygen and there would not be any CO2,
CO or perfluorocarbon (PFC) emissions at all from a carbon-free electrolytic process.
This means that we can eliminate the emissions from the electrolysis step and also the
perfluorocarbon emissions. This would result in a reduction of between 1.4 and 1.7 t
CO2-e/t Al.
There are three entities which have been working on this [47, 48]:
• Elysis consortium: ELYSIS is a joint venture between Alcoa, Rio Tinto and Apple to
create inert anodes to produce low-carbon aluminium. ELYSIS has begun construc-
tion of its first commercial-scale prototype at one of Rio Tinto’s smelters in Canada.
The companies hope to demonstrate the viability of the technology by 2024 and
commercialize it for the rest of the aluminium sector.
• Rusal: RUSAL has produced more than 1500 tonnes of the lowest-carbon alumin-
ium ever (0.01 tonnes CO2-e/tonne aluminium in direct emissions) using inert anode
technology and renewable electricity sources.
• Arctus Aluminium R & D team, in cooperation with the Innovation Centre Iceland,
has produced high-purity (99.9%) low-carbon aluminium in a test cell with vertical
inert anodes and wetted cathodes in a low temperature (800 °C) electrolyte.
Using zero-C electrical power to supply thermal energy would reduce CO2 emission
further. Alternatively, green hydrogen or blue hydrogen could be used for thermal
energy. Thermal energy could also be supplied by a nuclear power plant associated with
the smelting operation, if nuclear power option is adopted. For example, Rusal is plan-
ning to build an aluminium refinery in Saratov region complete with two nuclear reac-
tors [49] Planning, permitting and construction of a nuclear power plant could take
10 years in free-market economies.
The capital cost of replacing existing cells with inert anodes is not known. However,
if we assume it to be half of the greenfield capex for the refinery, that is $2000/t annual
installed capacity, then the replacement cost for the existing 67.5 Mt/year capacity
would be approximately $67.5 billion.
The total cost to the industry is thus ~$265 billion over a period of 30 years. This will
be mostly back-end loaded due to the time required for installing zero-carbon power
generation, as well as time required for adopting new technologies. If the capital is not
deployed soon enough, aluminium prices could skyrocket in the near future.
Scrap Recycling The World consumes four billion aluminium cans per week, most of
which are recycled. Brazil recycles 98.2% of its aluminium can production. The overall
recycling rate for aluminium beverage cans in the European Union, Switzerland,
Norway and Iceland in 2017 rose 2.3% from 2016 (72.8%), to reach an all-time record
74.5% in 2017. Japan’s recovery rate is 82.5%.
Currently, scrap constitutes 33% of aluminium supply. It is anticipated that this
could be increased to 50%. The total scrap consumption is thus anticipated to increase
from the current 33 Mt to 90 Mt/year. Scope for increasing scrap consumption exists in
China and other Asia where the recycling rate is low. However, this requires significant
improvements in scrap collection and recycling, which requires improved and expanded
infrastructure.
932 A. D. Dalvi
6 Decarbonization of Copper and Nickel Production
Copper and Nickel are both industrial base metals with a critical role to play in decar-
bonizing due to their use in electrification. As such, their demand is expected to rise,
and with that the CO2 emissions associated with their production, unless efforts are
made to decarbonize production
6.1 Copper
The world mine production of copper in 2020 was 20.6 million tonnes [50] and 21 Mt
in 2021 [51] Chile (5.7 Mt) and Peru (2.2 Mt) are the two largest producers of copper,
followed by China (1.7 Mt), DR Congo (1.3 Mt) and the USA (1.2Mt) [51]. A total of
about 8.5 Mt is produced from recycled material. Total production was 29.1 Mt. The
proportion of the recycled copper is thus ~30%. World copper production has been
increasing at about 2.75% a year for over a hundred years. Glencore expects demand for
copper to double to 60 Mt by 2050 [52]. Ayman Elshkaki et al. [53] project copper
demand to rise by 275 to 350% by 2050. The copper production capacity needs to
increase by 1 Mt per year to keep up with demand, compared to 0.5 Mt per year during
the last decade, even at the lower end of the projections.
With the increasing production, the copper grades have been falling, necessitating
the processing of higher quantities of ore. Nilsson et al. [54] have made an assessment
of carbon intensity of copper. Copper ores come in two types: Sulphide ores, and Oxide
ores. They are processed differently. The sulphide ores are upgraded to high-grade con-
centrate by grinding and flotation, which is then smelted in flash furnaces which use the
concentrate itself as a fuel. The resulting matte is then processed further and refined by
electro-refining. The oxide ores are generally treated by leach-SX-EW (Solvent
Extraction - Electrowinning) process.
The carbon intensity of copper varies based on ore grade and ore chemistry, source
of power, as well as the processing route. Copper ore grades have been consistently
falling as copper production has increased. The average ore grade is currently 0.6%
(compared to 40–60% for iron), requiring large amount of materials handling during
mining and milling. The falling grade has resulted in an increased carbon footprint.
Nilsson et al. [54] show a low carbon footprint between 1 and 2 t CO2/t Cu to a high
of 8.5 t CO2/t Cu for primary production. For pyrometallurgy route the carbon intensity
varies between 1 and 5 t CO2/t copper, with one operation at 8.5 t CO2/t copper. For the
hydrometallurgy route the emissions are in the range 1–3.5 t CO2/t copper. The global
average is close to 6.5 t CO2/t Cu for primary copper production. Copper scrap for sec-
ondary copper production comes from a wide variety of sources with different grades,
requiring different processing. The carbon intensity is therefore variable between 0.1
(refinery scrap), 0.5 (high grade scrap) and up to 2 (low grade scrap) t CO2/t Cu. The
total CO2 emission attributed to copper would be ~150 Mt/year. Doubling of production
by 2050 would entail increasing this to ~250 Mt CO2/year by 2050, unless the carbon
intensity is reduced.
The main contributor to the environmental impact is the production and transporta-
tion of copper concentrate that accounted for around 80% of the total energy consump-
tion and CO2 emissions. Therefore, electrification and decarbonization of copper

mining and concentrate transport become critical to achieve decarbonization of copper
production. The smelting technologies (in the pyrometallurgical route) only accounted
for around 20% of the total impact. Use of zero-carbon electricity in electrorefining
(5%) and slag cleaning (1%) would contribute to lowering the carbon footprint. The
carbon intensity of smelting is in the range 0.5–0.8 t CO2/t blister copper. There is scope
to reduce the carbon footprint of the smelting technologies by the choice of technology
as well as switching to carbon-free power and H2 fuel. Increasing the proportion of cop-
per scrap would aid reduction in carbon intensity.
Increased material efficiency is critical in the case of copper, not only from the point
of view of carbon footprint, but also to meet the growing demand for copper in the com-
ing decades.
6.2 Nickel
The world production of nickel in 2021 is expected to be 2.67 million tonnes rising to
3.12 Mt in 2022 [55, 56]. In 2019, the world’s leading nickel producing countries were,
Indonesia, Philippines, Russia, New Caledonia, Australia, Canada, and China. Around
68% of all nickel available from consumer products is recycled (reference year 2010);
another 15% enters the carbon steel loop. However, around 17% still ends up in landfill,
mainly in metal goods and in waste electrical and electronic equipment. It is estimated
that nickel bearing scrap totalling 4.4–4.6 million tonnes per year is collected and recy-
cled, containing almost 350,000 t of nickel annually which is mainly used by the
stainless-steel industry.
Demand for nickel is expected to increase from 2.67 Mt in 2021 to somewhere in the
range of 3.5 million to 4.0 million tonnes by 2030 and to >4 million tonnes by 2040.
IEA estimates the demand could increase further by >50% if deeper decarbonization
occurs (for example due to increased production of battery electric vehicles (BEV) and
batteries for power storage) [57].
Nickel occurs in the form of sulphide as well as oxide (laterite). Identified world’s
nickel resources consist of 60% laterite ores and 40% sulfide ores. The average nickel
grade for laterite and sulphide are similar (~1.8% Ni). But the sulphides could be con-
centrated by grinding and flotation by a factor of 5–10. The laterite ores have very
limited upgradability. The carbon footprint of nickel is variable based on the nickel
grade, type of ore (sulphide or laterite), and processing route (smelting or leaching) as
well as individual plant practices [57, 58].
Both pyrometallurgical (smelting) and hydrometallurgical (leaching) routes are fol-
lowed in the extraction of nickel. In the smelting of sulphides, the concentrate is smelted
in either flash furnaces or electrical furnaces to produce a matte which is further pro-
cessed in Pierce-Smith converters to produce a high-grade Ni-Cu-Co matte which is
refined by various methods to produce pure metals. In the hydrometallurgical route the
concentrate is leached in autoclaves and processed in solution to produce pure metal.
Laterites are processed in RK-EF (Rotary Kiln- Electric Furnace) route where the
ore is dried in rotary dryers and heated and calcined in rotary kilns; the kiln product is
smelted in electric furnaces to produce either ferronickel or a matte which are further
934 A. D. Dalvi
processed. Alternatively, low-nickel limonitic ores are smelted in blast furnaces to pro-
duce low-grade pig iron which is directly used in production of stainless steel.
The average intensity is 40 t CO2/t Ni [58, from SKARN Associates]. For the most
efficient sulphide operations the intensity is in the range 2–6 t CO2/t Ni, while for high
energy consuming laterite operations the carbon intensity is in the range 75–175 t CO2/t
Ni, giving a large range. The total CO2 emission attributed to nickel would be ~104 M t/
year [59]. One hundred and fifty percent increase in production by 2050 would entail
increasing this to ~250 Mt CO2/year unless the carbon intensity is reduced.
Decarbonization Pathway for Nickel Fukuzawa [58] and McKinsey [60] have
reviewed nickel industry emissions and have suggested pathways for decarburization.
The following options exist for the decarbonization of nickel production:
• Incremental process improvements: Dalvi et al. [61] have described process improve-
ments at P.T. Inco (now P.T. Vale) in Indonesia to reduce energy intensity by >25%.
Similarly, Inco (now Vale) have made a substantial reduction in energy intensity
through process and operational improvements at their Sudbury facility.
–– There is a wide variation in carbon intensity for the laterite ores. Best practices
would substantially reduce the carbon intensity.
• Decarbonization of electrical energy source
–– This would reduce carbon intensity in mining as well as the processing and is
applicable across all the process routes followed.
–– Energy sources in nickel processing vary from hydro, renewables, as well as fos-
sil fuels. In pyrometallurgical operations, coal and oil are used for both heating
and reduction.
–– Falconbridge (now Glencore) have reduced the carbon intensity of their Raglan
mine in Quebec by installing wind turbines in combination with energy storage
and management [62].
• Substitution of natural gas for coal
• Substitution of hydrogen for coal and oil
• Decarbonization of reagents used; for example, in the production of lime used for
neutralization in the hydrometallurgical processes
• In general, the hydrometallurgical route has lower carbon intensity. Therefore, for
new facilities and for major plant refurbishment it is an option.
7 Conclusion
Decarbonization of the mining sector as well as four major metals/alloy production,

namely steel, aluminium, copper and nickel representing a total of >12% of global
GHG emissions have been discussed. Decarbonization of the Mining and Metals indus-
try requires decarbonization of the energy sector. The pathways include incremental
improvements, electrification, process innovation, increased recycling as well as the
use of carbon-free hydrogen and CCS technology. However, infrastructure for both
hydrogen production and distribution as well as for CCS is yet to be developed at any
significant level. The capital cost required between 2020 and 2050 is on the order of
several trillion dollars. This would lower the carbon intensity of energy production and
metal mining and production, but it will be far from net-zero. To reach net-zero signifi-
cantly more technological developments and capital investments are required. To make
the technologies viable a carbon price in the range $50–150 per tonne of CO2 is required.
It should be understood that this would necessarily entail substantial increase in the cost
of energy and metals.
References
1. Our World in Data. https://ourworldindata.org/energy-production-consumption. Last Accessed 27

Feb 2022.
2. US Energy Information Agency (EIA). 24 Sept 2019. https://www.statista.com/statistics/859178/
projected-world-electricity-generation-capacity-by-energy-source/. Last accessed 27 Feb 2022.
3. US Energy Information Agency (EIA). Annual energy outlook 2021 with projections to 2050.
https://www.eia.gov/outlooks/aeo/. Last accessed 27 Feb 2022.
4. Projected electricity generation capacity worldwide from 2018 to 2050. Statistica. https://www.
statista.com/statistics/859178/projected-world-electricity-generation-capacity-by-energy-source/.
Last accessed 27 Feb 2022.
5. Gonzales V, Krupnick A, Dunlap L. Carbon Capture and Storage 101, Resources for the future.
May 2020. https://www.rff.org/publications/explainers/carbon-capture-and-storage-101/. Last
accessed 27 Feb 2022.
6. Rubin ES, Davison JE, Herzog HJ. The cost of CO2 capture and storage. Int J Greenh Gas Control.
2015;40:378. https://doi.org/10.1016/j.ijggc.2015.05.018.
7. Mining companies based on carbon dioxide emissions worldwide in 2017. Statistica. https://
www.statista.com/statistics/729396/co2-emissions-levels-of-mining-companies-worldwide/. Last
8. Kholod N, Evans M, Piltcher RC, et al. J Clean Prod. 2020;256:120489. www.elsevier.com/
locate/jclepro
9. Global Mine Methane, Global Methane Initiative. Sept 2011. www.globalmethane.org
10. A revamped cost curve showcases the biggest carbon-cutting opportunities. Environmental Defence
Fund. 2020. http://blogs.edf.org/climate411/2021/08/31/a-revamped-cost-curve-showcases-the-
biggest-carbon-cutting-opportunities/. Last accessed 27 Feb 2022.
11. Creating the zero-carbon mine. In: Metals and mining. McKinsey & Company. 29 June 2021.
12. Zero-emission copper mine of the future. The Warren Center Report WC4975–20. May 2020.
https://www.sydney.edu.au/content/dam/corporate/documents/faculty-o f-e ngineering-a nd-
information-technologies/industry-and-government/the-warren-centre/zero-emissions-copper-
mine-of-the-future-report-1-the-warren-centre.pdf
13. Glencore Raglan mine renewable electricity smart grid pilot demonstration, natural
resources Canada. https://www.nrcan.gc.ca/science-and-data/funding-partnerships/funding-
opportunities/current-investments/glencore-raglan-mine-renewable-electricity-smart-grid-pilot-
demonstration/16662. Last accessed 27 Feb 2022.
14. Varaschin J, De Souza E. Economics of diesel fleet replacement by electric mining equipment.
2015. https://www.scribd.com/document/358985531. Last accessed 27 Feb 2022.
15. Varaschin J. The economic case for electric mining equipment and technical considerations relat-
ing to their implementation. Thesis submitted for Master’s degree, Queens University, Kingston,
Sept 2016.
16. 2021 World steel in figures. World Steel Association. 2021.
17. McKinsey Report. Decarbonization challenges for steel. 3 June 2020. https://www.mckinsey.com/
industries/metals-and-mining/our-insights/decarbonization-challenge-for-steel
936 A. D. Dalvi
18. He K, Wang L. Review of the energy consumption and production structure of China’s steel
industry: current situation and future development. Metals. 2020;10:302. https://doi.org/10.3390/
met10030302.
19. Global crude steel production by process route and scenario, 2019–2050.
IEA. 8 Oct 2020. https://www.iea.org/data-and-statistics/charts/
global-crude-steel-production-by-process-route-and-scenario-2019-2050
20. Hasanbeigi A, Springer C. How clean is the U.S. steel industry? An international benchmarking of
energy and CO2 intensities. San Francisco: Global Efficiency Intelligence; 2019.
21. Nicholas S, Basirat S. New from old: the global potential for more scrap steel recycling. Institute
for Energy Economics and Financial Analysis (IEEFA). Dec 2021.
22. Bekir BC. World steel, the future of world scrap availability. https://worldsteel.org/media-centre/
blog/2018/future-of-global-scrap-availability/. Last accessed 27 Feb 2022.
23. Latest developments in steelmaking capacity. OECD. 2021. www.oe.cd/steelcapacity. Last
24. Available and Emerging Technologies for Reducing Greenhouse Gas Emissions from the Iron and
Steel Industry, US EPA. Sept 2012.
25. Guevara OL, Mahmoud M, Bene C, Rademaekers K. Zero Emission Steel, Technologies Available,
Prospects, Timeline and Costs, Trinomics, PE695.484. Dec 2021.
26. Fan Z, Friedmann J. Low carbon steel production, options and assessment. Columbia Centre on
Global Energy Policy. 2 Mar 2021.
27. Hornby S. Hydrogen based DRI EAF steelmaking- fact or fiction? https://www.researchgate.net/
publication/353328019. Jan 2021.
28. De Santis M, et al. Green Steel for Europe- Investment Needs. Mar 2021. https://www.estep.eu/
assets/Uploads/GreenSteel-D2.2-Investment-Needs-Publishable-version.pdf
29. Vogl V, Olsson O, Nykvist B. Phasing out the blast furnace to meet global climate targets. Joule.
5(10):2646–62. Cell Press, 20 Oct 2021.
30. IEA. Iron and steel technology roadmap: towards more sustainable steelmaking. Paris: OECD
Publishing; 2020. https://doi.org/10.1787/3dcc2a1b-en.
31. Iron and Steel Technology Road Map, IEA. Oct 2020. https://www.iea.org/reports/
iron-and-steel-technology-roadmap
32. Arcelor Mittal Decarbonization Project in Hamilton, ON Canada Announcement. 15 Feb 2022.
https://corporate.arcelormittal.com/media/press-releases/arcelormittal-decarbonisation-project-in-
hamilton-canada-confirmed-with-the-announcement-of-a-cad-500m-investment-by-the-governm-
ent-of-ontario. Last accessed 27 Feb 2022.
33. ArcelorMittal Europe to produce ‘green steel’ starting in 2020. 13 Oct 2020. https://corporate.arce-
lormittal.com/media/news-articles/arcelormittal-europe-to-produce-green-steel-starting-in-2020.
Last accessed 27 Feb 2022.
34. Algoma steel announces final investment decision for electric arc steelmaking. 21 Nov 2021. https://
www.globenewswire.com/news-release/2021/11/11/2332532/0/en/Algoma-Steel-Announces-
Final-Investment-Decision-for-Electric-Arc-Steelmaking.html
35. Starting our journey towards CO2-reduced steel production. https://www.thyssenkrupp-steel.com/
en/company/sustainability/climate-strategy/. Last accessed 27 Feb 2022.
36. A fossil-free future, hybrit fossil-free steel. https://www.hybritdevelopment.se/en/a-fossil-free-
future/. Last accessed 27 Feb 2022.
37. Rethink energy Webinar – the future of green steel. 11 Oct 2021. https://www.youtube.com/
watch?v=D_fKcr1ZF9o&t=1328s
38. International Aluminium Institute, Statistics, 2021. https://international-aluminium.org/statistics/
primary-aluminium-production/. Last accessed 27 Feb 2022.
39. IAI material flow model – 2021 update. https://international-aluminium.org/wp-content/
uploads/2021/05/IAI-Material-Flow-Model-2021-Update.pdf. Last accessed 27 Feb 2022.
40. International Aluminium. 25 May 2022. https://international-aluminium.org/
post-consumer-aluminium-scrap-tops-20-million-tonnes-for-the-first-time/
41. Closing the gap for aluminium emissions. World Economic Forum. Dec 2021. https://www3.wefo-
rum.org/docs/WEF_AFC_Unlocking_Technological_Scale_up_2021.pdf
42. Saevarsdottir G, Kvande H, Welch BJ. Aluminum production in the times of climate change.
JOM. 2020;72(1). https://doi.org/10.1007/s11837-019-03918-6.
43. Brough D, Jouhara H. The aluminium industry: a review on state-of-the-art technologies. Int J
Thermofluids. 2020. https://doi.org/10.1016/j.ijft.2019.100007.
44. Aluminium for climate: exploring pathways to decarbonise the aluminium industry. World Economic
Forum. Nov 2020. https://www3.weforum.org/docs/WEF_Aluminium_for_Climate_2020.pdf
45. Aluminium sector greenhouse gas pathways to 2050. International
Aluminium Institute. Sept 2021. https://international-aluminium.org/resource/
aluminium-sector-greenhouse-gas-pathways-to-2050-2021/
46. Kawatra SK, Young C, editors. SME mineral processing and extractive metallurgy handbook:
Society for Mining, Metallurgy & Exploration (SME); 2019.
47. Clemence C. Leaders emerge in the aluminium industry’s race to zero carbon. Aluminium Insider.
25 Feb 2022.
48. Magnusson JH. Clean aluminium production process, Arctus. Sept 2015. https://www.stjornar-
radid.is/lisalib/getfile.aspx?itemid=e2d429a6-6721-11e7-9416-005056bc4d74
49. Nuclear to power world’s largest aluminium smelter. World Nuclear News. 9 Oct 2007. https://
www.world-nuclear-news.org/
50. The World Copper Fact Book 2021, International Copper Study Group, www.icsg.org. Last
51. Basov V. Kitco News. 1 Feb 2022. https://www.kitco.com/news/2022-02-01/Top-10-largest-
copper-producing-countries-in-2021-report.html
52. Reuters Staff. Commodities News. 22 June 2021. https://www.reuters.com/article/
us-glencore-metals-copper-idUSKCN2DY25N
53. Elshkaki, A, Graedel T, Ciacci, L, Reck B. Science for environmental policy. (470). 16 Sept 2016.
https://doi.org/10.1016/j.gloenvcha.2016.06.006.
54. Nilsson AE, et al. A review of the carbon footprint of cu and Zn. Fortschr Mineral. 2017;7:168.
https://doi.org/10.3390/min7090168.
55. The World Nickel Market in 2021 INSG; World Stainless Steel. June 2021. www.stainless-steel-
world.net
56. Total Nickel Demand by Sector and Scenario 2020–2040. IEA. Jan 2022. https://www.iea.org/
data-and-statistics/charts/total-nickel-demand-by-sector-and-scenario-2020-2040
57. Wei W, et al. Energy consumption and greenhouse gas emissions of nickel products. Energies.
2020;13:5664. https://doi.org/10.3390/en13215664.
58. Fukuzawa R. Climate change policy to foster pollution prevention and sustainable industrial
practices – a case study of the global nickel industry. Miner Eng. 2012;39:196–205. https://doi.
org/10.1016/j.mineng.2012.07.011.
59. Skarn Associates. Carbon emission curves for iron ore, copper, met coal
and nickel; Northern Miner. 10 Sept 2020. https://www.mining.com/
carbon-emission-curves-for-iron-ore-copper-met-coal-and-nickel/
60. Azavedo M, et al. How clean can the nickel industry become? McKinsey. 11 Sept 2020.
61. Dalvi AD, et al. Developments at P.T. Inco’s Indonesian Nickel Project, CIM, 25th Annual
Conference of Metallurgists. 1986. p. 334–55.
62. Glencore’s Raglan mine to add energy-producing turbines. CIM Magazine. 11 Mar 2021.
Decarbonization at the Anglo-American Barro Alto
Smelter Through Implementation
of the Ecombustible Technology
Daniel Brosig1 (*), Paykan Safe2, Matt Russell2, Jorge Arevalo3, Carlos Kiyan3,
Mario Rodrigues4, and Marcelo Miranda4
1
GCT Engineering Inc., Vancouver, BC, Canada
brosig@gcteng.com
2
Gas Cleaning Technologies LLC, Irving, TX, USA
3
Ecombustible Energy LLC, Fort Lauderdale, FL, USA
4
Anglo American Niquel Brasil, Ltda, Barro Alto Goias, Brazil
Abstract. As part of the effort to decarbonize the world economy, the advent of the use
of hydrogen has awakened interest within the metallurgical industry. Sustainability-
driven strategic goals sponsored by the Anglo-American Barro Alto industrial complex
call for the displacement of fossil fuels with hydrogen to reduce carbon dioxide (CO2)
emissions in existing thermal applications. The environmental and business case strength
for the use of hydrogen in thermal processes can vary significantly based on the hydrogen
source and the local economic conditions, such as the cost of raw materials and electricity
and the price of incumbent fossil fuels. Important safety considerations require close
attention as the use of hydrogen as an industrial fuel grows toward becoming ubiquitous
in all industries. This paper presents an overview of an ongoing project to implement a
novel technology that will deliver a hydrogen-based fuel to thermal applications within
the Anglo-American Barro Alto smelter.
Keywords: Decarbonization · Hydrogen · Environmental sustainability
1 Introduction
Ecombustible Energy LLC has developed a patented technology that can generate a
customizable hydrogen-based fuel from a novel process which consists primarily of a
water electrolyzer and a magnetic reactor. The proprietary hydrogen-based fuel has a
higher calorific value than conventional molecular hydrogen fuel on a volume basis.
This gaseous product has been named eCombustible.
Ecombustible Energy LLC has signed an agreement with Anglo American Niquel
Brasil Ltda to deliver eCombustible fuel with the goal of decarbonising some thermal
applications at the Barro Alto Industrial Complex which currently consumes Liquified
Petroleum Gas (LPG). Gas Cleaning Technologies LLC and Ecombustible Energy LLC
are engineering the implementation of an eCombustible fuel generation plant to be
installed within the Barro Alto Industrial Complex as part of Anglo American’s strate-
https://doi.org/10.1007/978-3-031-17425-4_104
940 D. Brosig et al.
gic objective of producing green ferronickel. This paper presents an initial overview of
this project. Process details around the eCombustible technology implementation will
be presented at a later date.
2 eCombustible Technology
The eCombustible technology consists of a unique industrial electrolyzer that requires

less power to decompose water into hydrogen and oxygen. The thyristor or SCR con-
troller is tuned in such a way that it dissipates a minimum amount of energy in the
microcell units without affecting the quantity of hydrogen generation. Another unique
feature of this technology involves the conversion of ortho-hydrogen into para-hydrogen
with the use of a magnetic reactor to generate eCombustible fuel. Both the produced
oxygen and eCombustible fuel are collected, compressed and stored prior to transfer to
an end-user. The system’s key process equipment includes:
• Water Treatment System
• Electrolyzer
• Water Recirculation Cooling System
• Gas Collection and Handling
• Magnetic Reactor
• Buffer Tanks, Compressors, and Storage Tanks for Hydrogen and Oxygen
• Drain Collection and Wastewater Handling System
3 Barro Alto Industrial Complex
The Anglo-American Barro Alto Industrial Complex is a nickel production facility

located in the municipality of Barro Alto, Goias, in Brazil’s mid-west region. Figure 1
shows location details.
The nickel (contained in ferronickel) production process adopted in the Barro Alto
Industrial Complex consists of ore crushing, homogenization, drying, calcining, reduc-
tion and refining. The Rotary Kiln – Electric Furnace (RKEF) technology is employed
for calcination and reduction of the ore. Figure 2 shows the Barro Alto Industrial
Complex general process flowsheet.
The ore extracted from the Barro Alto mine is initially processed in the ore prepara-
tion process, where the stages of crushing, homogenization and ore drying take place.
The prepared ore is fed to two rotary kilns (calcination stage), where the load is com-
pletely dried and calcined. The energy required for the operation comes from the use of
pulverized coal together with heavy fuel oil. With the rotary kiln feed, a charge of
reducing carbon is also added, which participates in nickel reduction reactions during
the reduction stage. In addition, dry pellets are fed into the calciners, which are pro-
duced using ore fines that are recovered from the calcination and reduction processes.
Out of the rotary kilns, the calcined ore is fed into two rectangular Electric Furnaces,
Decarbonization at the Anglo-American Barro Alto Smelter Through Implementation… 941
Fig. 1. Location of the Barro Alto Industrial Complex within Brazil.
Fig. 2. Barro Alto Industrial Complex process flow diagram.
where the charge is smelted to produce the pay metal (ferronickel or FeNi) and slag.
The ferronickel is transferred in ladles to the refining stage, where part of the impurities
existing in the alloy are removed before metal shotting and packaging for sale.
Within the Barro Alto Industrial Complex, there is an LPG fuel supply facility. The
LPG is used in the pelletizing area (pellet drying), in the coal pulverization plant (coal
drying), and also in the refinery for pre-heating of ladles and tundishes, and FeNi drying
(after shotting with water).
At the pellet drying plant, LPG is supplied to each of the two hot gas generators
(HGGs) to produce hot gas, which is used as a heat source to dry the pellets. The coal
plant processes coarse wet coal, simultaneously grinding and drying it inside a coal
mill. Dry pulverized coal is used in the drying and calcination stages as a fuel source.
The coal plant HGG produces hot gases which are used to directly dry the wet coal. The
refinery operates five LPG-fueled burners including one vertical ladle pre-heater, one
horizontal ladle pre-heater, two tundish pre-heaters, and one rotary dryer used to dry
FeNi.
4 Thermal Application
The eCombustible fuel will be produced on-demand to displace the use of LPG cur-
rently consumed in the pelletizing, pulverizing and refining stages, generating environ-
mental benefits for the Barro Alto Industrial Complex in addition to eliminating the
operation’s exclusive dependence on LPG. Similar to LPG, eCombustible will be
burned in the existing burners resulting in zero CO2 emissions. Unlike conventional
hydrogen, the unique eCombustible fuel provides similar flame characteristics and cal-
orific values as fossil fuels such as LPG which allow it to replace such fuels with mini-
mal modifications to existing equipment. The eCombustible fuel production plant’s
installed capacity is expected to be approximately 22 MW of thermal energy in deliv-
ered fuel. After installation of the eCombustible production plant, reserves of LPG will
be maintained only as a backup fuel. The existing combustion systems will be adapted
to operate dually with either of the two fuels.
The pellet dryers HGGs will be prioritized as the first stage to demonstrate the tech-
nical performance of the eCombustible fuel operation. A ramp-up program will be
implemented to ensure that the combustion performance of the new fuel satisfies safe
control standards and that the hot gas generated complies with the target process condi-
tions required to enable effective and efficient drying of the pellets. Figure 3 shows a
pellet dryer HGG.
Ecombustible Energy LLC is currently executing a testwork program to demonstrate
the applicability of the use of eCombustible fuel as an industrial fuel in common ther-
mal applications. A key priority in this testing program is to test the eCombustible fuel
combustion performance with a burner design equivalent to the one currently installed
at the commercial pellet dryer HGGs. The results of this testing will be evaluated and
considered as input for potential retrofit requirements at the Barro Alto pellet dryer
HGGs.
Fig. 3 Pellet dryer hot gas generator
5 Key Environmental Benefits
Replacing LPG fuel with eCombustible, which is more competitive from an environ-
mental and economic point of view, will provide benefits that are in line with Anglo
American Nickel’s strategic objectives, that is, the search for an increasingly “cleaner”
operation, in terms of reduction of CO2 emissions, as well as the reduction of operating
costs. The reduction in CO2 emissions with the use of eCombustible is estimated at
approximately 2869 t per month based on historical records. Table 1 shows a summary
of the historical average monthly CO2-eq emissions by area (pellet dryers, coal mill and
refinery) from 2016 to 2020.
The implementation of this project will be a milestone in the fulfillment of Anglo
American’s strategic objective of making its ferronickel operations the first in the world
to be operated 100% with renewable resources and 100% with internally reused water.
Upon successful completion of this project, Anglo American will consider implement-
ing the eCombustible technology at the Codemin ferronickel smelter located in
Niquelandia.
Table 1. Average monthly CO2-eq emissions (t CO2-eq/month).

Year Pellet dryers Refinery Coal mill Total
2016 2047 364 530 2941
2017 2151 376 228 2754
2018 1961 362 530 2853
2019 1993 362 465 2820
2020a 2027 362 588 2976
Average 2036 365 468 2869
Data range from January 2020 to August 2020
a
6 Environmental Impact
Potential impacts on flora and fauna are negligible as the plant and fuel will be installed
and used in an area already industrially developed. Clearing of vegetation will not be
necessary during the implementation of this project. The combustion of eCombustible
fuel within the Barro Alto smelter burner applications generates water vapor as the
main constituent of the combustion product. Production of other potentially controlled
emissions (e.g., nitrogen oxides or NOx) will be minimized by implementing appropri-
ate engineering and/or operational solutions.
The main waste stream resulting from the operation will be wastewater from the
water treatment system. The physicochemical characteristics and flow rate of this
wastewater are estimated to fall within the environmentally-acceptable permissible lim-
its to allow for its safe disposal within Barro Alto’s existing process water handling
system.
Regarding the mitigation of environmental impacts, Anglo American already prac-
tices control measures aimed at controlling impacts on the industrial site. These mea-
sures take place throughout Anglo American’s site in Barro Alto and will extend to the
location of the eCombustible plant.
7 Technology Integration
The eCombustible production plant will be built in a greenfield area conceded by Anglo
American within the Barro Alto industrial complex as part of the agreement with
Ecombustible Energy LLC. The project requires earthworks and construction of new
foundations and reinforcements to support the equipment. An electrical equipment
room, control room and small warehouses dedicated to the proper storage of chemical
reagents will be included at the plant’s construction site. eCombustible electrolyzers
will not be permanently installed on site. Instead, they will be installed in containerized
modules that allow for easier transport, installation and subsequent mobility (when
necessary). The same containerized approach (or alternatively, a module-assembled
approach) will be followed for other components (e.g., electrical equipment room, con-
trol room, warehouses) and auxiliary process systems (e.g., electrolyte cooling system
and water treatment system). A security system will be installed around the perimeter
of the eCombustible plant to prevent unauthorized persons from entering the area, as
well as bollards around all tanks to prevent accidental vehicular collisions.
Existing pipeline infrastructure (including gas feed trains) currently used to trans-
port LPG to all HGGs and burners will remain unaltered. This will allow the Barro Alto
smelter to continue using LPG normally and at its discretion. New dedicated pipelines
will be installed to transport the eCombustible fuel and oxygen to all end-users. These
systems will be equipped with all the necessary instrumentation to guarantee the distri-
bution of gases with quality and safety (e.g., flow meters, pressure, and gas leak detec-
tors). Additional piping requirements will include pipelines to supply raw water to the
eCombustible system from existing infrastructure.
The eCombustible system consumes electrical energy for the operation of its various
equipment and instruments. Most of the energy will be consumed by the electrolyzer
cells to carry out the electrolysis process. Other devices with lower electricity con-
sumption within the process are the magnetic reactor, the gas compressors and various
pumps. The Barro Alto smelter will supply power via a high-voltage transmission line
that will be installed from the existing main electrical substation to the eCombustible
plant. This new line will be connected to a new local transformer, to be acquired and
installed by Ecombustible Energy LLC, within the perimeter of the eCombustible pro-
duction facility.
Inside the eCombustible production plant, there will be a facility to receive and treat
the raw water necessary for the production of the eCombustible fuel. This subsystem is
designed to treat the entire volume of water intended for consumption by the plant and
will be responsible for reducing water hardness and removing other impurities that
could affect the performance of the water electrolysis process. The technology to treat
this water consists primarily of ultrafiltration and reverse osmosis systems.
The eCombustible production plant’s location has been selected in collaboration
with Barro Alto personnel by identifying the least environmentally impactful area (no
need for vegetation suppression, no impact on fauna, flora and water courses), and the
potential for future production expansion. Figure 4 shows the area selected for the con-
struction of the new plant.
8 Safety Considerations
The design of the eCombustible fuel production and combustion systems follows a
series of codes and standards to ensure compliance with safe design requirements.
International, Brazilian and local codes and standards are used in the design process to
minimize risks in the production, storage, handling and combustion of the eCombusti-
ble fuel. Additionally, multiple technical peer reviews have been undertaken by inde-
pendent professional safety consultants. Anglo American personnel have also actively
participated in the development of Hazards Identification and Hazards and Operability
workshops. The following subsections describe basic design features to address safety
considerations for the main process areas.
Fig. 4. eCombustible production plant location.
8.1 Water Treatment System
Water holding tanks are installed to ensure a constant supply of raw water to the reverse
osmosis system and clean water to the electrolytic cells to prevent fluctuations in water
flow which can damage and/or affect the performance of electrolytic cells. The tanks
themselves are equipped with level transmitters and drain valves to monitor the water
level and drain as needed. To prevent damage to the electrolytic cells, the quality of the
water (raw and treated) is monitored through various safeguards such as redundant
conductivity meters, pH meters and manual water hardness tests. Both automatic and
manual valves are installed in case of deviations and/or shutdowns. There are conduc-
tivity meters in the water recirculation line and pH meters in the treated water supply
line installed to ensure that incorrect KOH solution dosing does not lead to low hydro-
gen production and/or contamination of cell membranes.
8.2 Electrolysis
High temperatures can lead to evaporative loss of electrolyte solution, reduced process
efficiency and premature failure of cell stacks. An electrolyte cooling system is installed
to remove excess heat from the electrolyte (with temperature transmitters at the inlet
and outlet of the heat exchangers to monitor operating temperatures). Low electrolyte
flow in pumps can lead to cavitation and a lower volume of electrolytes in cells. The
latter, in turn, can lead to the burning of cell components, resulting in permanent dam-
age. In addition, a lower circulating volume of electrolytes would result in a lower level
of electrolytes in the oxygen and hydrogen holding tanks, which could lead to the mix-
ing of oxygen and hydrogen gases. The process is equipped with engineering controls
to avoid mixing oxygen and hydrogen and therefore mitigating the risk of explosions.
Level transmitters are installed on each gas holding tank accompanied by an automated
shutdown procedure when a low-level set point is reached. Sight glasses are installed in
the tanks for manual checking of liquid levels. Pressure transmitters are installed at the
pump discharge to monitor the discharge pressure. Electrical components are properly
insulated and non-conductive material is used in piping and hoses to prevent personal
injury. The electrical components of the electrolysis module are also physically shielded
to protect against potential electrolyte leakage.
8.3 Gas Collection and Handling
Water separators are installed in the ventilation lines to capture water as electrolyte car-
ryover in the gas can result in the accumulation of KOH (precipitate from the solution)
in downstream ducts and equipment. There are pressure relief valves and emergency
stacks with flame arrestors installed in the gas storage tanks in case of any mechanical
blockage causing pressure build-up. Hydrogen and oxygen are kept in dedicated, ade-
quately spaced holding tanks to minimize the risk of mixing and possible subsequent
explosions. There are pressure sensors installed in both tanks, and the pressure differ-
ential is monitored between them to always ensure that the hydrogen holding tank has
a higher pressure than the oxygen tank. This will reduce the risk of oxygen flowing
from the oxygen holding tank to the hydrogen holding tank.
8.4 eCombustible Fuel Combustion
The flow of the eCombustible gas supply is monitored through redundant pressure
transmitters and flow meters. Flame stability is monitored using UV/IR flame sensors
and exhaust temperature sensors. The air to gas ratio is controlled to prevent flame loss
and maintain optimal combustion conditions.
8.5 Electrical and Instrumentation System
All instrumentation is designed for Class 1, Division 2 per the National Electrical Code
(NEC) in the areas of electrolyzer and gas storage and processing. Standard security
system design is in accordance with applicable codes. All equipment will be properly
secured to reinforced concrete bases that will prevent damage from rain, and concrete
barriers are installed around key areas to prevent accidental vehicle collisions. There
are hydrogen and oxygen sensors installed around each gas storage tank in the generat-
ing plant to detect potential leaks.
9 Path Forward
The implementation of the novel eCombustible technology at the Anglo American

Barro Alto industrial complex promises to greatly enhance Anglo American’s position
as a producer of environmentally-sustainable nickel. Given that the technology is new,
significant risk reduction and mitigation steps are being considered to ensure that risks
to personnel, assets and the environment are kept as low as possible. The project team
is currently developing final preparatory work to execute the construction of the eCom-
bustible production plant in 2022.
Ni/Co–Cu Sulphide Ores, Key Contributors
to Sustainability: A Review of Applications
and Processing Challenges
S. Kelebek1, S. Arslangil1 (*), and C. E. Gibson1

1
The Robert M. Buchan Department of Mining, Queen’s University, Kingston, ON, Canada
s.arslangil@queensu.ca
Abstract. Nickel and cobalt are among the strategic metals contributing to sustainability.
Recent statistics have indicated that their domestic production gradually declined,
although Canada still ranks among the top five countries in nickel production linked to the
co-production of cobalt and precious metals. The sustainability of nickel and cobalt pro-
duction requires efficient processing, whether this relates to recycling from secondary
sources or primary resources. One of the main driving forces in the growing demand for
nickel is related to the increasing importance of nickel and cobalt-bearing lithium-ion
electric-car batteries to reduce dependence on fossil fuels as part of a sustainable future.
In addition to their use for stainless steel production, alloys of these metals are essential
for solar power facilities towards the same goals.
Nickel-copper sulphide ores in Canada are in the category of complex sulphides since
they are available with large quantities of pyrrhotite which is undesirable from an environ-
mental point of view. Thus, the rejection of pyrrhotite in the processing of Ni-Cu sulphide
ores will continue to be important in the Canadian mining industry due to its dominance
in the disseminated matrix of these ores. In addition, in some cases, these ores also have
serpentines and talc problems. Challenges related to mineralogy are reviewed together
with reagent functionality related to pyrrhotite rejection as well as non-sulphide gangue,
specifically talc and serpentines with an overview of promising new reagents.
Keywords: Demand · Productions and recycling trends · Processing challenges ·

Pyrrhotite rejection update
1 Introduction
1.1 Co-Presence of Nickel and Cobalt in Nature
The atomic properties of nickel and cobalt are very similar to one another as they are
neighbours in the periodic table. In the list of abundance [1], nickel ranks as #27 com-
pared to cobalt (ranking as #31). Due to the fundamental similarity, they often coexist
in the same minerals, and their behaviour typically goes together in mineral and metal-
lurgical processing until their separation using solvent extraction techniques. In the
https://doi.org/10.1007/978-3-031-17425-4_105
950 S. Kelebek et al.
production of nickel from primary sources such as sulphides of magmatic origin and
stratiform sediment-hosted deposits, cobalt is a typical byproduct [2].
1.2 Demand and Production of Nickel/Cobalt
Trends for Demand Between 2020 and 2050, demand for cobalt, nickel, and most
other minerals is expected to grow by factors of 17–19, 28–31, and 15–20, respectively,
necessitating a dramatic expansion of cobalt and nickel supply chains [3]. By 2020,
overall cobalt consumption is predicted to exceed 120,000 tonnes per year, an increase
of about 30% over the 93,950 tonnes consumed in 2016. Battery utilization is estimated
to account for 60% of total cobalt needs in 2020, up from 58% in 2016 [4]. Global stain-
less steel usage accounted for 70% of total primary nickel consumption in 2019, a
proportion that has been continuously increasing over the last decade. This dominant
position has been steadily consolidated over the last few decades due to China’s eco-
nomic stimulus, which fueled a construction boom. As shown in Fig. 1, following stain-
less steel, numerous additional industries utilize comparable amounts of primary nickel
[5]. Stainless steel is expected to continue to be the most popular material in 2040,
while batteries may begin to challenge stainless steel’s domination. In addition, the bat-
tery sector may massively increase demand share from 6% in 2020 to 36% in 2040, a
sixfold increase.
Trends for Production Production of Ni/Co is increasing every year due to the high
demand for Ni/Co. Nickel production was 2.51 million tonnes in 2020, then rose to 2.7
million tonnes in 2021 worldwide, and 130,000 tonnes of this amount were produced in
Canada. While 142,000 tonnes of cobalt were produced globally in 2020, this amount
of production grew to 170,000 tonnes in 2021, and Canada generated 4300 tonnes of
this total [6].
In addition, at least 300 million tonnes of nickel are contained in identified land-
based resources with an average nickel content of roughly 0.5% or more, with approxi-
mately 60% in laterites and 40% in sulphide deposits. Nickel resources are therefore
Fig. 1. Demand for nickel, by first-use sector [5].

Ni/Co–Cu Sulphide Ores, Key Contributors to Sustainability: A Review of Applications… 951
abundant on the ocean floor in manganese crusts and nodules. Furthermore, the large
proportion of these resources are centred in sediment-hosted stratiform copper deposits
in the Democratic Republic of the Congo and Zambia; nickel-bearing laterite deposits
in Australia and neighbouring island countries and Cuba; and magmatic nickel-copper
sulphide deposits hosted in mafic and ultramafic rocks in Australia, Canada, Russia,
and the United States [6].
1.3 Recycling as Part of Sustainability of Nickel/Cobalt Production
As the production from primary sources has been increasing to satisfy demands from
new markets such as the electric vehicle battery industry, recycling may aid in conserv-
ing natural resources, as expected from a sustainability point of view. Despite the com-
plicated composition of rechargeable batteries and the intricate worldwide production
chain, it turns out that the recycling scenario results in natural resource savings of
51.3%, not only because of the reduction in mineral ore reliance but also because of the
reduction in fossil resource demand (45.3% reduction) and nuclear energy need (57.2%
reduction) [7].
Several types of Co and Ni-bearing material are available for recycling, including
alloys, leftovers from stainless steel manufacture, and wasted batteries. Nonetheless,
the majority of procedures that use solvent extraction for the selective recovery of Co
and Ni are used to recycle spent batteries, which may include lithium-ion batteries
(LIBs), nickel-cadmium batteries (Ni-Cd), and nickel-metal hydride batteries (NiMH).
These sorts of batteries are a rich source of rare earth elements (REEs) and critical met-
als such as Ni, Co, Mn, and Cu [8].
Over the past several years, the use of LIBs for electronic-drive vehicles (EDVs) has
considerably risen in North America; in 2030, 676,000 LIBs for EDVs will be marketed
in Canada, Mexico and the United States, a more than 13-fold increase over the demand
in 2015 [9].
Recycling Co, particularly from used Li-ion batteries has added significance while
being a problem for countries with little local Co supplies. Apart from the technical and
logistical difficulties, the recycling process needs to be competitive with the new supply
produced from ore in terms of volume, cost, and environmental effect [2]. Recently, up
to 69% of end-of-life cobalt has been estimated to go to landfilling sites on a global
basis [10]. The recycling process and its success will be determined by the end-of-life
product containing cobalt and the price of cobalt [11]. The amount of cobalt that is
recycled is thought to be 68%, and the recycled content rate is thought to be 32% in
2011 [12]. Previously, the EU set an aim of recycling 40% of batteries by 2016 to mini-
mize the Co demand from primary ore [13]. For the electric vehicle (EV) industry
specifically, it is predicted that by 2030, recycled cobalt which comes from used LIBs
may fulfill up to 10% of EV cobalt demand [14]. The use of organic acids might be
considered as a technique of recycling that is environmentally favourable from an envi-
ronmental point of view [15].
2 Ni/Co Production from Primary Sources and Challenges
Domestic production of nickel and cobalt from primary sources relies on the processing
of nickel-copper sulphide ores, which dates back over a century in the rich history of
the mining culture of northern Ontario. Historical records relating to Canadian nickel
production indicate (S.O. Fekete, private communication, 2000) that the Sudbury area
was visited by Thomas Edison from the U.S.A, who predicted that Ni/Co would be
critical material in the making of efficient batteries to power electric engines. However,
later on, the electric engine lost its developmental competition to the combustion
engine, as gasoline from fossil fuels became available at low prices. Production of Ni/
Co from sulphide ores has its challenges that are fundamentally related to the mineral-
ogy of this type of ores.
2.1 Mineralogy of Nickel-Copper Sulphide Ores
Studies on the geology and mineralogy of domestic nickel-copper sulphide ore deposits
are reported in several sources in detail [16–19]. Although over 20 minerals have been
identified in these complex sulphide ores having primarily nickel, copper, and iron met-
als in a variety of stochiometries, these metals are contained in three principal sul-
phides, namely, pentlandite, chalcopyrite and pyrrhotite, respectively. The typical
composition of pentlandite is given by (Ni, Fe)9S8 with the nickel content fluctuating in
the range 34–36% depending on the location and conditions of genesis. Cobalt is pres-
ent as part of pentlandite (~1%). In addition, other nickel-bearing minerals such as
millerite (NiS), violarite (Ni2FeS4), niccolite (NiAs) are found in certain zones.
However, their average content in the plant feed is regarded as a minor amount.
Pyrrhotite (with compositions of Fe8S9 and Fe7S8) is found in monoclinic (predominant
type in the Sudbury area), which is moderately magnetic and hexagonal (non-magnetic)
commonly found in the Voisey’s Bay ore. Both types of pyrrhotites are nickeliferous in
nature. The nickel in pyrrhotite is found as a solid solution in the crystal structure in
variable amounts (up to about 1.4% Ni in the case of Sudbury ores). Although this
nickel content is much smaller than that of pentlandite, the overall nickel share of pyr-
rhotite in a typical plant feed could be pretty significant (at about 10%). This is because
pyrrhotite is 5–15 times more abundant in these ores than pentlandite. The ratio of pyr-
rhotite/pentlandite is a critical variable in the processing of these ores. It is a determin-
ing factor for the recovery of nickel. At the same separation efficiency of nickel-bearing
minerals, the higher the pyrrhotite/pentlandite ratio, the lower the nickel recovery in
daily operations. Other iron minerals such as pyrite (FeS2) and magnetite (Fe3O4) are
also present in small amounts.
Chalcopyrite (CuFeS2), with 36.4% Cu, is the main copper-bearing mineral. In some
nickel-copper sulphide ores, chalcopyrite is much more abundant than pentlandite. As
an additional Cu-bearing sulphide mineral, the deposits also have cubanite (CuFe2S3) in
smaller amounts. Other minerals that are present in small amounts include marcasite,
galena and precious metal minerals.
Global trends such as decreasing ore grade and stricter environmental regulations
are very much felt in the Sudbury region where two major companies have been
o perating with long histories. In general, Vale’s ore grades are relatively lower, which
can be attributed to the use of bulk mining methods [20]. The recent nickel grade range
can be as low as 1.0–1.20%. The ore grades at Raglan and Voisey’s Bay are much
higher than in the Sudbury region. However, since the production from Nickel Rim
South Mine was brought online, Glencore’s flotation plant Strathcona started process-
ing a higher grade ore with 1.57% Ni and 2.85% Cu ([21]).
2.2 Challenges in Processing of Nickel-Copper Sulphide Ores
The issues in the domestic Ni-Co production are related to the complexity of the sul-
phide portion of these ores, as noted above. A major problem is centred around the
abundance of pyrrhotite in these ores. Even if large levels of pyrrhotite rejection can be
achieved using a mineral processing technique, which is not easy, its low recovery lev-
els can still make a significant negative impact on the quality of final nickel concen-
trates [22]. This is a common problem in the processing of these ores regardless of the
type of deposits; layered mafic, ultramafic or volcanic. Since the abundance of pyrrho-
tite has been reflected in the amount of SO2 emissions from the Sudbury area smelters,
emission control strategies proposed included an option which practically meant stop as
much pyrrhotite as possible reaching the smelting furnaces [23]. Both companies devel-
oped their SO2 abatement programs as part of this strategy, which is considered as the
most reasonable strategy for a sustainable nickel industry in Canada [24].
Pyrrhotite Problem and Consequential SO2 Abatement Program Although it has
nickel in its structure, pyrrhotite is considered a gangue mineral since the amount of
nickel is small compared to the amount of sulphur it carries. In the process, it responds
to the same collectors intended for pentlandite and because of the high degree of dis-
semination of fine pentlandite in the pyrrhotite matrix, its efficient separation has been
difficult from the pentlandite concentrates going to the smelters. Thus, any excess
amount of pyrrhotite reporting to the smelter feed has been a significant contributor to
SO2 emissions. From a metallurgical point of view, however, pyrrhotite was a source of
additional nickel, plus it was a significant contributor to the energy input of the smelting
process. In the 60s and 70s, the consequence of environmental damage that can be
caused by the abundant presence of pyrrhotite in the smelter feed was not realized as
much as in the subsequent decades. Thus, in this earlier period, the environmental cost
of nickel production was relatively heavy. Trends of high levels of emissions continued,
which led to imposed limits on the emission levels. Thus, the companies initiated SO2
abatement programs. In 1970, about 2300 kilotonnes of SO2 had been emitted into the
Sudbury air, which was drastically reduced in the subsequent decades due to process
developments in mineral and pyrometallurgical processing practices. Figures 2 and 3
show the SO2 emission limits and accomplishments of Inco (now Vale) [27] and
Falconbridge (now Glencore) in reducing the level of emissions over the years, respec-
tively. As can be noted, both companies can meet the requirements of regulations and
emission limits imposed.
R & D projects undertaken resulted in the implementation of new processes [28, 29].
The new practice involved the use of diethylenetriamine (DETA) as a chelating reagent
that was effective in minimizing the copper/nickel activation of pyrrohotite for
Fig. 2. SO2 emission reduction at the Copper Cliff, Sudbury [25].
Fig. 3. SO2 emission reductions 1960–2019 – Sudbury INO Smelter-Falconbridge/Glencore [26].
co-flotation with pentlandite. The latter work had SO2 as an additional ingredient that
improved the consistency in the separation of pyrrhotite from pentlandite. Given the
challenges of pyrrhotite depression, the research work on DETA continued with TETA
(triethylenetetramine), which is a stronger chelating agent for the potential activating
species of copper and nickel for pyrrhotite. The ore sample investigated was the Ni-Cu
sulphide ore from Voisey’s Bay [30], which indicated an excellent pentlandite-pyrrhotite
selectivity with the combined use of SMBS and TETA. However, under the conditions
of highly efficient pentlandite-pyrrhotite separation, the flotation of chalcopyrite was

noted to be slowed down. Thus, it can be concluded that chalcopyrite should be removed
effectively before the pentlandite-pyrrhotite separation stage, which is the normal
practice.
Problems with Non-sulphide Gangue Minerals Another challenge with Ni-Cu sul-
phide ores is related to the process behaviour of non-sulphide minerals present in the
host rock. These are Mg-bearing complex silicates with a very low background concen-
tration of nickel or copper. Broadly, they are represented within the phyllosilicate
group, or sheet silicates, as an important group of minerals that includes the micas, talc,
chlorite, and serpentines. Talc, Mg3Si4O10(OH)2 in this group, exists in some base metal
sulphides, and some molybdenite ores, where separation issues can arise due to co-
flotation tendencies [31]. Another phyllosilicate is chlorite, Na0.5Al4Mg2Si7AlO18
(OH)12.5(H2O) (webmineral.com) [32]. These silicates, despite the dominant ionic
bonding in their structure, exhibit hydrophobic characteristics due to layered structures
that hold their crystals together by weak Van der Walls forces forming cleavage planes
without rupture of ionic bonds [33]. Thus, due to inherent hydrophobicity, they tend to
float without needing any collector and report to the concentrates of metal sulphides.
Another group of Mg-bearing silicates are serpentinized forms of silicates such as
chrysotile and lizardite, both of which are examples of polymorphous crystal lattices
corresponding to the same composition of Mg3Si2O5(OH)4 (or 3MgO.2SiO2.2H2O) but
they, unlike talc and chlorite, are essentially hydrophilic. These are specific to Ni-Cu
sulphides hosted in ultramafic (and volcanic) types of deposits that are mined and pro-
cessed domestically and in other countries like Australia [34]. Canadian examples
include Thompson operations in Manitoba [35], Raglan deposit in northern Quebec
[36], and Montcalm deposit near Timmins [37].
The metallurgical concerns with these silicates are distinctly different. The hydro-
phobic group of phyllosilicate minerals, when inefficiently depressed in the flotation
process, report to the nickel concentrates causing dilution and finding their ways to
smelter furnaces and build up at the bottom. If the MgO level becomes too high in the
smelter feed, they increase viscosity problems in the slag. High viscosity is undesirable
since it influences matte separation negatively causing elevated metal losses in slag. In
addition, they react with refractory brick liners in the furnaces fusing with their
MgO. Thus, they gradually accumulate in the furnaces occupying an unnecessary space
and causing serious maintenance issues and ultimately shortening the life of brick liners
[18].
The hydrophilic group (e.g., containing asbestos-like silicates such as lizardite and
chrysotile) does not present a serious issue for the smelter operations, since they do
not report to the concentrates in significant amounts. The main concern with them is
their adverse effect on flotation separations of sulphide minerals, primarily their
potential depression of pentlandite. Fundamental research work by Edwards et al.
[38], demonstrated this effect through electrokinetics investigations. These mineral-
ogical challenges initiated a great deal of subsequent research work, which are high-
lighted below.
3 Selected Results from Recent Investigations
3.1 Flotation and Depression Behaviour of Pyrrhotite Types
Since control of pyrrhotite’s flotation behaviour is central to the successful processing

of Ni-Cu sulphide ores and pyrrhotite is variable from deposit to deposit in terms of not
only its amount but also the crystal type, recent research focused on a better understand-
ing of the floatability of pyrrhotite types encountered in these ores. Although there are
at least four different types of natural pyrrhotite structures stable at room temperature,
a variety of monoclinic and hexagonal crystal structures are most common, compared
to other forms. Based on an analysis of pyrrhotites from 82 sources, Arnold [39]
reported that 73% were intergrown between hexagonal and monoclinic forms. The fre-
quencies of these as singular phases were 13% and 9%, respectively, and 5% of the
pyrrhotite samples studied indicated intergrowths of hexagonal pyrrhotite and hexago-
nal FeS (troilite). Various authors investigated the floatability of main types of pyrrho-
tites under various conditions. Becker et al. [40], reported that hexagonal (non-magnetic)
pyrrhotite from the Sudbury area was less reactive to oxidation and floated better than
monoclinic pyrrhotites investigated. In a recent review, Multani and Waters [41]
reported comparative results on these two types of pyrrhotite which suggested similar
behaviour of hexagonal pyrrhotite with respect to pentlandite recovery [42] and on size
by basis [43]. Results reported by Wercholaz [42] are shown in Fig. 4.
However, in a more recent study, Zhao et al. [44], reported that monoclinic pyrrho-
tite is more floatable than hexagonal pyrrhotite. These authors conclude that contrary to
the previous work by Becker et al., hexagonal pyrrhotite is more reactive than mono-
clinic pyrrhotite based on their density functional study supported by the results of
flotation experiments. Since the surfaces of pyrrhotite are sensitive to oxidation, it is not
surprising to see differences in flotation behaviour that is also affected by its nickel
content. It should be noted that results reported by both Becker et al. [40] and Zhao
Fig. 4. Flotation behaviour of hexagonal and monoclinic pyrrhotite relative to that of pentlandite
[42].
et al. [44, 45] are based on small-scale experiments with pure minerals, whereas those
reported by Wercholaz [42] and Lawson et al. [43] are based on testing of Ni-Cu sul-
phide ore sample.
3.2 Reagents and Assessments in Operating Plants
In view of different flotation behaviours of hexagonal and monoclinic pyrrhotite, recent

work has included results on reexamination of DETA/TETA for pyrrhotite-pentlandite
separation without and with sodium metabisulphite (SMBS) or sodium sulphite with
respect to the type of pyrrhotite. Recovery by size data reported on the flotation of
Ni-Cu sulphide ore from Voisey’s Bay rich in hexagonal pyrrhotite indicated an excel-
lent depression of the pyrrhotite in a wide particle size range [46]. Results from this
work are shown in Fig. 5.
On a Ni-Cu sulphide ore in the Sudbury Basin with 42% hexagonal pyrrhotite and
58% monoclinic pyrrhotite, Multani and Waters [47] demonstrated DETA/SMBS.
The combination was equally effective on a mixed pyrrhotite using a relatively high-
grade ore with 1.6% Ni, as can be noted below in Fig. 6. These results further d emonstrate
the consistency of DETA/SMBS reagents system on rejection of pyrrhotite whether it
is mixed type with hexagonal type or fully magnetic (monoclinic) type [28]. The shift
of selectivity curves is also in line with the results of theoretical analysis on the effect
of xanthate dosage relative to depressant action [48].
In a more detailed study on a better understanding of the pyrrhotite depression
behaviour under plant conditions, it has been shown that the hexagonal pyrrhotite more
100
Blank
90 SMBS
80 TETA
SMBS/TETA
70
Pyrrhotite Recovery (%)
60
50
40
30
20
10
0
1 10 100 1000
Average Particle Size (m)
Fig. 5. Hexagonal pyrrhotite recovery by particle size from Voisey’s Bay Ni-Cu sulphide ore under
depression reagent conditions stated [46].
Fig. 6. Po-Pn flotation selectivity using DETA/SMBS/PIBX under various conditions [47].
Table 1. The ratio of metal species on pyrrhotite surfaces in various streams [49]a.
Process stream Hexagonal pyrrhotite Monoclinic pyrrhotite
Mg/Fe Ca/Fe Ni/Fe Cu/Fe Mg/Fe Ca/Fe Ni/Fe Cu/Fe
Nickel ore feed 0.40 0.57 0.17 0.09
Primary rougher con. 0.37 0.39 0.26 0.47 0.16 0.30 0.57 0.32
Primary rougher tail 0.44 0.59 0.14 0.04 0.44 0.60 0.12 0.03
Nickel con. 0.53 258 0.25 0.12
Pyrrhotite tail 0.29 1.61 0.17 0.09
From data reported by Qi et al. [49]
a
easily floats reporting to the concentrates compared to monoclinic type that tends to
stay depressed and reports to the tailing due to relatively heavier oxidation [49]. This
study also demonstrated through cyclic voltammetry that hexagonal pyrrhotite is less
reactive (while showing a greater tendency of getting Cu-activated) than the monoclinic
type that has an easier tendency for oxidation leading to ferric hydroxy species. The
results reported on the potential activating effects of Cu and Ni ions on pyrrhotite flota-
tion are complementary to their suspected roles reported previously [50–53]. A TOF-
LIMS study carried out in this period indicated similar results on the relative amount of
Cu and Ni ions on the pyrrhotite reporting to the concentrates, however, with no specific
reference to the two types of pyrrhotite. Table 1 shows the ratio of metallic species
suspected of causing depression (e.g., Mg, Ca, Fe) and activation (Ni & Cu) of both
hexagonal and monoclinic pyrrhotite [49]. Although the data on monoclinic pyrrhotite
for the Ni concentrate and pyrrhotite tailings are not reported, it is clear that pyrrhotite
reporting to the primary rough concentrate has had a greater share of activating
species.
In some Ni-Cu sulphide ores with low sulphide mineral content, but rich in precious
metal levels, processing requires activation rather than depression. Recent work sug-
gests that the ammoniacal solution of copper sulphate is more effective on pentlandite
than the copper sulphate itself in the alkaline range [54]. The effectiveness was related
to the lack of hydroxide precipitate formation and improved electrochemistry leading
to the formation of a greater level of hydrophobicity of the pentlandite surface.
3.3 Side Effects of Reagents Used in Pyrrhotite Deactivation/Depression
The use of complexing agents such as ammonia and DETA or its mixtures with TETA
can undoubtedly be useful for certain objectives to improve flotation separations of
sulphide minerals as documented in various publications reviewed. However, in closed-
loop process water systems through stored water in tailings areas, they will tend to build
up and reach elevated levels which can interfere with the practice of the effluent treat-
ment plants since a simple pH control may not be enough to meet the allowed limits of
discharge flows. Since most of the water is recycled from tailing areas, recent research
also directed to the adsorption and stability of metal chelates [55, 56] and their biode-
gradability in the environment [57, 58].
3.4 Flotation and Depression Behaviour of Non-sulphide Minerals
Depression of Hydrophobic Phyllosilicates As noted previously, talc and chlorite

need to be efficiently depressed since they are natural floaters. For this purpose, CMC
is commonly used in plant practice. It is an anionic macromolecular modifier, a car-
boxymethyl substituted polysaccharide (see below), which can act both as a depressant
and dispersant. The response of Raglan Ni-Cu ore to CMC depression of these hydro-
phobic gangue minerals was very good, as shown in Fig. 7 [36]. However, the Ni grade
recovery was not completely restored considering the behaviour of ore from the under-
ground zone with little or no talc. The operating mechanism involves various modes of
binding of large CMC molecules onto talc which includes hydrophobic and hydrogen
bond formation with the required orientation to expose a hydrated surface towards the
water phase [59].
Its derivatives like carboxymethyl starch (CMS) [60] and substituted varieties were
suggested for applications in the processing of other ores [61]. It is effective in mildly
Fig. 7. The effect of CMC on talc depression in the flotation of Raglan ore with 8–10% pentlandite
containing 6.3% talc and chlorite [36] and schematic structure of CMC (right).
alkaline pH range, but in some cases, CMC’s depressive effect is pH-dependent and
gradually diminishes as the pH increases to high levels such as 11 [62].
Alternatively, talc depression can be achieved using guar gum. Rath et al. [63],
reported a greater efficiency with it compared to dextrin. However, some polymeric
varieties of dextrin were closely examined using adsorption isotherms for talc depres-
sion by Beattie et al. [64], who found that their behaviour was quite selective in depress-
ing talc with only a slight influence on the behaviour of the nickel and copper sulphide
minerals.
Lignosulfonates were also found useful for the depression of talc in various sulphide
ores through preferential adsorption [31, 65]. The use of lime as a source of calcium
ions was particularly useful at higher pH levels for more effective adsorption of ligno-
sulfonate onto the talc surface resulting in a stronger depression [65].
As another polysaccharide from natural sources, galactomannan (KGM), has been
reported to be a selective talc depressant in the Ni-Cu Sulphide ores. Zhao et al. [44,
45], found it to be more effective than guar gum and CMC even at the half dosage levels
of these conventional depressants. They attributed its superior performance to the
chemical adsorption between KGM and talc and the lack of physical adsorption between
KGM and pentlandite.
More recent studies on the depression of talc in the flotation of sulphide minerals
involved chitosan. Interestingly, its effectiveness can shift with the manipulation of pH
between 9 and 3 [66].
Another type of gum, known as tragacanth gum, has indicated promising results
[67]. Results show that its depressive impact on talc is higher than that of guar gum at
various dosages. In addition, the depression effect is pH-independent, as in the case of
guar gum.
Most of the depressants touched upon above were investigated on a small scale using
pure minerals, and thus require further studies with industrial ore samples to assess
their potential.
Control of Hydrophilic Gangue Causing Slime Coating Edwards et al. [38] hypoth-
esized that in serpentinic Ni-Cu sulphide ores, poor recovery of pentlandite is mainly
linked to electrostatic mechanisms which involve attraction between positively charged
MgO bearing silicate minerals such as lizardite and pentlandite which is negatively
charged under processing conditions. They indicated that dispersants, such as CMC,
can effectively reduce pentlandite slime coating, improving grades and recoveries. In
the subsequent investigations followed, these effects have been demonstrated in various
Ni-Cu sulphide ores which had serpentine gangue minerals such as chrysolite and liz-
ardite on nickel sulphide ores [34, 35, 68].
The effect of CMC on the pentlandite-lizardite system as reported by Bremmell
et al. [69] is shown in Fig. 8. Lizardite particles are positively charged in the pH range
(3–10) based on the results of zeta potential determinations, while pentlandite was neg-
atively charged above pH 4. As a result of electrostatic interaction between the two,
pentlandite is adversely affected. They also reported that the presence of lizardite did
not affect the xanthate adsorption onto pentlandite and that lizardite did not consume
the collector.
Subsequently, Peng and Bradshaw [70] noted that the separation of ultrafine pent-
landite (i.e., −8 microns) and lizardite is more effective in the highly saline bore water
Fig. 8. The zeta potential of lizardite and pentlandite vs pH (left) and CMC conc. mg/L (right) [69].
electrostatic attraction between the two minerals induced by double layer

compression.
The same concept was tested using sodium hexametaphosphate (SHMP) with a
stronger dispersant action compared to CMC. The SHMP has been found ideal for the
cleaning stage of nickel flotation to increase the concentrate grade [71].
In more recent research work, sodium phytate was tested as a dispersant together
with SHMP for comparison of their effectiveness for MgO control [72], who reported
that the grade-recovery performances for both nickel sulphide and copper sulphide
minerals were better with sodium phytate than SHMP at the same dosage level.
Investigations on the control of MgO level in the Ni-Cu concentrates included non-
sulphide minerals to exploit their surface charge (e.g, quartz, SiO2) as well as the
magnetic properties (magnetite, Fe3O4) that can interact with either serpentine minerals
or nickel-bearing sulphides such as pentlandite to alleviate their undesirable interac-
tions causing poor grade-recoveries [73, 74].
4 Summary
Due to the higher demand for nickel and cobalt needed for sustainable production of
stainless steel and new applications such as car batteries, copper-nickel sulphide ores
have regained importance. In Canada, the major source of these strategic metals is pent-
landite which is intimately associated with large amounts of pyrrhotite. The behaviour
of pyrrhotite in mineral processing operations is critical to continued reductions of SO2
emissions. Some Canadian Ni-Cu sulphide ores also have variable amounts of serpen-
tines and/or talc.
In this paper, trends in demand, production and recycling of Ni/Co have been
reviewed together with recent developments in dealing with processing challenges of
Ni-Cu sulphide ores. The behaviour of pyrrhotite is examined with respect to the main
polymorphic types; hexagonal and monoclinic. Functionalities of reagents such as
DETA/SMBS applicable for depression of pyrrhotite and CMC for depression of talc
together with the dispersion of problematic serpentines such as lizardite and chrysotile
are presented. An overview of new reagents proposed for the control of MgO-bearing
silicates is provided.
Acknowledgements The authors are thankful for continued support and funding from the Natural
Sciences and Engineering Research Council of Canada (NSERC) and for the sponsorship of S. Arslangil
by the general directorate of mineral research and exploration of Turkey.
References
1. Lide DR, editor. CRC handbook of chemistry and physics. 85th ed. Boca Raton: CRC Press; 2004.
2. Dehaine Q, Tijsseling LT, Glass HJ, Törmänen T, Butcher AR. Geometallurgy of cobalt ores: a
review. Miner Eng. 2021;160:106656.
3. Xu C, Dai Q, Gaines L, Hu M, Tukker A, Steubing B. Future material demand for automotive
lithium-based batteries. Commun Mater. 2020;1(1):1–10.
4. Global Energy Metal Corporation. Cobalt demand. 2022. Retrieved March 8, from https://www.
globalenergymetals.com/cobalt/cobalt-demand/.
5. Fraser J, Anderson J, Lazuen J, Lu Y, Heathman O, Brewster N, Bedder J, Masson O. Study on
future demand & supply security of Ni for electric vehicle batteries. Luxembourg: Publications
Office of the European Union (Rosskill report, JRC); 2021.
6. U.S. Geological Survey. Mineral commodity summaries 2022: U.S. Geological Survey. 2022.
https://doi.org/10.3133/mcs2022.
7. Dewulf J, Van der Vorst G, Denturck K, Van Langenhove H, Ghyoot W, Tytgat J, Vandeputte
K. Recycling rechargeable lithium ion batteries: critical analysis of natural resource savings.
Resour Conserv Recycl. 2010;54(4):229–34.
8. Alvial-Hein G, Mahandra H, Ghahreman A. Separation and recovery of cobalt and nickel from end
of life products via solvent extraction technique: a review. J Clean Prod. 2021;297:126592.
9. CEC. Environmentally sound management of end-of-life batteries from electric-drive vehicles in
North America (978–2–89700-110-0). Commission for Environmental Cooperation; 2015. http://
www.cec.org/files/documents/publications/11637-environmentally-sound-management-end-
life-batteries-from-electric-drive-vehicles-en.pdf
10. Harper E, Kavlak G, Graedel T. Tracking the metal of the goblins: cobalt’s cycle of use. Environ
Sci Technol. 2012;46(2):1079–86.
11. Petavratzi E, Gunn G, Kresse C. Commodity review: cobalt. British Geological Survey; 2019.
12. Graedel T, Allwood J, Birat J, Reck B, Sibley S, Sonnemann G, Buchert M, Hagelüken C. UNEP
(2011) Recycling rates of metals-a status report, a report of the Working Group on the Global
Metal Flows to the International Resource Panel, WGotGMF (Ed.). United Nations Environment
Programme; 2011.
13. Roberts S, Gunn G. Cobalt. In: Gunn G, editor. Critical metals handbook. 1st ed. John Wiley &
Sons; 2014. p. 122–49.
14. Alves Dias P, Blagoeva D, Pavel C, Arvanitidis N. Cobalt: demand-supply balances in the transition
to electric mobility. European Commission, Joint Research Centre, EUR-Scientific and Technical
Research Reports Publications Office of the European Union, 10, 97710. 2018.
15. Golmohammadzadeh R, Faraji F, Rashchi F. Recovery of lithium and cobalt from spent lithium
ion batteries (LIBs) using organic acids as leaching reagents: a review. Resour Conserv Recycl.
2018;136:418–35.
16. Boldt JR, Queneau P. The winning of nickel; its geology, mining, and extractive metallurgy.
Toronto: Longmans; 1967.
17. Hawley JE, Stanton RL. The Sudbury ores, their mineralogy and origin; part 2, the facts; the ores,
their minerals, metals and distribution. Can Mineral. 1962;7(1):30–145.
18. Kerr A. An overview of recent developments in flotation technology and plant practice for nickel
ores. In: Mular AL, Halbe DN, Barratt DJ, editors. Mineral processing, plant design, practice and
control proceedings, vol. 1. Littleton: Society for Mining, Metallurgy and Exploration; 2002. p.
1142–58.
19. Naldrett AJ, Singh J, Krstic S, Li C. The mineralogy of the Voisey’s Bay Ni-Cu-Co deposit,
Northern Labrador, Canada: influence of oxidation state on textures and mineral compositions.
Econ Geol. 2000;95(4):889–900. https://doi.org/10.2113/gsecongeo.95.4.889.
20. Kossatz E. Inco boosts productivity using bulk mining methods. Can Min J. 1983;104(12):23–5.
21. XPS. A mineralogical study of the Nickel Rim Ore. Sudbury, Glencore, XPS R&D Centre; 2010.
https://www.xps.ca/services/process-mineralogy#publications, Last accessed 16 Sept 2014
22. Cherevaty M, Lostritto P, Agar GE, Stewart K, Labonté G. Laboratory flotation of Voisey’s Bay
ovoid zone massive sulphide. 30th annual meeting of CMP, Ottawa. 1998.
23. Bouillon DF. Developments in emission control technologies/strategies: a case study, Chapter 21.
In: Gunn JM, editor. Restoration and recovery of an industrial region: smelter damaged landcape
near Sudbury, Canada. New York: Springer; 1995. p. 275–85.
24. Rezaei S, Liu F, Marcuson S, Muinonen M, Lakshmanand VL, Sridhar R, Barati M. Canadian pyr-
rhotite treatment: the history, inventory and potential for tailings. Can Metall Q. 2017;56(4):410–7.
25. Vale. SO2 emissions reductions. 2018. http://www.vale.com/canada/EN/aboutvale/communities/
sudbury/Clean-AER/Pages/default.aspx. Accessed 23 Feb 2022.
26. Ransom C. Environmental management system & community engagement report 2019.
Glencore. 2020. https://www.glencore.ca/.rest/api/v1/documents/2d864503e57e129842a459670
ad6e272/2019%20Annual%20EMT%20Report.pdf.
27. Inco. Annual sustainability report 2005. I. Ltd; 2006.
28. Kelebek S, Fekete SO, Wells GW. Selective depression of pyrrhotite using sulphur dioxide–diethy-
lenetriamine reagent combination. In: Proceedings of the XIX international mineral processing
congress, San Francisco, CA, vol. 3. Littleton: SME; 1995. p. 181–7.
29. Marticorena M, Hill G, Kerr A, Liechti D, Pelland D. INCO develops new pyrrhotite depressant.
Innov Miner Process. 1994;3:15–33.
30. Kelebek S, Tukel C. The effect of sodium metabisulfite and triethylenetetramine system on pent-
landite–pyrrhotite separation. Int J Miner Process. 1999;57:135–52.
31. Kelebek S, Yoruk S, Smith GW. Wetting behaviour of molybdenite and talc in lignosulphonate/
MIBC solutions and their separation by flotation. Sep Sci Technol. 2001;36(2):145–58.
32. Webmineral.com. Mineralogical databae. http://www.webmineral.com/data/Tosudite.shtml#.
Ym24a-1E3DE. Accessed 15 Feb 2022.
33. Gaudin AM. Flotation. New York: McGraw-Hill; 1957.
34. Wellham E, Elber L, Yan D. The role of carboxy methyl cellulose in the flotation of a nickel sul-
phide transition ore. Miner Eng. 1992;5(3–5):381–95.
35. Mani H, Xu M, Quinn P, Stratton-Crawley R. Proceedings of the UBC-McGill Bi-annual interna-
tional symposium on fundamentals of mineral processing, Sudbury, 17–19 Aug 1997.
36. Holmes JA, Langlois P. Process development at the Raglan concentrate. 33rd annual meeting of the
Canadian Mineral Processors (CMP), Ottawa. 2001.
37. Gibson C, Kelebek S. Sensitivity of pentlandite flotation in complex sulfide ores towards pH con-
trol by lime versus soda ash: effect on ore type. Int J Miner Process. 2014;127:44–51.
38. Edwards C, Kipkie W, Agar G. The effect of slime coatings of the serpentine minerals, chrysotile
and lizardite, on pentlandite flotation. Int J Miner Process. 1980;7(1):33–42.
39. Arnold RG. Range in composition and structure of 82 natural terrestrial pyrrhotites. Can Mineral.
1967;9:31–50.
40. Becker M, Bradshaw D, De Villiers J. The mineralogy of pyrrhotite from Sudbury CCN and
Phoenix nickel ores and its effect on flotation performance. Can Metall Q. 2011;50(1):10–9.
41. Multani RS, Waters KE. A review of the physicochemical properties and flotation of pyrrhotite
superstructures (4C-Fe7S8/5C-Fe9S10) in Ni-Cu sulphide mineral processing. Can J Chem Eng.
2018;96:1185–206.
42. Wercholaz CR. Non-magnetic pyrrhotite and its effect on ore processing. BSc thesis, University of
Toronto, Toronto, ON. 2009.
43. Lawson V, Hill G, Kormos L, Marrs G. The separation of pentlandite from chalcopyrite, pyrrhotite
and gangue in nickel projects throughout the world. 12th AusIMM Mill Operators’ Conference,
Townsville, 1–3 Sept 2014.
44. Zhao CH, Wu BZ, Chen JH. Electronic structure and flotation behavior of monoclinic and hexago-
nal pyrrhotite. J Cent South Univ. 2015a;22(2):466–71.
45. Zhao K, Gu G, Wang C, Rao X, Wang X, Xiong X. The effect of a new polysaccharide on the
depression of talc and the flotation of a nickel–copper sulfide ore. Miner Eng. 2015b;77:99–106.
46. Kelebek S, Tukel C. Separation of nickeliferous hexagonal pyrrhotite from pentlandite in Ni-Cu
sulphide ores: recovery by size performance. Miner Eng. 2018;125:223–30.
47. Multani RS, Waters KE. Pyrrhotite depression studies with DETA and SMBS on a Ni-Cu sulphide
ore. Can J Chem Eng. 2019;97(7):2121–30.
48. Tukel C, Kelebek S, Yalcin E. Eh-pH stability diagrams for analysis of polyamine interaction with
chalcopyrite and deactivation of Cu-activated pyrrhotite. Can Metall Q. 2010;49(4):411–8.
49. Qi C, Liu J, Malainey J, Kormos LJ, Coffin J, Deredin C, Liu Q, Fragomeni D. The role of Cu ion
activation and surface oxidation for polymorphic pyrrhotite flotation performance in Strathcona
Mill. Miner Eng. 2019;134:87–9.
50. Kelebek S, Wells PF, Heinrich G. Metal ion characterization of monoclinic pyrrhotite from nickel-
copper ores using EDTA extraction. In: Kemal M, et al., editors. Changing scopes in mineral
processing: proceedings of the 6th International Mineral Processing Symposium. Rotterdam:
A.A. Balkema; 1996a. p. 295–300.
51. Kelebek S, Wells PF, Fekete OS. Differential flotation of chalcopyrite, pentlandite and pyrrhotite
in Ni-Cu sulphide ores. Can Metall Q. 1996b;35(4):329–36.
52. Xu ZH, Rao SR, Finch JA, Kelebek S, Wells P. Role of diethylene triamine (DETA) in pent-
landite-pyrrhotite separation. 1. Complexation of metals with DETA. Trans Inst Min Metall
C. 1997;106:C15–20.
53. Yoon RH, Basilio CI, Marticorena MA, Kerr AN, Stratton-Crawley R. A study of the pyrrhotite
depression mechanism by diethylenetriamine. Miner Eng. 1995;8(7):807–16.
54. Cao J, Qi L, Tian X-D, Jia J-J, Hu X-Q, Luo Y-C, Xu PF. Investigation on the promoted flotation
behavior of pentlandite by ammoniacal copper(II) solution. Colloids Surf A Physicochem Eng
Asp. 2018;546:67–73.
55. Tian X, Furnell E, Bobicki ER. Predicting the fate of diethylenetriamine in pyrrhotite tailings
management. Miner Eng. 2022;76:107335.
56. Yalcin E, Cushing AML, Kelebek S. Solution equilibria and adsorption tendencies of complexed
Cu ions in mining environment. Paper presented at 7th conference on sustainable development of
energy, water and environment systems, July 1–7, Ohrid, Republic of Macedonia. 2012. p. 1–9.
57. Cushing A, Kelebek S, Yue S, Ramsay JA. Biodegradation of CuTETA, an effluent by-product in
mineral processing. Environ Sci Pollut Res. 2018;25:17393–401.
58. Furnell E, Tian X, Bobicki ER. Diethylenetriamine as a selective pyrrhotite depressant: properties,
application, and mitigation strategies. Can J Chem Eng. 2021;99(6):1316–33.
59. Pugh R. Macromolecular organic depressants in sulphide flotation—a review, 1. Principles, types
and applications. Int J Miner Process. 1989;25(1–2):101–30.
60. Liu G, Feng Q, Ou L, Lu Y, Zhang G. Adsorption of polysaccharide onto talc. Miner Eng.
2006;19(2):147–53.
61. Mierczynska-Vasilev A, Beattie DA. Adsorption of tailored carboxymethyl cellulose polymers
on talc and chalcopyrite: correlation between coverage, wettability, and flotation. Miner Eng.
2010;23(11–13):985–93.
62. Feng B, Lu Y, Feng Q, Zhang M, Gu Y. Talc–serpentine interactions and implications for talc
depression. Miner Eng. 2012a;32:68–73.
63. Rath RK, Subramanian S, Laskowski J. Adsorption of dextrin and guar gum onto talc. A compara-
tive study. Langmuir. 1997;13(23):6260–6.
64. Beattie D, Huynh L, Mierczynska-Vasilev A, Myllynen M, Flatt J. Effect of modified dextrins on
the depression of talc and their selectivity in sulphide mineral flotation: adsorption isotherms, AFM
imaging and flotation studies. Can Metall Q. 2007;46(3):349–58.
65. Ma X, Pawlik M. The effect of lignosulfonates on the floatability of talc. Int J Miner Process.
2007;83(1–2):19–27.
66. Feng B, Peng J, Guo W, Zhu X, Huang W. The stimulus response of chitosan and its depression
effect on talc flotation. Miner Process Extr Metall. 2018;127(1):56–61.
67. Wei G, Bo F, Jinxiu P, Wenpu Z, Xianwen Z. Depressant behavior of tragacanth gum and its role in
the flotation separation of chalcopyrite from talc. J Mater Res Technol. 2019;8(1):697–702.
68. Peng Y, Seaman D. The flotation of slime–fine fractions of Mt. Keith pentlandite ore in de-ionised
and saline water. Miner Eng. 2011;24(5):479–81.
69. Bremmell KE, Fornasiero D, Ralston J. Pentlandite–lizardite interactions and implications for their
separation by flotation. Colloids Surf A Physicochem Eng Asp. 2005;252(2–3):207–12.
70. Peng Y, Bradshaw D. Mechanisms for the improved flotation of ultrafine pentlandite and its separa-
tion from lizardite in saline water. Miner Eng. 2012;36:284–90.
71. Feng B, Wang P, Lu Y, Feng Q. Role of sodium hexametaphosphate in flotation of a nickel ore.
Physicochem Probl Miner Process. 2016;52(1):170–81. https://doi.org/10.5277/ppmp160115.
72. Zhao K, Yan W, Wang X, Wang Z, Gao Z, Wang C, He W. Effect of a novel phosphate on the flota-
tion of serpentine-containing copper-nickel sulfide ore. Miner Eng. 2020;150:106276.
73. Feng B, Feng Q, Lu Y. A novel method to limit the detrimental effect of serpentine on the flotation
of pentlandite. Int J Miner Process. 2012b;114:11–3.
74. Lu J, Yuan Z, Liu J, Li L, Zhu S. Effects of magnetite on magnetic coating behavior in pentlandite
and serpentine system. Miner Eng. 2015;72:115–20.
Water Recycling and Seasonal Water Quality Effects
in Mineral Processing
Patrick W. Rankin1 (*), Sadan Kelebek2, Antonio Di Feo1, and Jennifer R. Taylor3
1
CanmetMINING, National Resources Canada, Ottawa, ON, Canada
patrick.rankin@queensu.ca
2
The Robert M. Buchan Department of Mining, Queen’s University, Kingston, ON, Canada
3
CanmetMINING, National Resources Canada, Ottawa, ON, Canada
4
Kidd Operations, Glencore Canada Corporation, Timmins, ON, Canada
Abstract. Water management is a crucial aspect of sustainability in the mining industry.

Increased awareness of the lack of freshwater resources pushes mining companies to use
water from alternate sources and increase the proportion of recycled process water.
Global trends such as decreasing ore grade, climate change, water scarcity, and stricter
environmental regulations have led to the recirculation of greater quantities of water in
mineral processing plants. As a result, the ionic strength of process water increases.
Detrimental or beneficial effects to mineral flotation depend on the sensitivity of ores in
the process. Seasonal variations in temperature, ionic strength, and microbiological activ-
ity can contribute to a decrease in metallurgical performance in concentrators during
certain months of operation. The seasonal performance of the Kidd Creek concentrator
located in Timmins, Ontario is investigated in some detail. The purpose of this paper is to
provide an overview of water chemistry and seasonal variation effects on flotation. The
review will serve as fundamental background knowledge for future research focused on
the effects of water quality variation on a complex Cu-Zn sulphide ore.
Keywords: Flotation · Water recycling · Seasonal variation · Water management ·

Ionic strength · Water quality
1 Introduction
The importance of sustainable water management in mining and mineral processing has
increased in significance due to global trends such as climate change, water scarcity,
stricter environmental regulations, and decreasing ore grades [1–3].
Process water quality is affected by several factors and managing this can have an
impact on process performance (concentrate grade and recovery) and the environment
[4]. Factors that affect water quality in flotation include the feed ore composition, min-
eralogy, degree of oxidation, reagents added during flotation, the degree of water recy-
cling, process water ionic strength, operating conditions, the quality of make-up water
used, seasonal changes, reactions between process water constituents or in tailings
dams, biological growth, or seasonal changes.
https://doi.org/10.1007/978-3-031-17425-4_106
968 P. Rankin et al.
Water quality changes can have a detrimental or beneficial effect on flotation perfor-
mance, specifically concentrate grade and metal recovery [5, 6]. The effect on flotation
depends on process water constituents created or transferred from the water quality
changes described above [7]. Inorganic compounds and organic constituents can
increase the complexity of the flotation water and affect flotation results [5, 8]. Water
quality can change very quickly, sometimes in a matter of hours, and these changes are
often undetected [8]. Different inorganic ions can have vastly different effects on a
process and by extension, the ionic strength of the process water is an important param-
eter of study [7].
Two methods used to reduce the quantity of freshwater used in mineral processing
include process water recirculation and substitution of freshwater with alternative
sources (known as make-up water) [3]. Seawater and other saline water sources have
been used in several mineral processing operations worldwide including Las Luces in
Chile and Raglan Mine in Canada [9]. Ore is processed in Raglan operations without
the use of frother during flotation due to the high salinity level of local water. It was
shown that salt solutions with high ionic strength (0.4 M) achieved similar froth proper-
ties during flotation compared to using 10 ppm methyl isobutyl carbinol (MIBC) [10].
Recycling process water or using an alternative water source for flotation can result in
an increase in the concentration of process constituents [8]. Negative impacts of using
water with low quality include corrosion and scale formation, as well as deleterious
mineral separation resulting from the loss of concentrate recovery and/or grade.
1.1 Importance of Water as a Variable in Flotation Systems
Flotation has a multivariable nature, and it is the main mineral separation process used
in the mining industry [11]. The type and number of variables in flotation have been
categorized based on three principal aspects: chemistry, equipment, and operational
factors, which are commonly depicted in the corners of a triangle [12]. Water is not
explicitly mentioned as a variable within this interactive system. However, water is the
medium for flotation chemistry while at the same time it is a raw material used in opera-
tions. Water deserves greater consideration given the growing importance of sustain-
ability in mining and mineral processing. Accordingly, an additional triangle has been
proposed where water is highlighted as a separate parameter affecting flotation systems
[13]. Figure 1 shows the updated version of this model with more detail on the types of
variables characterizing the flotation system and potential interrelationships among
variables. Water is recognized as a cornerstone phase, where interactions between min-
erals and reagents are enabled and affected by the composition in the flotation slurry
system which is also affected by temperature.
The importance of water and the impact of seasonal changes on water chemistry is
clear, although mechanisms affecting the metallurgical performance are often poorly
understood.
Water Recycling and Seasonal Water Quality Effects in Mineral Processing 969
Fig. 1. Flotation system variables and interrelationships [13].
1.2 Objectives
This paper will review the effects of water quality, recycling, and seasonal trends on
flotation. The purpose of this paper is to provide an overview of water chemistries that
are coupled with seasonal changes towards the development of a fundamental back-
ground for a study on the effects of variation of water quality on a complex Cu-Zn
sulphide ore.
2 Water Quality Effects on Flotation
Types and concentrations of dissolved ions and species determine the quality of water.
The effect of ionic strength on flotation depends on the ore type sensitivity and concen-
trator [14]. The process may experience a detrimental or beneficial effect, which can be
due to several factors such as a change in froth characteristics, inadvertent activation of
certain minerals or gangue, precipitation of colloidal particles, the interaction of pro-
cess constituents, or a change in flotation kinetics [8, 15].
Water quality is affected by recycling in flotation (or degree of recycling), seasonal
variation, changes in process make-up water, flotation conditions, ore mineralogy/oxi-
dation level, reactions between organic or inorganic constituents during flotation, reac-
tions in tailings dams, and biological growth [3, 4, 15]. Various water uses and recycling
in mining and mineral processing are shown in Fig. 2.
Fig. 2. Water flows and recycle in mining/mineral processing operations [16].
As is evident, the use of water in mining is integral to operations. The water chem-
istry of the streams used and recycled is important to the concentrator’s performance
and environmental impact.
2.1 Ionic Constituent Effect
It has been shown in various studies that the effect of individual ions can be either ben-
eficial or detrimental to flotation [5]. Generally, for monometallic flotation, the effects
of recycled water are negligible or favourable. However, for polymetallic and complex
ores, the effects of recycled water on metallurgical performance can be detrimental.
Many studies have shown a link between ionic strength and froth stability [17–19].
Researchers have studied the effect of ionic strength on the flotation of platinum group
metal (PGM) ore from the Merensky reef [17]. They found that froth stability increases
as ionic strength increases which causes greater water and solid recovery. The improved
stability results in greater metal recovery and lower metal grade.
Saline or seawater is used in many processing operations for flotation and there have
been several studies investigating the use of saline water as a flotation medium [20]. For
instance, Peng et al. (2012) investigated the effects of using bore water (TDS = 77 g/L)
for fine particle copper and nickel flotation [21]. Results indicate that using bore water
increases metal recovery and reduces gangue entrainment due to particle screening by
electrolytes in the solution. Mu and Peng (2019) also reported on the effects of using
saline water in chalcopyrite and pyrite flotation [22]. Seawater use in flotation signifi-
cantly increases pyrite recovery and reduces selectivity due to copper activation on
pyrite surfaces. The increased ionic strength causes chalcopyrite oxidation and dissolu-
tion and results in a greater concentration of copper ions in the solution. Drelich and
Miller (2012) studied the kinetics of air bubble attachment of chalcopyrite, bornite, and
gold using seawater and other solutions [23]. Synthetic seawater was found to have a
positive effect on air bubble attachment kinetics compared to solutions with lower ionic
strength. For chalcopyrite and gold, the positive effect on bubble attachment was fur-
ther increased in the presence of potassium xanthate.
Threshold concentrations at which the flotation response is dictated by a specific ion
concentration have been reported by many authors [7, 18, 19]. The flotation perfor-
mance is positively affected below a threshold concentration and negatively affected
above the threshold concentration. Specific ions present in process water have a greater
effect on flotation response than the total ionic strength of the water [24]. Polyvalent
cations and anions with large hydration layers have a positive flotation response whereas
ions with small hydration layers have a negative response [7]. Ions present in process
water and saline water such as sulfate, sodium, calcium, base metals, chloride, and
thiosulphates, have threshold concentrations and these can be important parameters
when determining suitable make-up water sources and recycle percentage in flotation
practices [8].
Dzingai et al. (2020, 2021) investigated the effects of various anions and cations on
PGM ore flotation by spiking synthetic process water [18, 19]. Similar to previous
results from this research group, increasing the ionic strength causes an increase in
froth stability and positive flotation effects. Ion threshold concentrations were found to
be 400 mg/L, 720–1200 mg/L, and 528–880 mg/L for calcium, sulphate, and nitrate
ions, respectively. The threshold concentrations for magnesium and thiosulphate were
theorized to be above the range of the study at 700 mg/L and 78 mg/L, respectively.
Collectorless flotation recovery and flotation rate for sphalerite increase with an
increase in copper ion concentration to a critical concentration due to copper activation
[25]. With additional copper addition after 1 × 10−5 M (or 0.95 mg/L), the flotation rate
and recovery of sphalerite decrease. Negative effects on flotation performance are due
to copper hydroxide precipitation, which covers hydrophobic mineral surface sites and
inhibits flotation.
Dissolved salts in a suspension can alter the electrokinetic behaviour of mineral
particles [26]. Precipitation at high pH and adsorption of cations at low pH onto mineral
particle surfaces can cause a charge reversal. Coupled with solution equilibria dia-
grams, it is possible to predict the causes for a change in electrokinetic behaviour.
Changes in surface charge can occur in solutions with dissolved inorganic ions due to
surface reactions or bulk precipitation.
Literature research was conducted to compile sources that reported water quality in
flotation. The ionic strength of each source was calculated and will be detailed in a
subsequent publication. The summary statistics are shown in Table 1.
The average ionic strength value (0.25 M) is largely influenced by an outlier value
(hypersaline water with ionic strength of 8.9 M) which is much higher than any other
value found in the literature.
Ionic organic species affect the electrokinetic behaviour of particles. Biosolids gen-
erally shift zeta potential to negative values for chalcopyrite and galena [27]. The nega-
tive zeta potential shift is more pronounced when humic acid is present in the solution
Table 1. Summary statistics from ionic strength literature search.

Variable in literature n Mean Median Mode Minimum Maximum
Ionic strength (M) 93 0.25 0.064 0.24 5.7 × 10−5 8.9
Note: “n” refers to the quantity of water quality data cited in flotation studies
compared to other biosolids. Increasing concentrations of humic acid in the solution

caused the zeta potential to decrease.
The effect of humic acid, stearic acid, and tannic acid (commonly found in rivers) on
chalcopyrite and molybdenite flotation was investigated [28]. These compounds had a
negative effect on copper and molybdenum grade, and humic acid was the most detri-
mental to flotation, followed by tannic acid and stearic acid.
2.2 Organic Matter
Organic compounds present in flotation water can include microorganisms (including

compounds from metabolic processes), reagents, compounds from wastewater, and
organic compounds from make-up water [5, 8]. Freshwater resources have been
replaced with wastewater effluent in some mining operations [3]. Additionally, make-
up water can contain high levels of bacteria if untreated sewage is discharged to surface
waters near mine sites.
Some studies have reported a decrease in flotation efficiency due to biological matter
[29, 30], however, this phenomenon is not as widely studied as inorganic constituent
effects. The flotation efficiency of chalcopyrite and gold is negatively affected as the
concentration of model bacteria cells (E. coli) increases [29]. The mechanism for flota-
tion depression was attributed to bacteria adhering to mineral surfaces and creating a
hydrophilic layer that impeded collector adsorption. The extracted DNA concentration
in process water from a Sudbury concentrator shows a clear seasonal trend [6]. A study
conducted by Anglo Platinum compared the use of Merensky Waterval process water
and treated sewage effluent from the Rustenburg Municipality in platinum ore flotation
[30]. A decrease in overall platinum recovery of 2% was observed using treated sewage
relative to process water. The decrease in recovery could have been due to a relatively
higher bacterial count in the treated sewage water, or possibly the higher ionic strength
of the process water.
2.3 Seasonal Effects
Several zinc concentrators located in Canada and Sweden have experienced a seasonal
variation in zinc recovery, with lower recovery in the winter compared to the summer
months [31]. The effect of temperature on flotation is complex, with some gaps in the
available literature. Temperature changes can affect flotation rate, kinetics, mass trans-
fer rate, adsorption selectivity, hydrodynamic properties, reagent dissolution, surface
tension, wetting angle, dissolved oxygen concentration, and interfacial chemistry, all of
which are interdependent and can affect flotation performance [31].
A notable seasonal effect in the zinc concentrate grade produced at the Matagami
concentrator (formerly operated by Noranda Inc., now Glencore) occurs in winter
(December–April) compared to the summer (July–August) [32]. Pulp temperatures
range from 5 °C in the winter to 30 °C in the summer. Using a novel gas dispersion
technology, it was found that the optimal gas flow conditions were distinct, and that
bubble properties were markedly different depending on the season. For instance, bub-
ble size decreased in the winter compared to the summer, causing an increase in surface
area flux and a decrease in zinc grade. Interestingly, the zinc recovery between seasons
was unchanged. It was found that the zinc grade was inversely proportional to the sur-
face area attributed to the fraction of bubbles smaller than 1 mm.
3 Case Study: Kidd Creek Concentrator
Kidd Creek is a Cu-Zn-Ag mine in Timmins, ON, Canada. The main valuable minerals
are chalcopyrite and sphalerite together with a minor amount of galena and a significant
quantity of silver. There are considerable amounts of pyrite as a sulphide gangue in the
ore, and the main non-sulphide gangue mineral is quartz. Figure 3 shows the B Division
grinding and Cu flotation circuit simplified from Taylor et al. [33].
The ore is ground in two stages comprising a rod mill and ball mill combination
producing a cyclone overflow with a P80 of 100 microns, which is further ground in a
secondary ball mill circuit to a P80 of 44 microns. The reagents used in the copper circuit
are lime (added to the rod mill to obtain a pH of 10.5 in the copper circuit), 3418A,
R208 as collectors, and MIBC as a frother. More details are reported elsewhere [33].
The concentrator primarily uses two water resources for processing operations: pro-
cess water, which is recycled in the plant, and raw water from a local river as fresh.
Batch flotation of Kidd Creek ore using real process water has been studied previ-
ously [34]. It was found that recycled water was beneficial to flotation performance
compared to tap water due to elevated concentrations of thiosalt and calcium ions.
Fig. 3. Kidd Creek B Division of the grinding & Cu flotation circuits adapted from Taylor et al. [33].
Fig. 4. Monthly copper recovery in the copper concentrate in the Kidd Creek concentrator from 2017
to 2020 [38].
3.1 Seasonal Effects Impacting Metallurgical Performance in Flotation
The beneficial effect of elevated temperatures on flotation performance has been known
for many years [35]. High temperatures were also found to be critical for the reverse
flotation of pyrite in upgrading the zinc concentrate at Kidd Creek [36]. Seasonal effects
on flotation have also been observed in the upgrading of copper concentrate in the
Sudbury region [37]. The rejection of the nickel impurity from the copper concentrate
was significantly better in the summer months than in the winter months. The monthly
copper recovery between 2017 and 2020 at Kidd Creek is shown in Fig. 4 [38].
It is notable that during the monthly transitions from winter to spring (e.g., January
to March), copper recovery is adversely affected. This is undesirable since it will lead
to copper losses to the zinc concentrate and tailings. The cumulative sum (cusum) plot
of the plant data is shown in Fig. 5. The target copper recovery is the average copper
recovery during operation between 2017 to 2020 (Fig. 4).
When the difference between the observed copper recovery and the target copper
recovery is negative, this implies that the copper is below the expected value. This is not
desirable since a considerable mass of copper is lost. What is evident from Fig. 5 is that
the difference between the copper recovery and target copper recovery decreases
(becomes more negative) starting in January to April every year.
4 Conclusion
In this paper, the water quality and seasonal effects of process water are discussed with
respect to flotation effects and solution chemistry. Ionic strength can have detrimental
or beneficial effects on flotation depending on the operation of the study. A case study
highlighting the seasonal effects experienced in a Cu-Zn complex sulphide ore concen-
trator is discussed in some detail.
Fig. 5. Cusum plot for monthly copper recovery in the copper concentrate in the Kidd Creek concen-
trator from 2017 to 2020.
Future work in this field will include batch flotation testing complemented by statis-
tical analysis to study the effects of variations in water quality on the flotation of a
Cu-Zn-Ag ore from the Kidd Creek concentrator. Surface analysis testing will be done
to determine which water constituents are predominantly affecting flotation. Finally,
water treatment options will be investigated for the process water.
Acknowledgements Thanks to Lucie Morin, André Demers, Meghan De Souza, Samir Samhat, and
Maziar Sauber for the valuable support to this project. The Kidd Operations management is also
thanked for their in-kind support and sharing of plant data used in this work.
References
1. Prosser I, Wolf L, Littleboy A. Water in mining and industry. In: Water: science and solutions for
Australia. Collingwood: CSIRO; 2011. p. 135–46.
2. Aitken D, Rivera D, Godoy-Faúndez A, Holzapfel E. Water scarcity and the impact of the mining
and agricultural sectors in Chile. Sustainability. 2016;8(2):128. https://doi.org/10.3390/su8020128.
3. Liu W, Moran CJ, Vink S. A review of the effect of water quality on flotation. Miner Eng.
2013;53:91–100. https://doi.org/10.1016/j.mineng.2013.07.011.
4. Levay G, Smart RSC, Skinner WM. The impact of water quality on flotation performance. J South
Afr Inst Min Metall. 2001;101(69–75):8.
5. Rao SR, Finch JA. A review of water re-use in flotation. Miner Eng. 1989;2(1):65–85. https://doi.
org/10.1016/0892-6875(89)90066-6.
6. Di Feo A, Mortazavi S, Langley S, Morin L, Prabhakar G, Demers A, Bedard I, Volchek K. The

effects of water recycling on flotation at a North American concentrator – part 1. J Miner Mater
Charact Eng. 2020;8(4):240–76.
7. Pugh RJ, Weissenborn P, Paulson O. Flotation in inorganic electrolytes; the relationship between
recover of hydrophobic particles, surface tension, bubble coalescence and gas solubility. Int J
Miner Process. 1997;51(1–4):125–38. https://doi.org/10.1016/S0301-7516(97)00021-5.
8. Levay G, Schumann R. A systematic approach to water quality management in the minerals pro-
cessing industry. Water in Mining Conference, Brisbane; 2006. p. 277–89.
9. Wang B, Peng Y. The effect of saline water on mineral flotation – a critical review. Miner Eng.
2014;66–68:13–24. https://doi.org/10.1016/j.mineng.2014.04.017.
10. Quinn JJ, Kracht W, Gomez CO, Gagnon C, Finch JA. Comparing the effect of salts and frother
(MIBC) on gas dispersion and froth properties. Miner Eng. 2007;20(14):1296–302. https://doi.
org/10.1016/j.mineng.2007.07.007.
11. Rao SR, Leja J. Surface chemistry of froth flotation, Volume 1: Fundamentals. 2nd ed. New York:
Springer Science+Business Media, LLC; 2004.
12. Klimpel RR, Dhansen R, Fee BS. Chapter 26: Selection of flotation reagents for mineral flotation.
In: Mular AL, Anderson MA, editors. Design and installation of concentration and dewatering
circuits. Littleton: Society of Mining Engineers of AIME; 1986. p. 384–404.
13. Kelebek S. Internal report, Private Communication (2022) The Robert M. Buchan Department of
Mining, Queens University, Kingston ON, Canada, unpublished. 2019. p. 5.
14. Di Feo A, Hill-Svehla CM, Hart BR, Volchek K, Morin L, Demers A. The effects of water recy-
cling on flotation at a North American concentrator. Miner Eng. 2021;170:107037. https://doi.
org/10.1016/j.mineng.2021.107037.
15. Rao F, Lázaro I, Ibarra LA. Solution chemistry of sulphide mineral flotation in recycled water
and sea water: a review. Miner Process Ext Metall. 2017;126(3):139–45. https://doi.org/10.108
0/03719553.2016.1192794.
16. Government of Canada. Environmental code of practice for metal mines: mine life
cycle activities. 23 Mar 2017. Available: https://www.ec.gc.ca/lcpe-epa/default.
asp?lang=En&n=CBE3CD59-1&offset=4. Accessed 2020/3/9.
17. Manono MS, Corin KC, Wiese JG. An investigation into the effect of various ions and their ionic
strength on the flotation performance of a platinum bearing ore from the Merensky reef. Miner
Eng. 2012;36–38:231–6. https://doi.org/10.1016/j.mineng.2012.03.035.
18. Dzingai M, Manono M, Corin K. Simulating the effect of water recirculation on flotation through
ion-spiking: effect of Ca2+ and Mg2+. Fortschr Mineral. 2020;10(11):1033. https://doi.org/10.3390/
min10111033.
19. Dzingai M, Manono MS, Corin KC. Probing the effect of water recycling on flotation through
anion spiking using a low-grade Cu–Ni–PGM ore: the effect of NO3−, SO42− and S2O32−.
Fortschr Mineral. 2021;11(4):340. https://doi.org/10.3390/min11040340.
20. Gálvez ED, Cisternas LA. Innovative solutions for seawater use in mining operations. In: Moya
BL, de Gracia MDS, Mazadiego LF, editors. Case study of innovative projects – successful real
cases. InTech; 2017. https://doi.org/10.5772/intechopen.68191.
21. Peng Y, Zhao S, Bradshaw D. Role of saline water in the selective flotation of fine particles. In:
Proceedings of the first international symposium. Englewood: Society for Mining Metallurgy and
Exploration; 2012. p. 61–71.
22. Mu Y, Peng Y. The effect of saline water on copper activation of pyrite in chalcopyrite flotation.
Miner Eng. 2019;131:336–41. https://doi.org/10.1016/j.mineng.2018.11.032.
23. Drelich J, Miller JD. Induction time measurements for air bubbles on chalcopyrite, bornite, and
gold in seawater. In: Proceedings of the first international symposium. Englewood: Society for
Mining, Metallurgy, and Exploration; 2012. p. 73–85.
24. Shackleton NJ, Malysiak V, Vaux D, Plint N. Effect of cations, anions, and ionic strength on the
flotation of pentlandite-pyroxene mixtures. In: Proceedings of the first international symposium.
Englewood: Society for Mining, Metallurgy, and Exploration; 2012. p. 197–210.
25. Albrecht TWJ, Addai-Mensah J, Fornasiero D. Critical copper concentration in sphalerite flotation:
effect of temperature and collector. Int J Miner Process. 2016;146:15–22. https://doi.org/10.1016/j.
minpro.2015.11.010.
26. Acar S, Somasundaran P. Effect of dissolved mineral species on the electrokinetic behaviour of
sulfides. Miner Eng. 1992;5(1):27–40. https://doi.org/10.1016/0892-6875(92)90004-S.
27. Reyes-Bozo L, Escudey M, Vyhmeister E, Higueras P, Godoy-Faúndez A, Salazar JL, Valdés-
González H, Wolf-Sepúlveda G, Herrera-Urbina R. Adsorption of biosolids and their main com-
ponents on chalcopyrite, molybdenite and pyrite: zeta potential and FTIR spectroscopy studies.
Miner Eng. 2015;78:128–35. https://doi.org/10.1016/j.mineng.2015.04.021.
28. Sinche-Gonzalez M, Fornasiero D, Zanin M. Flotation of chalcopyrite and molybdenite in the pres-
ence of organics in water. Fortschr Mineral. 2016;6(4):105. https://doi.org/10.3390/min6040105.
29. Liu W, Moran CJ, Vink S. The adverse effect of disrupted water-borne bacteria cells on flotation.
Int J Miner Process. 2016;157:128–33. https://doi.org/10.1016/j.minpro.2016.10.007.
30. Slatter KA, Plint ND, Cole M, Dilsook V, De Vaux D, Palm N, Oostendorp B. Water management
in anglo platinum process operations: effects of water quality on process operations. In: Abstracts
of the international mine water conference, Pretoria, South Africa. 2009. p. 46–55.
31. Lin IJ. The effect of seasonal variations in temperature on the performance of mineral processing
plants. Miner Eng. 1989;2(1):47–54. https://doi.org/10.1016/0892-6875(89)90064-2.
32. Nesset JE, Gomez CO, Finch JA, Hernandez-Aguilar J, Di Feo A. The use of gas dispersion mea-
surements to improve flotation performance. In: Proceedings of the Canadian mineral processors
2005, Ottawa, Canada. 2005. p. 401–421.
33. Taylor J, White S, Leggett T. Glencore-the Kidd concentrator. In: Bobicki ER, Gibson CE,
Boulanger JF, editors. Canadian milling practice. Montreal: Canadian Institute of Mining,
Metallurgy and Petroleum; 2020.
34. Liu L, Rao SR, Finch JA. Technical note laboratory study of effect of recycle water on flotation of a Cu/
Zn sulphide ore. Miner Eng. 1993;6(11):1183–90. https://doi.org/10.1016/0892-6875(93)90095-5.
35. Hukki RT. Hot flotation improves selectivity and raises mineral recoveries. World Min.
1973;March:101–18.
36. Kızılırmaklı AV. Application of reverse pyrite flotation at Kidd Creek Mines Ltd. In: Processing
of complex ores, Amsterdam. 1989. p. 215–226. https://doi.org/10.1016/B978-0-08-037283-
9.50024-X.
37. Kelebek S, Beauchamp JB, Murphy DA, Marrs G, Macnamara DD. Column practice in Ni-Cu
separation at Falconbridge. In: Column ’96: proceedings of the international symposium on col-
umn flotation, Montréal, Canada. 1996. p. 119–134.
38. Taylor JR, et al. Internal report, Kidd Creek concentrator, Timmins, ON, Canada, 2017–2020.
Production of Spodumene Concentrate
from the North Carolina Piedmont Lithium Project
Brian Kawenski Cook1 (*), Massoud Aghamirian1, Hao Li1, Nima Nabiri1,
Jim Nottingham3, Jarrett Quinn4, Chris Gunning1, Charlotte E. Gibson2,
and Patrick Brindle3
1
SGS Canada Inc., Lakefield, ON, Canada1
brian.cook@sgs.com; 9bc25@queensu.ca
2
3
Piedmont Lithium Inc., Belmont, NC, USA
4
Primero Group Americas Inc., Montreal, QC, Canada
5
Queen’s University, Robert M. Buchan School of Mining, Kingston, ON, Canada
Abstract. The global push toward sustainable energy has created a surge in demand for
electric vehicles powered by lithium-ion batteries. To meet projections, the mining indus-
try is rapidly expanding to develop new lithium mineral projects focused on producing
high-quality lithium concentrates. The Piedmont Lithium Project in North Carolina is
located along the Carolina Tin-Spodumene Belt and contains high-grade lithium pegma-
tites well-suited for producing spodumene concentrate. The study investigated chemical-
grade spodumene concentrate production using dense media separation (DMS), magnetic
separation, and froth flotation. Three composite samples were selected with varying ratios
of pegmatite to host rock and different mineralogical compositions. A two-stage DMS
circuit processed the −6.4/+3.3 mm and −3.3/+1.0 mm fractions of each sample and
achieved 24–37% with lithium grades ranging from 6.0% to 6.4% Li2O. The flotation
feed was comprised of DMS middlings and the −1.0 mm fines fraction which were stage-
ground to −300 μm. Magnetic separation was used to reject iron silicate minerals prior to
flotation. Flotation conditions were optimized in a series of bench-scale flotation tests.
Locked-cycle tests were conducted under optimized flotation conditions and resulted in
global lithium recoveries ranging from 37% to 53% with lithium grades between 5.7%
and 6.4% Li2O. The combined DMS and flotation flowsheet produced lithium recoveries
of 74%, 77%, and 80% at combined concentrate grades of 5.8%, 6.4%, and 6.0% Li2O,
respectively.
Keywords: Spodumene concentrate · Flotation · Dense media separation
1 Introduction
The growing desire for electric vehicles powered by lithium-ion batteries has created a
projected lithium demand that will be difficult to meet given the current state of global
lithium production. Lithium is found in three major sources: pegmatite mineral depos-
its, brine deposits, and clays, but is currently only produced from minerals and brines.
Lithium-bearing pegmatites are an increasingly important resource, for example, in
https://doi.org/10.1007/978-3-031-17425-4_107
980 B. K. Cook et al.
2021 mineral operations in Australia accounted for about 55% of the world’s lithium
production, up from under 50% in 2020 [1, 2]. While lithium-bearing pegmatites are
widespread, they must have a high enough lithium content for economic extraction.
Pegmatite deposits can contain several lithium minerals like spodumene, petalite, lepid-
olite, amblygonite, and zinnwaldite. Of these, spodumene (LiAl(Si2O6)) is the most
important lithium-bearing mineral due to its high Li2O content, up to 8% depending on
crystal impurities [3]. To ensure economic lithium extraction, spodumene concentrates
typically target >6.0% Li2O, but this can be challenging due to similarities in specific
gravity and physicochemical properties of spodumene and gangue minerals.
In recent years, there has been significant exploration and development of spodu-
mene deposits in North America. In the United States, the North Carolina Piedmont
Lithium Project is of significant interest as the deposit is well-suited to the production
of spodumene concentrates. The project lies within the Carolina Tin-Spodumene belt, a
25-mile long and 2-mile wide lithium-rich region that once supplied most of the world’s
lithium. The vision of this project is to produce spodumene concentrates for conversion
to battery-grade lithium hydroxide [4].
Spodumene concentration is a complex process involving dense media separation
(DMS), flotation, and magnetic separation. DMS typically processes coarse material
(>850 μm) and can produce a coarse final spodumene concentrate while often also act-
ing as an upstream upgrading stage for the flotation feed. This stage serves to remove
gangue minerals with a lower specific gravity (S.G.) than spodumene (S.G. 3.1–3.2),
like quartz and feldspar. If desired, two cut-points may be used to produce a tailings
stream, a middlings stream, and a saleable spodumene concentrate. The DMS mid-
dlings and fine fraction (commonly >850 μm) are often combined as the feed for flota-
tion. Not only does this stage produce final spodumene concentrate, but it also serves to
significantly reduce the footprint of the flotation circuit as a large portion of the mass is
rejected to the DMS tailings [5].
Spodumene flotation circuits will often use upfront magnetic separation and gangue
flotation to remove magnetic susceptible minerals and mica-type minerals, respectively.
The spodumene flotation stage uses fatty acid collectors, known for being strong collec-
tors but with poor selectivity. However, selective flotation of spodumene can be
improved through desliming, a proper conditioning stage, and using the optimum col-
lector dosage. The presence of slimes is detrimental to spodumene flotation with fatty
acids as they are susceptible to hydraulic entrainment and most likely collector adsorp-
tion. It is important to minimize fines production as they can negatively impact flotation
performance and imply higher lithium losses during the desliming stages.
In addition to Li2O grade, the iron content of the spodumene concentrate is one of
the major specifications for downstream lithium extraction from spodumene
concentrates, typically targeting under 1% Fe2O3. Iron is mainly present in lithium
deposits as iron-silicate minerals like amphiboles, but also as biotite and iron substitu-
tions in the spodumene crystal matrix. The presence of iron-silicate minerals may have
some negative effects during the downstream calcining process for spodumene phase
transformation, but iron existing in the spodumene crystal structure as the solid solution
has no effect on the downstream hydrometallurgy operations. The most effective meth-
ods for rejecting iron-bearing minerals are ore sorting and magnetic separation.
Magnetic separation can be applied to the final DMS and flotation concentrates, as well
Production of Spodumene Concentrate from the North Carolina Piedmont Lithium Project 981
as on the flotation feed. Unfortunately, the presence of iron in the crystal structure can
make spodumene weakly prone to magnetic separation, thus it is important to maximize
iron rejection and minimize lithium losses. This can be achieved using staged magnetic
separation with gradually increasing intensities [6].
This paper investigates the production of a final spodumene concentrate for the con-
version to battery-grade lithium hydroxide with material from the Piedmont Lithium
Project. The discussion will focus on testwork with three primary ore composite sam-
ples, all testwork and analyses were conducted at SGS in Lakefield. Spodumene con-
centration was produced using DMS, magnetic separation, and flotation. Optimized
flotation conditions were developed through bench flotation testwork and validated
with locked-cycle testing.
A total of nine composite samples were evaluated as part of a larger testwork program,
but the discussion of this paper will only focus on the three primary samples. The com-
posite samples came from drill core intervals that were selected to capture realistic
dilutions of waste rock to pegmatite. It should be noted that these samples contained
higher than average dilution levels to better evaluate the impact of waste rock on spodu-
mene concentrate production. For the purposes of this paper, these composites are
referred to as COMP A, B, and C. A simplified flowsheet of the spodumene beneficia-
tion testwork is presented in Fig. 1. The DMS testwork comprised pilot-scale testing to
produce spodumene concentrate and to prepare the material used as the feed to the
flotation circuit.
The material was crushed to 100% passing 6.4 mm and classified into three distinct
size fractions: −6.4/+3.3 mm, −3.3/+1.0 mm, and −1.0 mm. The two coarse fractions
were fed to DMS and the fines were reserved for flotation. DMS was conducted on each
sample and size fraction using a Multotec Dense Media Cyclone in two stages. The
dense media used was a combination of ferrosilicon and magnetite in water. DMS on
all three composite samples was performed with a first pass to reject silicate gangue as
a “floats” product at a cut-point S.G. of 2.65. This was followed by a second DMS pass
of the 2.65 “sinks” product at S.G. values between 2.85 and 2.95, depending on the
composite sample, initial heavy liquid test results, and size fraction. The sinks from the
second stage were then passed through a dry Eriez Laboratory Magnetic Roll Separator
to reduce iron and produce a final DMS spodumene concentrate.
The DMS middlings and −1.0 mm fines were recombined, homogenized and split
into charges. Subsamples of about 40 kg were taken from each composite sample and
stage ground to 100% passing 300 μm for bench and locked-cycle testing (LCT). The
ground material was then homogenized and processed through a wet Eriez Model LWD
Medium Intensity Magnetic Separation (MIMS) at 5000 gauss. The non-magnetic
MIMS products were combined and homogenized, then split into representative 2-kg
charges for flotation tests. Prior to each flotation test, charges were passed through
Eriez Model L-4-20 Wet High-Intensity Magnetic Separation (WHIMS) at consecu-
tively higher intensities of 5000, 10,000, and 13,000 gauss. The non-magnetic products
Crushing Stage Grinding
Size < 1 mm fines Magnetic Magnetic

Classification (to flotation) Separation Product
(MIMS + WHIMS) (to tailings)
DMS 1st Pass, Slimes

Desliming (to tailings)
DMS Floats Silicate Gangue
(to tailings) Rejection
Mica Mica
Flotation Concentrate
DMS Middlings
DMS 2nd Pass (to flotation)
Scrubbing & Slimes
Desliming (to tailings)
DMS Sinks
Dry Magnetic
Mag.
Separation High Density
(to tailings)
Conditioning
Spodumene DMS Concentrate

(DMS Sinks Non-Magnetic Spodumene Spodumene
Flotation Float Tailings
Product)
Spodumene Flotation
Concentrate
Fig. 1. Simplified DMS and flotation flowsheet used in the testwork for spodumene upgrading.
were then deslimed to remove −15 μm particles prior to mica flotation, which was
performed with a cationic amine collector at alkaline pH. Next, the mica tailings were
scrubbed with sodium hydroxide and lignin sulphonate before a second desliming stage
and pulp washing. After scrubbing and desliming, the material was conditioned at high
density with a fatty acid collector. Bench scale and locked-cycle flotation consisted of
two stages of spodumene rougher-scavenger flotation and two stages of spodumene
cleaner flotation. All flotation stages were conducted at a pulp pH of 8.5.
The final spodumene concentrates and other test products from DMS, flotation, and
magnetic separation work were filtered and dried prior to taking representative sub-
samples that were then submitted for chemical analysis. Whole Rock analysis was
determined by X-ray Fluorescence (XRF) and lithium was analyzed through sodium
peroxide fusion and Inductively Coupled Plasma (ICP-OES). The metallurgical perfor-
mance of each stage will be discussed throughout this paper and a final section will
review the combined global performance for each sample combining the DMS and
flotation testwork.
3 Characterization of the Head Sample
Subsamples of each of the three primary composite samples were submitted for chemi-
cal analysis and semi-quantitative X-ray Diffraction (XRD). QEMSCAN analysis was
performed on the flotation feed after MIMS and WHIMS magnetic separation. The
head assays for each sample are presented in Table 1.
Table 1. Head assays of the three primary composite samples.

Assays (%)
Sample Li Li2O SiO2 Al2O3 K2O Na2O CaO P2O5 Fe2O3
COMP A 0.52 1.12 70.2 16.5 2.51 3.46 0.95 0.30 1.92
COMP B 0.49 1.05 71.2 16.4 2.53 3.71 1.13 0.36 1.40
COMP C 0.49 1.05 69.9 16.6 2.49 3.70 1.26 0.33 2.01
3.1 XRD Analysis of DMS Feed
The XRD results for the DMS feed of the three primary composites are presented in
Fig. 2. The major gangue minerals in each sample were plagioclase, at about 28–40%
mass, quartz at about 23–31%, followed by 7–12% microcline (K-feldspar). The mica-
type minerals, muscovite and biotite were present in amounts between 5.7% to 10.3%
and 1.7% to 5.4%, respectively. Muscovite content was highest in COMP A and about
equal in COMP B and C. Meanwhile biotite content was lowest in COMP B, but about
equal in COMP A and C. The Fe-Al silicate gangue, or amphibole, the content was
determined to be between 1.2% and 1.6% with COMP A having the highest content and
COMP C the lowest.
The lithium-bearing minerals in the composite samples included spodumene from
8.7% to 12.2% mass, holmquistite from 0.5% to 2.8%, and minor amounts of petalite
between 0.2% and 1.4%. COMP C had the lowest spodumene content, about 8.7%, and
contained the most holmquistite and petalite at 2.8% and 1.4% respectively. Of the total
lithium distribution, it was determined that 95% is in the form of spodumene in COMP
A and B. Due to the higher petalite and holmquistite content, and comparatively lower
spodumene content, only 80% of the total lithium distribution was found as spodumene
in COMP C.
Based on this mineralogy analysis, it was anticipated that any petalite would be lost
to the floats fraction during DMS due to its low specific gravity. It is also expected that
holmquistite, a lithium-bearing gangue mineral likely associated with the waste rock,
will report to the DMS sinks due to its high S.G but should be rejected downstream
during the DMS magnetic separation stage.
3.2 QEMSCAN Analysis of Non-Magnetic Flotation Feed
QEMSCAN analysis of the non-magnetic flotation feed is presented in Fig. 3 and

reports a spodumene content in the range of 11.3–16.5% by mass, a notable upgrade
from the DMS feed samples. There is considerable mica remaining in all samples, with
the highest in COMP A once again. It should be highlighted that these minerals are
expected to be removed during the mica flotation stage. The distributions of plagioclase
and quartz are similar in the flotation feed as they were in the DMS feed; COMP A has
a higher quartz content and lower plagioclase content than samples B or C. The amphi-
bole content was slightly enriched after DMS where it increased to 2.2% in COMP A
and C and remained at 1.4% in COMP B.
Of the lithium minerals, spodumene was well liberated in all samples, falling
between 85.9% and 87.3%. Petalite liberation varied from 56.7% to 77.3% but was only
Plagioclase
Quartz
Microcline
Spodumene
Muscovite
Biotite
Holmquistite
Amphibole
Diopside
Petalite
COMP A - DMS Feed
Chamosite COMP B - DMS Feed
Calcite COMP C - DMS Feed
0 5 10 15 20 25 30 35 40 45
Mineral Mass Distribution (%)
Fig. 2. Semi-quantitative XRD mineralogy of DMS feed for each composite sample.
Spodumene
Petalite
Quartz
Plagioclase
K-Feldspar
Micas
Amphibole
Other Silicates
Apatite
Oxides
Mn-Fe Phosphate COMP A - Flotation Feed Non-Mag
Calcite COMP B - Flotation Feed Non-Mag
COMP C - Flotation Feed Non-Mag
Other
0 5 10 15 20 25 30 35 40 45
Mineral Mass Distribution (%)
Fig. 3. QEMSCAN modal mineralogy of the non-magnetic flotation feed for each composite
sample.
reported to be 0.32–0.39% of the total sample distributions. The primary silicate gangue
minerals, quartz and feldspars, had very good liberation above 94% for all samples.
Both muscovite and biotite exhibited good liberation from 77.7% to 79.4% and 69.7%
to 74.3%, respectively.
The presence of iron-bearing minerals like biotite, tourmaline, and other amphiboles
or pyroxenes, can be an obstacle in producing concentrate grading >6% Li2O and <1%
Fe2O3 if they cannot be sufficiently rejected from the flotation feed with magnetic sepa-
ration. While biotite rejection is targeted during mica flotation, Fe-Al silicates, like
amphiboles, exhibit similar flotation behavior to spodumene and may be difficult to
selectively reject in the spodumene flotation stage.
4 Beneficiation Testwork Results
The spodumene concentrate from the Piedmont Lithium Project is expected to be pro-
duced using DMS and flotation with magnetic separation used at various points in the
process. As mentioned, the DMS results discussed in this paper are from a DMS pilot
plant program. The flotation results are from locked-cycle testwork that was completed
following a developmental bench flotation test program to optimize flotation condi-
tions. Finally, the combined performance of the DMS and flotation circuits are pre-
sented to evaluate global spodumene recovery and final concentrate quality.
4.1 Dense Media Separation
The cut-points selected for this DMS testwork were based on previously conducted
heavy liquid separation (HLS) testwork to produce a 6.0% Li2O concentrate with the
−6.4/+3.3 mm and the −3.3/+1.0 mm size fractions of each composite sample. The cut-
points determined from HLS testwork and used in the DMS pilot plant are presented in
Table 2.
As discussed, the second cut-point DMS sinks product for each composite sample
was passed through dry magnetic separation to reduce iron content from the final coarse
spodumene concentrate. For each composite sample, the results from the two-size frac-
tions for each composite sample were combined to determine the global DMS metal-
lurgical performance. The global recoveries and coarse concentrate Li2O and Fe2O3
assays for each composite sample are presented in Fig. 4. In addition, the global Li2O
and Fe2O3 assays and distributions in the DMS second pass sinks and the DMS tailings
are presented in Tables 3 and 4, respectively.
As shown in Fig. 4, a coarse spodumene concentrate grading >6% Li2O was pro-
duced with each sample. COMP A and C were able to meet the <1% Fe2O3 target, at
0.91% and 0.84% respectively, while COMP B fell just above the target with 1.15%
Fe2O3. All three composite samples achieved 15–16% lithium recovery in the coarse
−6.4/+3.3 mm size fraction, but performance differed significantly in the finer
Table 2. The selected media S.G. cut-points used for DMS testwork.
Media specific gravity
First cut-point Second cut-point
Composite sample −6.4/+3.3 mm −3.3/+1.0 mm −6.4/+3.3 mm −3.3/+1.0 mm
COMP A 2.65 2.65 2.95 2.90
COMP B 2.65 2.65 2.95 2.90
COMP C 2.65 2.65 2.95 2.85
Table 3. Assays and distributions for the non-magnetic products and magnetic concentrate of the
DMS second pass sinks.
Non-magnetic products Magnetic concentrate
Mass Dist. (%) Assay (%) Mass Dist. (%) Assay (%)
DMS second pass sinks (%) Li Fe2O3 Li2O Fe2O3 (%) Li Fe2O3 Li2O Fe2O3
COMP A 4.99 24.2 2.51 6.12 0.91 1.33 1.30 8.19 1.24 11.1
COMP B 5.27 29.9 4.04 6.33 1.15 3.26 2.54 25.0 0.87 11.5
COMP C 6.08 37.3 2.43 6.32 0.84 3.11 2.32 17.5 0.77 11.8
Table 4. Assays and distributions in the DMS tailings.

Mass Distribution (%) Assays (%)
DMS first pass floats (%) Li Fe2O3 Li2O Fe2O3
COMP A 23.8 3.47 6.34 0.18 0.48
COMP B 27.4 3.89 7.56 0.16 0.41
COMP C 29.2 4.26 6.00 0.15 0.43
40 8
Li2O or Fe2O3 Concentrate Assay (%)

35 7
Global Lithium Recovery (%)
30 6
25 5
20 4
15 3
10 2
5 1
0 0
COMP A COMP B COMP C
Li Rec. (%) -3.3/+1.0 mm Li Rec. (%) -6.4/+3.3 mm

Global %Li2O Conc. Assay Global %Fe2O3 Conc. Assay
Fig. 4. Global DMS lithium recovery and concentrate %Li2O and %Fe2O3 assay after dry magnetic
separation.
−3.3/+1.0 mm fraction. Lithium recovery in the finer fraction was 9.5%, 14.0%, and
20.8% for COMP A, B, and C, respectively. The best global DMS performance was
achieved with COMP C at 37.3% lithium recovery and the highest Li2O content (6.33%)
and the lowest Fe2O3 content (0.84%). It is possible the higher S.G. cut used in the sec-
ond DMS pass for COMP A and B resulted in a slightly increased spodumene deport-
ment to the middling in the −3.3/+1.0 mm fraction.
Table 3 highlights the benefits of magnetic separation on the coarse DMS concen-
trate. Lithium losses were relatively low, between 1.3% and 2.5%, with good Fe2O3 and
mass rejection. A significant amount of mass was rejected from COMP B and C which
largely contributed to the upgrading of Li2O in the final non-magnetic DMS concen-
trate. Magnetic separation removed most material from COMP B, rejecting nearly 25%
of the Fe2O3 and grading 11.5% Fe2O3 in the magnetic products.
The DMS tailings data presented in Table 4 indicates a large portion of the global
feed can be rejected with DMS at the S.G. cut of 2.65. COMP C had the lowest Li2O
tailings grade (0.15%) and saw the most material rejected with DMS at 29.2% mass and
only 4.3% lithium losses. Meanwhile, COMP A saw the lowest mass rejection and
lithium losses at 23.8% and 3.5%, respectively, but was still able to produce an on-
target coarse concentrate. Overall DMS combined with magnetic separation proved an
effective way to produce a portion of the final spodumene concentrate and was able to
reject a significant portion of material with low lithium losses upstream of flotation.
4.2 Flotation
The DMS middlings, or the floats product of the second DMS cut-point for each size
fraction, were combined with the −1.0 mm fines as the feed to the flotation circuit. The
mass and lithium distributions, as well as Li2O and Fe2O3 head assays for the flotation
circuit feed, are presented in Table 5.
The flotation feed head assays for COMP A and COMP B were upgraded in com-
parison to their bulk %Li2O head assays presented in Table 1. However, due to the
exceptional performance of DMS with COMP C, the %Li2O flotation feed assay for this
sample was reduced and %Fe2O3 was increased. The mass fractions in the flotation feed
for COMP B and C were similar, around 60%; however, a significantly higher mass
fraction around 70% reported to the flotation stage for COMP A.
Upstream Processing As shown in Fig. 1, the feed to the flotation circuit passes
through several steps before the spodumene flotation stage. Once the material was stag-
ground to 100% passing 300 μm, it had to pass through medium and high-intensity
magnetic separation, two desliming stages, and mica flotation, before spodumene was
ready to be recovered.
MIMS and WHIMS Multiple passes through magnetic separation were necessary to
reject iron-bearing minerals that would likely be recovered during spodumene flotation
and dilute the spodumene concentrate. The results of both magnetic separation stages
in the validation locked-cycle flotation testwork are presented in Table 6.
In general, magnetic separation was very beneficial for all three composite samples
and was able to reject between 35% and 58% of the global Fe2O3 content. MIMS was
most effective on COMP A, from which 14% of the Fe2O3 distribution was removed
with only 2.8% lithium losses. Meanwhile, WHIMS was best with COMP C removing
nearly 50% of the Fe2O3 distribution at 7.52% lithium losses. Between both magnetic
separation stages, COMP C achieved the highest global iron removal at 58.6% and
Table 5. Comparison of original DMS feed and flotation feed assays, including lithium distribution
and upgrade factors.
Flotation circuit head assays Original global head Upgrade
Mass Li dist. (%) assays (%) ratios
Sample (%) (%) Li2O Fe2O3 Li2O Fe2O3 Li Fe
COMP A 69.9 80.0 1.28 2.14 1.12 1.92 1.14 1.11
COMP B 60.1 67.4 1.25 1.59 1.05 1.40 1.19 1.14
COMP C 61.6 56.1 0.94 2.52 1.05 2.01 0.90 1.25
Table 6. Global assays and distributions of the MIMS and WHIMS magnetic concentrates.
MIMS global mag. conc. WHIMS global mag. conc.
Mass Dist. (%) Assay (%) Mass Dist. (%) Assay (%)
Sample (%) Li Fe2O3 Li2O Fe2O3 (%) Li Fe2O3 Li2O Fe2O3
COMP A 3.57 2.81 14.0 0.99 6.53 5.50 5.17 39.2 1.18 11.9
COMP B 2.57 1.42 4.03 1.03 4.16 4.31 5.10 31.5 1.39 12.2
COMP C 1.80 1.66 8.82 0.97 10.1 7.52 7.76 49.8 1.10 13.6
COMP B the lowest at 35.5%. The combined global lithium losses of the WHIMS and
MIMS separation stages ranged from 6.5% to 9.4%. These findings suggest that increas-
ing iron rejection from the flotation feed is possible, but will be accompanied by higher
lithium losses, as was the case for COMP C.
Desliming The desliming of the flotation feed during locked-cycle testwork was per-
formed twice, first before mica flotation, and again before spodumene flotation. The
global lithium losses to the slimes were relatively low for COMP A and B at 1.2% and
1.7%, respectively, with slightly higher lithium losses in COMP C at 4.2%. This can be
attributed to the high mass rejection of COMP C at 5.7%, while in COMP A and B only
1.6% and 2.1% of the mass was rejected, respectively.
Mica Flotation The performance of the mica flotation stage is evaluated primarily
through the K2O and mass distributions in the mica concentrate. For these three sam-
ples, COMP A saw the highest mass (8.6%) and K2O (24.2%) report to the mica con-
centrate with only 3.3% lithium losses. To some extent, this was expected based on the
QEMSCAN results shown in Fig. 3 indicating the high mica content of this sample.
COMP B and C display similar mica contents in Fig. 3 and reported similar K2O and
lithium distributions from 11.7% to 15.0% and 1.6% to 1.8%, respectively.
Spodumene Flotation Testwork The development bench scale flotation testwork for
the three composite samples was part of a bigger development test program. The focus
of this work was to produce a spodumene concentrate with >6.0% Li2O while maintain-
ing maximum lithium recovery. The major findings of this study related to rougher
collector dosage and type as well as the number of cleaning stages. Any effective vari-
able changes were optimized around one composite sample and then validated with the
remaining samples prior to locked-cycle testing.
The initial rougher collector dosage was set to 500 g/t with a 90:10 blend of two fatty
acid collectors. This was quickly changed to the single collector that composed 90% of
the original collector blend, as the difference in performance was considered negligible.
Similarly, the number of cleaning stages was quickly reduced from three to two, as
losses in the third cleaner outweighed its upgrading ability. It was determined through
a series of tests on COMP A that slightly lowering the collector dosage was beneficial
for producing a >6.0% Li2O concentrate grade, this change was maintained for locked-
cycle testing.
The importance of the conditioning stage in these systems has long been recognized
as an essential step to promote quick selective spodumene flotation [7, 8]. One observa-
tion to support the importance of spodumene conditioning was made during a brief
spodumene kinetics study. This work used three rougher flotation stages and ran one
kinetics test for each composite sample. As shown in Fig. 5, the kinetics curves for each
sample followed a first-order model. However, negligible change in concentrate %Li2O
grade was observed among the first, second, and third rougher concentrates for all com-
posites. The kinetics results suggest that around 80% of the spodumene in the flotation
circuit feed is floated in the first 1.5 minutes of rougher flotation and very little spodu-
mene is recovered in the final minute.
The impact of two alternate fatty acid collectors, as well as a 50:50 blend of the
baseline collector and the second alternate collector, were investigated with COMP
B. Of the three conditions, only the first alternate collector was unable to produce a final
concentrate grading >6.0% Li2O, while the second alternate collector and the blend
produced the highest Li2O concentrate grades. However, when tested on COMP A and
C for validation there was no benefit of the alternate collector or blend. The reason for
this inconsistency was not clear; however, the chemistry of the modified collector could
be involved. The initial unsaturated fatty acid collector was used in the remainder of
this testwork.
100
90
80
Lithium Recovery (%)
70
60
50
40
30
20
10
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (minutes)
Fig. 5. Flotation kinetics results for each composite sample.

The concentrate Li2O grade vs. lithium recovery curves for the final flotation condi-
tions determined by bench testing is presented in Fig. 6. The data points (from left to
right) represent the second cleaner, first cleaner, rougher, and rougher scavenger flota-
tion stages, respectively. The same parameters used in these three tests were applied to
the validation flotation LCT work.
As shown in Fig. 6, the best flotation recovery and concentrate grade were achieved
with COMP B at 70% lithium recovery and 6.1% Li2O concentrate grade. A similar
concentrate grade was achieved with COMP C, but with 5% lower lithium recovery.
COMP A achieved 69% lithium recovery and fell just shy of the 6.0% Li2O target con-
centrate grade. Successful upgrading was achieved in the first cleaner with only small
lithium losses. The second cleaner stage was needed to upgrade the concentrate to
>6.0% Li2O, but this was accompanied by a slight increase in lithium losses.
Locked-Cycle Testwork One locked-cycle test using six flotation cycles, each with a
2-kg charge, was conducted to validate the optimized flotation conditions. The stage
and global performances are presented in Table 7.
The stage LCT recoveries reported in Table 7 are considerably higher for COMP A
and B than the bench scale flotation test results presented in Fig. 6, at 79% and 82%,
respectively; similar bench scale and LCT recoveries were reported for COMP C at
65%. The LCT flotation concentrate grade for COMP B was the only on-target concen-
trate grading >6.0% Li2O at 6.36%, while the final flotation concentrates from COMP
A and C graded 5.7% Li2O. The concentrate grade of the COMP A bench results pre-
dicted a 6.1% Li2O concentrate grade, so there is potential that further refinement of the
flotation procedure could improve locked-cycle results. It should be reiterated that the
lower global lithium recovery from flotation with COMP C was heavily impacted the
high DMS recovery.
7.0
6.0
5.0
Li2O Grade %
4.0
3.0
2.0
Head
Assays
1.0 % Li2O
0.0
50 60 70 80 90 100
Lithium Recovery (%)
Fig. 6. Bench performance of the optimized flotation conditions for each composite.
Table 7. Stage and global assays and distributions of the LCT spodumene second cleaner concentrates.
Stage Global
distribution distribution
Mass (%) Mass (%) Assays (%)
LCT second cleaner concentrate (%) Li Fe2O3 (%) Li Fe2O3 Li2O Fe2O3
COMP A 17.9 78.8 14.2 11.6 53.1 8.73 5.71 1.24
COMP B 17.8 82.0 12.1 10.2 54.9 6.53 6.36 1.07
COMP C 11.0 65.0 3.2 6.76 36.9 2.39 5.70 0.72
Table 8. Combined table of DMS and LCT flotation results for each composite.
Mass Distribution (%) Assays (%)
Combined DMS and LCT flotation (%) Li Fe2O3 Li2O Fe2O3
COMP A 16.7 77.4 11.5 5.84 1.14
COMP B 12.9 80.5 11.01 6.36 1.10
COMP C 12.8 73.6 4.86 6.00 0.78
4.3 Global (DMS + Flotation) Metallurgical Performance
The combined metallurgical performance of DMS and LCT testing was balanced to
determine the potential global recovery and quality of the final combined concentrate
from the Piedmont Lithium Project. The summarized results for iron and lithium are
presented in Table 8. The global losses to desliming, mica flotation, and magnetic sepa-
ration during DMS and flotation have been presented in the previous sections and are
omitted here.
The global lithium recovery from all three composite samples was good at 73.6–
80.5%, with final concentrate grades >6.0% Li2O possible for COMP B and C. The
highest concentrate grade and recovery was achieved with COMP B at 80.5% lithium
recovery and 6.36% Li2O. COMP A also had high recovery but fell just short of the
<6.0% Li2O target concentrate grade. The lowest iron content was in the COMP C con-
centrate at 0.78% Fe2O3, while COMP A and B concentrates contained very close to
<1% Fe2O3. It is anticipated that further development of the WHIMS magnetic separa-
tion stages in the flotation circuit can reduce this content to below the target in COMP
A and B.
5 Conclusions
The results of this testwork prove the combined use of dense media separation and
flotation is an effective and successful approach to produce spodumene concentrate
from this deposit. The final combined products for each sample achieved from 73% to
80.5% lithium recovery with concentrate assays that approached or exceeded the 6%
Li2O target.
The two-stage DMS circuit with dry magnetic separation on the final DMS sinks
product produced on target coarse spodumene concentrates with >6.0% Li2O and low
iron content for all three composite samples. Multi-stage WHIMS magnetic separation
performed in the upstream flotation circuit was essential for lowering the iron content
of the spodumene flotation feed. This stage was most effective on COMP C; however,
it is anticipated this portion of the flowsheet can be better developed to further reduce
Fe2O3 in spodumene flotation concentrates of COMP A and B.
Spodumene flotation is very effective with all composite samples, contributing
around 54% of the global lithium recovery in COMP A and B. The lower global lithium
distribution of 37% in the spodumene flotation concentrate of COMP C is attributed to
the superior performance of DMS with COMP C. The overall metallurgical perfor-
mance of these three composites suggests Piedmont Lithium’s North Carolina Lithium
Project is well suited to produce spodumene concentrates for the conversion to lithium
hydroxide.
References
1. Jaskula B. Lithium. U.S. Geological Survey Mineral Commodity Summary, January 2022; 2022.
2. Bowell RJ, Lagos L, de los Hoyos CR, Declercq J. Classification and characteristics of natural lithium
resources. Elements (Quebec). 2020;16(4):259–64. https://doi.org/10.2138/gselements.16.4.259.
3. Bulatovic SM. Beneficiation of lithium ores. In: Handbook of flotation reagents: chemistry theory
and practice – volume 3: flotation of industrial minerals. Elsevier B.V; 2014. p. 41–56.
4. Our Project. Piedmont lithium. https://piedmontlithium.com/about/project. Accessed 27 Feb 2022.
5. Gibson CE, Aghamirian M, Grammatikopoulos T, Smith DL, Bottomer L. The recovery and con-
centration of spodumene using dense media separation. Minerals (Basel). 2021;11(6):649. https://
doi.org/10.3390/min11060649.
6. Gibson CE, Aghamirian M, Grammatikopoulus T. The removal of iron bearing silicate minerals
from a hard rock lithium ore. In: 49th annual Canadian mineral processors operators conference,
Ottawa, ON, Canada. January 17–19, 2017. p. 205–215.
7. Norman J, Gieseke EW. Beneficiation of spodumene rock by froth flotation. Technical Publication
No. 1161. American Institute of Mining and Metallurgical Engineers; 1940.
8. Munson GA, Clarke FF. Mining and concentrating spodumene in the Black Hills, South Dakota. In:
Transactions AIME November 1955, Mining Engineering. 1955. p. 1041–5.
Using Polymetallic Nodules to Innovate Manganese
Markets: A New Manganese Feed for SiMn Alloy
Production
M. N. Boulby1 (*), V. Canaguier2, and J. R. Donald1
1
The Metals Company (TMC), Vancouver, BC, Canada
jeff@metals.co
2
SINTEF Industry, Trondheim, Norway
Abstract. The Metals Company (TMC) is on track to becoming a leading producer of

base and strategic metals obtained from vast high-grade seafloor polymetallic nodule
deposits. The Metals Company believes the seafloor minerals industry is an opportunity
to develop a more environmentally and societally attractive way to produce cleaner base
and strategic metals.
Deep-sea nodules from the Clarion-Clipperton Fracture Zone (CCZ) in the Pacific
Ocean represent the largest known resource of nickel, cobalt and manganese. The
Metals Company is developing a metallurgical process based on existing nickel flow-
sheet technologies to maximize the business outcomes consistent with the highest
environmental standards, including targeting a near-zero solid waste facility.
This work presents an overview and context of the project with a focus on the
manganese silicate product and the downstream opportunities to produce silicon-
manganese alloy (SiMn) and other manganese products. The work presents an over-
view of polymetallic nodules and TMC, a flowsheet and pilot plant overview,
characterization of the manganese silicate, experimental and theoretical results to
produce SiMn, and a value-in-use assessment compared to conventional SiMn feeds.
The significance of the Manganese Silicate product to a zero-waste flowsheet will
be discussed.
Keywords: Polymetallic nodules · Manganese · Manganese silicate · Silicon-

manganese alloy · SiMn · Zero waste flowsheet
1 Background and Context
1.1 Polymetallic Nodules in the Clarion-Clipperton Zone
Containing nickel, cobalt, copper and manganese, polymetallic nodules occur in most
oceans of the world. The nodules lie on the seafloor sediment. The largest deposit in
terms of nodule abundance and metal concentration occurs in the Clarion-Clipperton
Zone (CCZ) on vast abyssal plains at a depth of 4000–6000 m. The CCZ is located
almost entirely in international waters.
https://doi.org/10.1007/978-3-031-17425-4_108
994 M. N. Boulby et al.
1.2 Historical Context and Present Status
In the 1970s and early 1980s, there was substantial commercial interest in polymetallic
nodules. This initial activity and the promise of development led to the establishment of
the International Seabed Authority (ISA), an autonomous international organization
established under the United Nations Convention on the Law of the Sea (UNCLOS).
The ISA controls activities on the ocean floor beyond the limits of national jurisdiction
with its primary focus being on the administration of the mineral resources within the
international seabed area (the Area). Resources within the Area have been designated as
the “common heritage of mankind”.
The ISA has developed and adopted Regulations on Prospecting and Exploration for
Polymetallic Nodules in the Area and is currently drafting the exploitation regulations
that will govern the commercial harvesting of seafloor minerals. As of March 31, 2022,
the ISA has entered into 19 exploration contracts for polymetallic nodules, 17 of which
are in the CCZ, one in the Indian Ocean and one in the Western Pacific Ocean.
1.3 The Metals Company
The Metals Company (TMC-NASDAQ) has the world’s largest commercial explora-
tion portfolio for polymetallic nodules. The company’s stated purpose is to source the
metals the world needs for a sustainable future with less environmental and social
impact than conventional land-based mining. Polymetallic nodules present a means to
enable this purpose because of their high grade, abundance and the fact they sit unat-
tached on the surface of the abyssal plain. As such, TMC intends to be the first company
to commercialize their recovery. TMC is also the only organization with a NI 43-101
compliant Preliminary Economic Assessment and SK1300 compliant Initial Assessment
and two NI 43-101 and SK 1300 compliant technical resource statements for an end-to-
end value chain from nodule seabed collection to processing and refining of products
with confirmed markets.
Licensed Areas and Resource Statement
The Metals Company through its subsidiaries holds exploration rights to three licensed
areas in the CCZ sponsored by the pacific islands of Nauru (NORI), Kiribati (Marawa),
and the Kingdom of Tonga (TOML). In total, these license areas comprise approxi-
mately 225,000 square kilometers. A map of the CCZ and TMC’s licensed exploration
areas can be found in Fig. 1. The figure also shows the 13 areas that have been desig-
nated by the ISA as areas of particular environmental interest (APEI). These areas
encompass 43% of the total CCZ area and are not available for development in order to
protect the biodiversity of the CCZ and the deep ocean [1].
Of this total area, TMC has conducted advanced exploration in the Nauru and Tonga-
sponsored zones. Together they contain nodule quantities of 1.22 billion wet tonnes
inferred, 411 million wet tonnes indicated and 7 million wet tonnes measured. The
NI 43-101 and SK 1300 compliant technical resource statement is summarized in
Table 1.
Using Polymetallic Nodules to Innovate Manganese Markets: A New Manganese Feed… 995
Fig. 1. TMC exploration areas (colored) amongst exploration areas issued to other parties (white) for
nodule exploration within the Clarion Clipperton Fracture Zone.
1.4 Impact of Polymetallic Nodules on Existing Manganese Markets
The development of polymetallic nodule projects will have a significant impact on the
global manganese markets. Figure 2 presents the world’s manganese mines with a 60
ktpa nickel production equivalent polymetallic nodule project overlaid in green. This
operation will represent the largest source of manganese globally at about 7% of global
production (2017 basis).
The primary uses of manganese are in the steel industry, which consumes upwards
of 90% of all production. Portable batteries and aluminium beverage cans are primary
non-steel uses. In each case, manganese plays a vital role in improving the properties of
the alloys and compounds. The chart in Fig. 3 shows an estimate of how much manga-
nese is consumed in each of its end-use applications.
In 2017, just over one-half of all Mn ore consumed went into SiMn production. An
additional 35% went into HC and Refined Ferromanganese (FeMn). Electrolytic
Manganese Metal (EMM) accounted for 13% of ore usage.
With such a significant volume of manganese impacting the manganese supply,
TMC needs to ensure the product is consistent with the market requirements. Although
there is significant growth in manganese sulfate for battery uses and a sizeable market
for EMM and other specialized manganese products, the production volumes from nod-
ules overwhelms the manganese demand in these markets.
Table 1. Summary of TMC’s NI 43-101 & SK 1300 compliant technical resource statement (2022).
TMC exploration contract area NORI TOML Marawa
Sponsoring state Republic of Nauru Kingdom of Tonga Republic of Kiribati
Exploration area 74,830 km2 74,713 km2 74,990 km2
Technical resource statement Yes Yes Work in Progress
Measured resource (wMt) 4.0 2.6
Indicated resource (wMt) 341 69.6
Inferred resource (wMt) 11 696
Manganese 29.5% 29.1%
Nickel 1.3% 1.3%
Copper 1.1% 1.1%
Cobalt 0.2% 0.2%
Fig. 2. TMC will represent the largest source of manganese globally [2]. The size of the bubble indi-
cates the total contained manganese in ore production. TMC’s project, equivalent to 60 ktpa Ni produc-
tion, is shown.
Thus, TMC’s strategy for manganese is to target the manganese alloy for steel
markets.
1.5 Polymetallic Nodule Flowsheet Options and Associated Manganese

Products
Several potential flowsheets for the processing of polymetallic nodules have been pro-
posed [3–8]. In the published investigations, manganese seems an afterthought with
most concentrating on the recovery of nickel, copper and cobalt [6].
Hydrometallurgical Flowsheets and Manganese Products
Recent work has recognized both the economic drivers and the need to reduce or elimi-
nate waste streams, and thus Mn must be treated as a product.
Acid leach hydrometallurgical-based flowsheets in which the manganese remains in
solution produce a rich manganese precipitate [6]. Conceptually, this product could be
fed to manganese alloying operations, but the fine and wet nature of the material is
unsuitable for direct feed and the product would have to undergo drying and agglom-
Mn ore consumption by product (2017) EMM/HPEMM breakdown

EMM end use
13% High purity EMM

12% (2.5Mt) 75%
95%
20%
EMM 5%
19.2Mt ore
23% 200 series SS
(contained Mn 5.0% AI and steel alloys
Batteries
basis)
EMM and High Purity EMM end use
EMM HPEMM
52%
200 series SS Battery grade MnSo4
Batteries Superalloys
HSLA 3000 series AI alloys
Non-magnetic steel Welding powder
Catalysts
Pharmaceuticals
SiMn HC FeMn Ref FeMn EMM
Fig. 3. Manganese ore consumption by product, 2017 [2].
eration. In these flowsheets, the silica and iron would be removed from the nodules and
likely have to be disposed of as residue waste. The manganese alloy producers who
would receive the resultant relatively pure manganese oxide would then have to add
silica and iron to their feed blend.
Ammonia leach-based flowsheets produce a manganese-rich residue. Most histori-
cally view this residue as waste [7], but Randhawa, et al. [9], reported the laboratory
scale production of a SiMn product from the residue containing 26% Mn, 10% Fe and
6% SiO2 blended with manganese ore. The manganese recovery was 77%.
Other manganese products proposed from hydrometallurgical flowsheets include
MnCl2 and MnCO2. These products require further processing to EMM, MnSO4 or
other saleable products. As discussed above, even a small polymetallic nodule opera-
tion would produce more manganese than can be consumed by the market for these
products.
Pyrometallurgical Flowsheet and Manganese Products
Processing of the nodules by pyrometallurgy produces a manganese silicate product
that can be further processed to manganese alloys [10–12]. The high grade of manga-
nese in the product rivaling conventional high-grade manganese ores, the dry, pre-
reduced nature and favourable impurity profile as well as the physical attributes of a
slag material (strong, dense) make this a potentially disruptive product in the produc-
tion of manganese alloys.
The manganese silicate slag product from the smelting unit operation represents
about 90% of the mass of the solids from the operation. Thus, early in the flowsheet, the
majority of the flow is removed from the flowsheet and only 10% of the solids volume
continues to nickel, copper, and cobalt separation and refining. This prevents the bulk
of the mass from proceeding through the entire flowsheet, allowing for a significant
reduction in the scale of equipment.
The manganese silicate also contains the stable oxides from the nodules, notably
silica and a portion of the iron which are required in the downstream manganese alloy-
ing process. These attributes of the product contribute to the zero waste flowsheet
objective and the product’s inherent value in use.
Successful laboratory-scale investigations into the production of SiMn from slags
generated by the smelting of polymetallic nodules have been undertaken by CSIR –
National Metallurgical Laboratory (India) [13, 14]. Fe-Si-Mn alloys were produced
from a 40–45% MnO manganese silicate from nodules by varying the temperature and
basicity of the slags.
2 TMC Flowsheet Selection and Overview
Extraction of base and strategic metals from the nodules is done through a custom flow-
sheet, created using a combination of existing nickel flowsheet technologies. Several
flowsheet objectives were established during the concept development phase – notably
a process which would yield high recovery rates of the battery grade metals (nickel,
copper, cobalt), a manganese product that is readily saleable to existing manganese
markets, and target zero solid waste (all products being useful materials with an estab-
lished market). The project objectives, the selection of the flowsheet and other early
aspects of the project development have been discussed previously in Donald and
McLachlin [15].
After arriving to the port, the nodules are first processed through a rotary kiln at high
temperatures. The purpose of the kiln is to dry and dehydrate the nodules, while also
beginning the reduction process. The kiln creates a low loss on ignition (LOI) calcine
product, which is then transferred and fed into an electric arc furnace. In the furnace,
the reduction of the calcine is completed, and this smelting operation creates two dis-
tinct phases: a dense, metallic phase composed of an Iron-nickel-copper-cobalt alloy,
and a less dense oxide phase containing the majority of the manganese content – known
as the manganese silicate. These immiscible phases are separated by simply tapping
(removing material) from the furnace at different elevations. The metallic phase is pro-
cessed downstream to separate the individual components – while the proposed use for
the silicate layer during concept development was to use it as an alternative to manga-
nese ores as feed for silicon-manganese production.
TMC conducted a pyrometallurgical pilot scale operation in collaboration with FL
Smidth (Rotary Kiln Calcining) and eXpert Process Solutions (XPS): A Glencore
Company (DC Arc Furnace Smelting, Sulfidation and Converting). Beginning with
approximately 70 metric tonnes of wet nodules collected from the CCZ, about 35
tonnes of calcine was generated at the FLS facility in Whitehall, Pennsylvania. This
was then shipped to Falconbridge, Ontario to undergo processing in a 250 kW DC elec-
tric arc furnace. Over two campaigns spanning 3 weeks, a total of about 3 metric tonnes
of alloy and 25.5 metric tonnes of manganese silicate were produced.
Employing benchscale testwork, commercial knowledge of ferronickel processing,
and thermodynamic analysis using FactSage™ a mass and energy balance was devel-
Table 2. Comparison of the expected manganese silicate composition and the composition of the
manganese silicate produced in TMC’s pilot operations.
Component Model composition (wt%) Pilot composition (wt%)

MnO 52.2 55.2
FeO 1.9 1.9
SiO2 25.6 23.2
P2O5 0.09 0.03
Al2O3 5.7 5.8
CaO 3.2 3.6
MgO 4.3 5.3
Others 7.0 5.6
oped in preparation for the pilot demonstration. The outcomes of the pilot operation
were remarkably similar to the predicted values which can be seen below in Table 2.
3 Manganese Silicate Characteristics
The manganese silicate from the pilot operations was characterized in several ways.
The first was a summary comparison of manganese ores and manganese-rich slag
(MRS), which can be seen in Table 3. Optical and Scanning Electron Microscopy
(SEM), as well as Electron Dispersive Spectroscopy (EDS), were also performed.
A sample of the manganese silicate was prepared and analysed with optical and
scanning electron microscopy, shown in Fig. 4.
There are four major phases apparent in the images. The primary phases are MnO,
MnO⋅SiO2 and a mixed oxide phase. The small bright particles are entrained alloys. The
surface was examined with a Keyence microscope using reflective light at 200X and
500X magnifications. Large area scans were made by stitching adjacent images together
(10 × 10 images at 200X). The areas of different phases of the material observed in
reflective light were estimated with the Keyence software. Table 4 shows the indicative
composition of phases in the pilot silicate product using the Keyence software estimate
combined with the EDS analysis.
4 Proposed Market for TMC’s Manganese Silicate Product
As discussed in Sect. 2, the most significant market for manganese is steelmaking alloy.
Manganese is known as an important additive in steel making, as it reacts with dis-
solved oxygen in the liquid steel melt and creates an oxide layer that can be removed.
Dissolved oxygen in the steel melt creates a porous structure when the melt eventually
solidifies. The removal of this dissolved oxygen with manganese creates a stronger and
more durable final solid steel product [16].
Table 3. A comparison of key data between TMC’s manganese silicate product (pilot outcomes) and
various manganese sources currently being used for silicomanganese production.
Mn source Typical grade (%) Mn:Fe Mn:P
TMC 40–44 29:1 1426:1
HG Mn ore 1 44–48 4:1 1200:1
HG Mn ore 2 40–44 10:1 420:1
MG Mn ore 1 36–39 7:1 1800:1
MG Mn ore 2 36–39 8:1 1200:1
LG Mn ore 25–30 35:1 700:1
Manganese rich slag 30 N/A N/A
Fig. 4. Optical image at 500× of the manganese silicate, where four distinct phases are labelled and
EDS results of indicative mineralogy are shown.
Table 4. Corresponding details for the four phase labels added to Fig. 4.
Image label Phase description Indicative mineralogy

A MnO⋅SiO2 42.3
B Mixed oxides 42.2
C MnO 12.6
D Entrained alloy 0.20
Historically ferromanganese was the primary product for this use, but in the past
decades, the market shifted to the use of silicomanganese for this process, as silicon
acts as a significantly more effective deoxidizer than manganese. SiMn was found to
introduce less impurities (phosphorus, carbon, aluminum, nitrogen, arsenic, etc.) into
the steel product. Due to these inherent advantages of using SiMn, ferromanganese
producers began treating their slag product (manganese-rich slag or MRS) to create this
more desirable deoxidizing agent in the Duplex process.
TMC’s manganese silicate has a high grade of manganese, is pre-reduced and con-
tains appropriate levels of iron and other oxides, as well as low levels of deleterious
impurities, notably phosphorus. It conceptually compares favourably in relation to both
manganese ore and manganese-rich slags as feed in the production of SiMn.
5 Production of SiMn from TMC Pilot Manganese Silicate
TMC began a testwork program to confirm that silicomanganese could be generated

using their pilot manganese silicate product as the sole manganese source and to deter-
mine the conditions under which this was possible. Success in this program would be
based on two factors: the ability to generate a silicomanganese alloy product with high
enough metallic silicon content consistent with silicomanganese currently being used
in the steel industry; and an impurity deportment to the silicomanganese alloy product
within the limits of typically accepted values. This bench scale program was carried out
at SINTEF Industry, a world-leading testwork laboratory for manganese ores based out
of Trondheim, Norway.
High-temperature reduction experiments were carried out in a thermogravimetric
furnace with 11.24 g of charge and continuous measurement of the weight. A charge
mixture of TMC’s product, quartz, lime, metallic iron and coke was used. The charge
was first heated to 1200 °C, and then held at 1650 °C until equilibrium was achieved.
The target slag phase deportment was 5 wt% manganese oxide and 40 wt% silica. The
TMC product does not contain high manganese oxides (MnO2, Mn2O3), resulting in
lower CO2 emissions from silicomanganese production with the polymetallic nodules
in comparison to production using traditional manganese ore feeds.
During testing, there was only 0.03 g of mass loss observed in the 30 minutes of
isothermal holding at 1200 °C, confirming the charge does not require any pre-reduction.
Upon completion of the experiment, the alloyed product and slag phases were analyzed
for composition. Table 5 shows the resultant assays for the SiMn alloy, derived from the
slag’s composition, and how they compare to a typical specification.
Table 6 shows a complete assay of several constituents in the resultant slag from the
SiMn alloy production process. It is evident from the deportment of impurities shown
in the slag that the impurity deportment to the alloy is within the reasonably expected
limits.
The results show that TMC manganese silicate can be used as feed to produce sili-
comanganese products whose composition falls within expected parameters. The reac-
tion rate is similar to traditional manganese sources, pre-reduction and its resultant CO2
emissions are avoided, as is the risk of carbon and energy losses. The product assays
show a high silicon content in the alloy, with the majority of impurities deporting to the
slag phase. The slag can be used as a mineral product, for example in the construction
industry [12].
Overall, the high MnO content and absence of high manganese oxides in the TMC
manganese silicate are significant advantages of the use of polymetallic nodules as the
Table 5. Comparison of TMC’s silicomanganese product with traditional specifications.

Component Typical specification (wt%) Alloy composition (wt%)
Mn 65.0–68.0 69.6
Si 18.5–21.0 18.1
Fe 10.8–11.3 10.7
C 1.5 1–2
Table 6. Composition of the slag after production of SiMn from TMC’s product.
Component Composition (wt%)
Al2O3 17.2
CaO 18.5
FeO 0.0
K2O 0.3
MgO 13.0
MnO 4.2
MnS 3.5
SiO2 45.4
SO3 1.0
TiO2 0.3
sole manganese source for silicomanganese production. Figure 5 shows the sample of
SiMn produced from the experiments at SINTEF.
Using the experimental outcomes as a basis for the TMC case, mass and energy bal-
ances can be prepared to compare the value-in-use of TMC’s manganese silicate to
conventional feeds. Five scenarios were assessed – the TMC manganese silicate, high-
carbon ferromanganese slags with and without metallics added, a high-grade manganese
content ore and mid-grade manganese content ore. Table 7 summarizes the quantities of
consumables for the production of 1 metric tonne of SiMn.
The CO2 emissions for each scenario were calculated based on the full conversion of
CO to CO2 in air, without energy recovery, with the amounts of CO2 already present in
the gas stream leaving the system added to this conversion. The results of these calcula-
tions can be seen in Table 8.
The “value-in-use” of TMC’s manganese silicate product can be estimated by equal-
izing the cost of feed and consumables for each of the scenarios. Table 9 presents the
cost savings that the operator realizes using TMC’s product when producing a unit of
SiMn in comparison to conventional feed blends. The savings increase with CO2 tax, as
the use of TMC’s manganese silicate product results in lower CO2 emissions. Therefore,
it is expected that TMC’s manganese silicate product will attract a premium to bench-
mark manganese ore prices.
Fig. 5. Microscopic image with labels showing the four distinct phases generated as a result of the
bench scale SiMn production from TMC’s manganese silicate product
Table 7. Comparison of feed and consumables required to produce a single metric tonne of
silicomanganese with various manganese feed options.
Cost assumption TMC Mn High grade Mid grade Mn rich Mn rich slag +
(USD) silicate Mn blend Mn blend slag metallics
TMC Mn $550/ta 1.71 – – – –
silicate
HG Mn ore $550/ta – 1.46 0.64 0.31 0.23
MG Mn ore $550/ta – – 1.02 – –
Mn rich slag $550/ta – – – 2.02 1.51
Metallics $800/t – – – – 0.22
Iron ore $120/t 0.12 – – 0.10 0.06
Quartz $50/t 0.23 0.46 0.51 0.56 0.53
Dolomite $50/t 0.03 0.08 – – –
Coke $250/t 0.48 0.63 0.64 0.50 0.41
Electric $0.1/kWh 3465 3500 3548 4034 3501
power
Total cost/t SiMn (USD) $894 $923 $933 $998 $986
a
Mn contained basis
Table 8. Comparison of quantified CO2 emissions for each of the identified scenarios.
TMC Mn High grade Mid grade Mn Mn rich Mn rich slag +
Mn source silicate Mn blend blend slag metallics
CO2 (g) emitted (kg/ 1275 1692 1792 1313 1128
tonne alloy)
Table 9. Value-in-use of TMC Mn silicate product compared to other scenarios assessed.

Cost savings for TMC Mn silicate in comparison to With $50/t CO2
Mn source conventional feeds (%) tax (%)
High grade Mn 7.2 12.4
blend
Mid Grade Mn 9.8 16.3
blend
MRS + Mn ore 25.9 26.3
MRS + metallics 22.9 21.1
6 Conclusions
The conclusions of the present work are:

• Polymetallic nodules represent an important source of future manganese supply;
• The only market that is large enough to consume the high volume of manganese
from polymetallic nodules is Mn-alloys for steelmaking;
• A pyrometallurgical flowsheet produces a manganese silicate product that is suitable
for alloy production and offers several significant advantages;
• Testwork confirmed that TMC’s manganese silicate product can be employed to
produce SiMn without the need for blending with other manganese feed sources;
• TMC’s manganese silicate product should attract a premium to manganese ore
benchmark prices from a value-in-use perspective.
Acknowledgments The authors would like to thank TMC for resourcing the work and for permission
to publish. Great work by SINTEF (Eli Ringdalen), Kingston Process Metallurgy (Chris Pelow, Trevor
Lebel, Boyd Davis), Hatch (Mike Kozlowski, Michael McCaffrey), Canadian Engineering Associates
(Apurva Patel, Cameron Harris), FLSmidth (Michael Prokesch, Reinaldo Penso), XPS (Mika
Muinonen, Abigail Sequeira) and CRU (Alex Laugharne) contributed to the outcomes presented and
the project development to date. Permission from CRU to use the manganese market graphics is
appreciated.
References
1. Smith CR, et al. Areas of particular environmental interest for ecosystem based management of
the Clarion Clipperton Zone: rational and recommendation to the International Seabed Authority,
design marine protected areas for seamounts and abyssal nodule province in Pacific high seas,
October 23–26, 2007, University of Hawaii.
2. CRU Consulting. Manganese market strategy. Technical report. 2019.
3. Haynes BW, Law SL, Maeda R. Updated process flowsheets for manganese nodule processing.
Bur Mines Inf Circ. 1983;8924:100.
4. Kojima K. Review of studies on manganese nodule processing. In: Ocean mining symposium,
International Society of Offshore and Polar Engineers, Seoul, Korea. November 24–26, 1997.
p. 19–22.
5. Agarwall JC, et al. Processing of ocean nodules: a technical and economic review.
JOM. 1976;28:24–31.
6. Pemchand, Jana RK. Processing of polymetallic sea nodules: an oveview. In: Ocean mining sym-
posium, International Society of Offshore and Polar Engineers, Goa, India. 1999. p. 237–45.
7. Hubred GL. Kennecott’s cuprion process for manganse nodules, marine minerals: crossroads of
science, engineering, and the environment. 35th Underwater Mining Institute, November 1–6,
2005, Califonia, USA.
8. Ochromowicz K, Aasly K, Kowalczuk PB. Recent advancements in metallurgical processing of
marine minerals. Minerals (Basel). 2021;11(12):1437.
9. Randhawa NS, Jana RK, Das NN. Silicomanganese production utilizing low grade manganese
nodules leaching residue. Trans Inst Min Metall C. 2013;122(1):6–14.
10. Sridhar R, et al. Extraction of copper, nickel and cobalt from sea nodules. JOM. 1976;28:32–7.
11. Sridar R, et al. United States Patent #4,049,438, Filed February 14, 1975.
12. Sommerfeld M, Friedmann D, Kuhn T, Friedrich B. “Zero-waste”: a sustainable approach on pyro-
metallurgical processing of manganese nodule slags. Fortschr Mineral. 2018;8:544.
13. Agarwall S, et al. Manganse distribution between alloy & slag phases in making Fe-Si-Mn alloy
from polymetallic nodules. Paper 164, IMPC 2012, New Delhi, India. p. 79–89.
14. Sahu KK, et al. Nickel, cobalt and copper recovery from sea nodules by direct smelting process.
In: Ni-Co 2013. The Minerals, Metals & Materials Society; 2013. p. 291–8.
15. Donald JR, McLachlan C. Development of a metallurgical process for polymetallic nodules. In:
Proceedings of COM 2017. 2017.
16. Kim PP. Reduction rates of SiMn slags from various raw materials. Paper, Norwegian University
of Science and Technology. 2018. p. 1.
Index
A Alternative reduction agents, 151–155

Ablation, 43–46, 83–85, 87–89 Alumina ceramics, 84, 86, 87
Acid bake, 691, 710, 712, 714, 715, 718, 719, Aluminium, 65, 67–70, 305, 307–309, 360, 395,
722–736 397, 461, 463, 493, 504, 506, 547,
Acid leaching, 516, 550, 591, 622, 662, 690, 629, 716, 718, 724, 726, 813, 917,
692, 710, 712, 723, 724, 728, 734, 929–931, 934, 995
735, 740, 756, 778, 798, 813, 814, Aluminium alloys, 306, 310, 311, 383–389, 493,
819, 838–845 495
Acid wastes, 811–813 Aluminium-based entropic alloys, 413–422
Acoustio ultrasonic-echo (AU-E), 183–186, Aluminium 6111 alloy, 426, 430, 432, 434,
188–192, 848–853, 855, 856 437–445
Activated carbons, 251–253, 265, 267 Aluminium smelters, 395–404, 930
Additive manufacturing (AM), 9–11, 25, 26, 39, Aluminium thin strips, 426, 430–432, 437–445
73, 74, 79, 92, 299–300, 302, 335, Anisotropy, 80, 344, 348, 353, 388, 389
343, 344, 355 Anode furnace, 211, 213, 215, 223, 224
Adsorption, 238, 251–253, 263, 265–267, 395, Aqueous Particle Sensor (APS III) System
397, 399, 623, 816, 817, 824, Asset management, 396, 404
829–831, 959, 960, 971, 972, 980 Astragaloside, 823–834
Advanced High Strength Steels (AHSS), 47–51 Asymmetrical rolling (ASR), 383, 384, 387–389
Ag addition, 270, 272–274, 276, 277 Atmospheric leaching (AL), 276, 754,
Agglomeration, 110, 114, 457, 477, 870, 777–785
996–997 Atomization, 299, 316, 897–906
Aging, 11, 15–17, 26, 74, 255, 256, 270–275, Austempering-tempering, 21–23
296, 298, 480, 481, 640 Automation, 401, 859, 866, 923, 924
Al alloys, 6–8, 17, 391, 410, 420–422, 448, 449, Auxetic, 355–357
475, 478–480, 483, 484, 494–496,
504
Al-20Ce, 471–473 B
Alloy development, 344, 413, 494, 504 Bastnaesite, 682–684, 686, 687, 690, 692, 696,
Alloying, 17, 29, 32, 36, 49, 81, 293, 295–297, 701, 703, 705, 727–729, 736
316, 337, 344, 345, 348, 351, 391, Battery, 106, 113, 255, 258–263, 401, 476, 493,
413, 429, 458, 471, 476, 478–480, 504, 550, 591, 601, 603, 604, 628,
483–485, 506, 996, 998 629, 631, 634, 645, 662, 676,
Al-Mg-Si conductor alloys, 269–277 737–751, 797, 860, 866, 875, 923,
Al-Si alloy, 479, 480 933, 950–952, 961, 979, 995, 998
https://doi.org/10.1007/978-3-031-17425-4
1008 Index
Bauxite residue (BR), 119–121, 123, 126, 127, Chelating ion, 260–262
811, 813, 815, 816, 819 Computational fluid dynamics (CFD) modelling,
Bio-surfactants, 907–909 307, 370, 372, 375–380, 426, 430,
Bio syngas, 146–148 432
Black mass, 261, 603, 604, 628, 630, 632 Characterization, 12, 29–36, 54, 55, 74–75, 78,
Brazilian Nickel, 550 79, 81, 95, 96, 152, 232, 237–242,
Brine, 259, 646, 649, 651, 738, 740, 744, 745, 255, 318, 340, 439–444, 489–491,
748, 751, 979 553, 682, 700–701, 779–782, 828,
Briquetting, 870, 871 838, 839, 845, 862, 881, 889–891,
Bubble fragmentation, 375, 377, 379, 380 982–985
Chlor alkali, 742, 744, 745, 751
Chloride leaching, 519, 646, 649, 651, 678,
C 787
Campaign Life extension, 175–192 Chromite, 178–179, 463–465, 468, 469
Canadian technology, 171–174 Circuit capacity, 618
Carbon, 21, 26, 30, 49, 53, 61, 74, 101, 109, Circular economy, 588, 880, 884, 906, 918
120, 129, 133, 145, 152, 158, 162, Circular recycling, 880, 884
171, 192, 193, 208, 212, 251, 265, Coal fly ash, 621–623
269, 294, 348, 360, 395, 453, 475, Coating, 25, 47–50, 297, 315–326, 361, 392,
500, 516, 550, 601, 628, 675, 738, 960
754, 781, 797, 860, 869, 875, 884, Cobalt, 174, 259, 262, 263, 516, 519, 521, 523,
905, 907, 918, 940, 1000 525, 550, 551, 628, 629, 632, 637,
Carbon capture, 54, 55, 159, 516, 525, 550, 918, 662–672, 675–678, 738, 740–742,
921 745, 748, 751, 753–772, 777–785,
Carbon capture and sequestration, 55, 56, 129 787–794, 797–807, 811, 949–952,
Carbon dioxide (CO2) 961, 993, 996–998
CO2-emissions, 65, 113, 114, 133–143, 145, Cobalt sulphate, 664, 788
152, 162, 169, 171, 172, 193, 200, Cold bonding, 869–874
503, 516, 550, 675, 798, 807, 875, Cold spray, 315–326
877, 917, 922–934, 942, 943, 1001, Comminution, 208, 543, 860, 876, 912, 914
1002, 1004 Compressive properties, 414, 418
CO2 sequestration, 516, 797, 801 Conductivity, 6, 60, 61, 105, 142, 255, 258, 285,
Carbon fibers (CF), 151–155 391, 457, 479–481, 586, 650, 664,
Carbon footprint, 101, 109, 120, 158, 192, 500, 946
525, 860, 866, 884, 926, 929, 930, Conversion, 55, 102, 113, 135, 136, 140, 146,
932, 933 153, 154, 159, 200, 201, 224, 231,
Carbon negative, 737–751 245, 266, 286, 316, 646–648, 650,
Carbon pricing, 877, 918 651, 724, 762, 788–794, 899, 906,
Carbonation, 54–57, 516, 519, 551, 553–558, 940, 980, 981, 992, 1002
798–802, 807 Cooling curve, 330, 331
Carbothermic reduction (CTR), 134, 152–154 Copper, 4, 5, 49, 59–62, 117, 152, 154, 155,
Carburization, 145–148, 193 171–174, 179, 185, 208, 211, 212,
Castability, 7, 8, 391, 414 215–217, 219, 220, 222–226,
Casting, 3–6, 16, 18, 91, 220, 232, 296, 302, 280–282, 284, 293, 305, 345, 347,
305, 306, 309, 311, 313, 329, 331, 414, 425, 427, 428, 430, 439, 441,
332, 344, 356, 357, 370, 425–435, 452, 464, 465, 468, 562, 583–588,
437–439, 444, 448, 452, 454, 753–772, 788, 823, 824, 833, 834,
458–460, 476, 479–482, 484, 860–863, 865, 875, 908, 909, 917,
494–498, 500, 504–510 922, 923, 932–934, 951–955,
Caustic crack, 692, 710–712, 719, 722–736 959–961, 970–975, 993, 996–998
Causticization, 646–649, 651, 726 Copper nanomaterials, 59–62
CCS/CCUS, 918, 925, 926, 929 Corrosion, 26, 47, 231–235, 237, 239, 240, 242,
Cellular lattice structures, 336 301, 302, 315, 316, 320, 343, 359,
Cement, 116, 121, 129, 152, 159, 172, 622, 360, 369, 371, 392, 428, 464–468,
737–751, 870, 872, 873, 880, 902, 928 471, 489, 495, 497, 545, 621, 968
Index 1009
Critical and strategic metals (CSMs), 675, 880, 785, 798, 804, 815, 818, 838–845,
882 970–972
Criticality, 396, 398, 399, 404, 637 Distribution, 10, 11, 16, 34, 65, 74, 79–81, 112,
Critical materials, 637, 640, 662, 738, 773, 952 134, 149, 232, 251, 257, 318, 319,
Critical solids density (CSD), 552–554 324, 337, 338, 396, 398–400, 406,
Cross-rolling (CR), 21–23, 384 407, 441–442, 452, 459, 476, 482,
Crystallographic texture, 336 489, 509, 543, 545, 567–571, 573,
Cuprion, 753–772 574, 578, 579, 611, 638, 775, 788,
Cuprite, 823–834 812, 864, 912, 914, 934, 945,
Cyanide, 279–289, 584, 641, 839, 840, 844, 879 983–988, 991, 992
Cyclic deformation, 15–19 DRI smelting, 110, 116, 117
Dynamic simulation, 608
D
Decanter centrifuge, 888, 889, 891–896 E
Decarbonization, 102, 131, 157–160, 201, 207, EDTA, 204, 205, 800–803
208, 737, 751, 773, 870–877, 918, Electric sensing zone–particle analyzer
919, 922–934 (ESZpas), 451–454, 456
Decarburization, 49, 50, 934 Electric vehicle (EV), 3–8, 259, 261, 447, 476,
Deep eutectic solvents (DESs), 561–564, 584, 503, 504, 601, 628, 629, 645, 661,
586, 676–678, 861, 863, 864 662, 738, 751, 797, 812, 875, 923,
Demand, 9, 26, 43, 84, 103, 109, 114, 152, 171, 933, 951, 979
174, 199, 200, 269, 395, 464, 547, Electrical conductivity, 4–8, 269–277, 391,
550, 584, 591, 628, 629, 634, 637, 478–480
645, 651, 663, 675, 715, 716, 730, Electrical heating, 103, 113, 159
731, 738, 751, 775, 777, 787, 797, Electricity generation, 919, 920, 922–925, 928
811–812, 819, 820, 859, 866, 867, Electrocatalysis, 245
875, 881, 882, 920–922, 925, 926, Electrodes, 49, 60, 62, 105, 106, 110, 116,
929, 930, 932, 933, 950–951, 961, 134–136, 142, 160, 177, 246, 255,
979, 995 280, 281, 284–287, 289, 295–297,
Dense media separation (DMS), 682–688, 690, 299, 406, 522, 562, 563, 602, 650,
691, 697, 912, 980–988, 990–992 748, 758, 930
Dental implants, 360, 361, 365–367 Electrolysis, 129, 159, 172, 194, 195, 200, 201,
Depressant, 706, 709, 719, 823–834, 957, 959, 280, 282, 284–289, 395, 396, 398,
960 516, 519, 523–525, 551, 553, 646,
Dewatering, 887–896 649–651, 744, 930, 945–947
Die soldering, 392–394 ElectroMagnetic Levitation (EML), 471–473
Direct Chill (DC) casting, 305 Electron microscopy, 74, 237, 270, 336, 489,
Direct energy deposition (DED), 30–36 491
Direct laser writing (DLW), 60, 62 Energy, 5, 9, 10, 30, 45, 51, 60–62, 65, 69,
Direct reduced iron (DRI), 110, 113, 115–117, 83–89, 102, 103, 105, 106, 110, 112,
145–149, 157–159, 172, 193, 113, 115, 129, 133, 134, 136, 159,
195–197, 199–201, 925, 927, 928 162–164, 168, 169, 171, 186–188,
Direct reduction, 109, 110, 113, 114, 116, 117, 195, 200, 208, 212, 231, 241, 245,
130, 131, 870–874 246, 251, 255, 257, 265, 266, 274,
Directed energy deposition (DED), 29–36, 298, 299, 316, 331, 355, 357, 377,
39–41, 335 409–411, 414, 430, 437, 441, 447,
Discrete event simulation (DES), 584, 587, 677, 448, 476–478, 480, 493, 503, 504,
678, 861, 864 506, 523, 525, 526, 550, 558, 566,
Dissolution, 94, 169, 205, 231, 233, 238, 241, 580, 584, 598, 601, 628, 629, 634,
242, 245, 246, 279, 280, 393, 465, 637, 645, 651, 661, 675, 738, 755,
466, 471, 480, 482, 504, 505, 507, 774, 778, 785, 797, 812, 819, 828,
508, 516, 521, 551, 563, 584, 831, 848, 860, 861, 863–866, 872,
586–588, 634, 646, 647, 649, 663, 873, 875–877, 883, 898–900, 907,
670, 671, 678, 710, 712, 716, 755, 909, 912, 913, 915, 917–926,
1010 Index
928–931, 934, 935, 939, 940, 942, G

944, 945, 951, 953, 998, 1001, 1002 Gas Metal Arc Welding (GMAW) process,
Energy consumption, 117, 131, 133, 141, 405–407
161–169, 176, 196–198, 437, Gas Treatment Center (GTC), 395–404
523–525, 651, 652, 726, 756, 859, Geology, 530–531, 952
863, 884, 919, 920, 932–933 Geostatistical simulation, 638
Environment, 55, 66, 102, 105, 120, 152, 155, Grade, 54, 74, 84, 110, 115, 146, 159, 172,
183, 185, 225, 237, 238, 240, 242, 246, 196, 204, 344, 345, 347–350, 353,
301, 316, 396, 398, 399, 402, 448–450, 360, 426, 438, 453, 454, 518, 535,
453, 456, 476, 575, 598, 602, 663, 737, 537, 541–543, 554, 567–574,
844, 851, 920, 948, 959, 967 576–579, 581, 592, 619, 638, 662,
Environmental sustainability, 476, 947 682, 684, 687, 690, 696, 697, 701,
Environmental uncertainty, 519, 861 706–709, 718, 719, 722, 726, 731,
Epoxy adhesive, 65–67, 69 743, 751, 774, 775, 823–826, 833,
Equal-channel angular pressing (ECAP), 5, 834, 859, 861, 863, 876, 887, 923,
361–364 932, 933, 952, 953, 959–961, 967,
Evaporative crystallization, 647, 648 968, 970, 972, 973, 980, 987,
Extractant, 592–597, 662–665, 670–672, 718, 989–991, 994, 997, 998, 1000, 1001,
770, 788, 818 1003, 1004
Extractive metallurgy, 812, 918 Grain refinement, 329, 331, 367, 388, 476–479,
482, 485
Grain size, 28, 80, 110, 297, 350, 361, 363, 365,
F 384, 385, 389, 429, 432, 437,
Fast weathering, 57 440–441, 444, 464, 476, 477, 482,
Fayalitic slag, 152 543, 701, 728
Fe-based metallic glass, 316, 319 Granulation, 897–906
Ferrochromium, 203–205 Greenhouse gas (GHG) emissions
Ferromanganese production, 162 Green hydrogen, 113, 131, 132, 159, 195, 199,
Finite-element (FE) model, 406, 407 200, 524, 925–927, 931
Fire refining, 211–226
Flexible devices, 43–46
Flexible electronics, 59–62 H
Flotation, 532, 539–541, 640, 706–710, 714, H13 tool steel, 29–36
716, 719, 722–736, 823–834, Hardness, 10, 11, 13, 22, 23, 27, 28, 30, 48, 49,
860–867, 887, 888, 907–909, 932, 78, 80, 92, 93, 95, 96, 261, 270–272,
933, 953, 955–961, 967–975, 275, 276, 296, 317, 321, 325, 362,
980–985, 987–992 363, 365, 366, 414, 432, 479, 496,
Flotation separations, 824, 834, 955, 959 532, 539, 900, 945, 946
Flow rate, 92, 102, 179, 214, 216, 217, 219, 220, HAZ softening
222, 224, 225, 262, 263, 282, 287, HBI, 113, 114, 117, 146
289, 370, 371, 374–377, 380, 430, Heap leaching, 517–521, 525, 550, 662
438, 456, 457, 522, 554, 622, 623, Heat treatment, 6, 7, 11–13, 15–18, 26–28, 31,
650, 766, 839, 892, 895, 944 34, 36, 49, 50, 73, 74, 76, 81, 92–96,
Fluid bed, 101–107, 110, 159 123, 138, 271, 296, 298, 325, 364,
Forging, 91, 296, 302 367, 384, 428, 429, 480–481, 494,
Formability, 364, 369, 383, 438, 497 504, 509
Fossil-free, 116, 145, 146 Heat-treated aluminum alloys, 409–411
Fractional factorial design, 779, 783, 785 Heavy liquid separation (HLS), 682–688, 697,
Fracture, 30, 51, 83, 238–242, 361–367, 409, 985
411, 417, 448, 494, 995 Heavy rare earth elements, 577
Frother flotation, 826, 907–909, 968 Heavy rare earths (HRE), 533, 534, 538, 543,
Fuel cell, 245, 246, 447, 448, 812, 923 547, 580, 700, 706
Furnace integrity monitoring System (FIMS), High pressure die casting (HPDC), 3, 6, 8, 16,
186–189, 192 504
Index 1011
High strength, 8, 17, 26, 47–51, 91, 297, 300, Ionometallurgy

336, 361, 367, 413–422, 426, 448, Iron and steel, 131, 172, 173, 448, 464, 870,
475, 495, 496, 591, 738, 872 872, 917, 925
Horizontal single belt casting (HSBC), 305–307, Iron ore, 105, 109, 110, 112–117, 130, 131, 145,
312, 425–435, 437–441, 443, 444 157, 159, 160, 172, 207, 774, 811,
Hot tearing, 7, 331 870, 872, 873, 922, 923, 1003
Hybrid additive/subtractive manufacturing, 10, Iron ore reduction, 157, 159
92, 96 Iron recovery, 120, 152
Hydrochloric acid (HCl), 678, 692, 695, 710, Iron segregation, 207, 208
712, 714, 723, 724, 745, 787, 789, Ironmaking, 145, 158, 160, 207, 208, 870, 872
791, 816, 825 Isotherm modelling, 266, 267
Hydrogen, 110, 112, 113, 117, 120, 121, 124, Isotherms, 126, 252, 266, 608, 611, 789–794,
126, 127, 129–132, 145, 146, 158, 816, 960
159, 172, 193–201, 208, 211–226,
232, 238, 245, 246, 447–450, 480,
481, 550, 561, 593, 602, 604, 650, K
676, 678, 737, 754, 778, 789, 918, Kimberlite, 911–914
925–929, 931, 934, 939, 940, 942, Kinetics, 5, 91, 129–131, 152, 162, 194,
946, 947, 959 199–201, 212, 232–235, 245, 246,
Hydrogen-based DRI, 926 252, 253, 266, 267, 280–282, 284,
Hydrogen economy, 918 347, 348, 377, 430, 465, 478,
Hydrogen embrittlement (HE), 448, 449 508–510, 516, 519, 551, 553, 555,
Hydrogen internal combustion engine (HICE), 557, 558, 586, 670, 785, 839, 969,
447–449 971, 972, 989
Hydrogen metallurgy, 129, 131 Kroll’s process, 295
Hydrometallurgy, 561, 583, 813–819, 881, 932,
980
L
Lanthanum(III), 591–597
I Laser cutting, 84
II-VI, 812–819 Laser powder bed fusion (LPBF/PBF-L), 9–11,
ImageJ, 372, 374 26, 27, 92, 94–96, 299, 300, 335–337,
Imaging, 31, 239, 241, 489, 491, 494, 574, 340, 343–353
852–856 Laser processing, 83
Impact load, 409–411 Laser reduction, 60, 61
Impulse atomization (IA), 471–473 Laterite, 550, 662, 777–785, 797–807, 933, 934,
Inclusions, 18, 19, 49, 198, 295, 369–380, 451, 950, 951
452, 454–461, 481, 482, 531, 785 Layer thickness, 10, 26, 29, 31, 32, 34, 39–41,
Inconel® 718, 91 92, 300
Increased demand, 259 Leaching, 120, 121, 204, 205, 279, 516, 518,
Inductively coupled plasma mass spectrometry 519, 521–523, 525, 550, 551, 564,
(ICP-MS), 245–246 591, 601–604, 607, 622, 634, 641,
Intelligent instruments, 401 646, 649, 676–678, 694, 695, 710,
Interface, 10–13, 26–28, 30, 49, 51, 69, 83, 148, 712, 714, 716, 723, 724, 728,
238, 242, 300, 301, 309, 320, 321, 738–746, 751, 754–756, 758, 762,
326, 347, 348, 364, 373, 420, 469, 765, 767, 771, 773, 778, 779,
662, 669, 670 782–785, 787, 788, 800–803,
Intermetallic compounds (IMCs), 294, 442–444 805–807, 811, 813–816, 819,
Intermetallic network, 421 837–845, 899, 900, 933
Investment casting, 335, 355–357 Li-garnet, 255–258
Ion exchange, 196, 260, 542, 546, 547, 550, 551, Light alloys, 476
607–612, 616, 619, 620, 622, 646, Light metals, 343, 434, 435, 489
695, 718, 741, 811 Light weighting, 65, 448, 481, 504, 510, 812
Ion exchange resin, 259, 261–263, 813 Li-ion conductivity, 257
Ionic strength, 967–972, 974 Linear circuit analysis, 861–863, 866
1012 Index
Liquid metal cleanliness analyser (LiMCA), 355, 361, 367, 369, 383, 384, 389,
451–456, 458–461 411, 414, 418, 420, 428, 429, 448,
Lithium, 255, 259–261, 484, 601–604, 628–632, 449, 471, 475–477, 480–484, 489,
637, 645–647, 650, 651, 675–678, 494, 495, 509
738, 875, 876, 979–992 Melting, 4, 5, 49, 80, 87–89, 92, 107, 130, 138,
Lithium chloride (LiCl), 646, 649, 651 146, 148, 149, 153, 157, 159, 179,
Lithium hydroxide (LiOH), 645–652, 980, 981, 204, 208, 293, 295–297, 299, 302,
992 316, 345, 346, 351–353, 392, 395,
Lithium-ion battery (LIB), 256, 628, 629, 631, 414, 426, 465, 477, 481, 924, 927,
634, 661, 662, 678, 860 929
Lithium-ion battery recycling, 661–672 Membrane electrodialysis, 651
Lithium sulfate, 646 Memristor, 61, 62
Low carbon, 26, 49, 51, 74, 102, 158, 171, 173, Meniscus, 306, 309–313, 430, 432, 433
877, 926, 931, 932 Metal deposition, 337, 340
Low carbon economy, 171–174, 675 Metal extraction, 396, 519, 521, 676–678, 731,
Low-modulus, 296, 359–367 734, 743, 745, 760–762, 764, 767,
Low temperatures, 12, 21, 113, 164, 196, 200, 769, 771, 784, 785, 801, 803–807
255, 293, 297, 298, 360, 478, 931 Metallurgy, 121, 183, 207, 293–295, 302, 335,
426, 519, 591
Mg-alloy, 231–235, 317, 321, 477, 482, 494,
M 506, 510
Machining, 92–94, 96, 296, 297, 300, 302, 336, Microbubble generator, 370–373, 380
392, 775 Micro-bubbles, 370
Magnesium, 18, 231, 261, 295, 315–326, 441, Microstructure, 11, 16, 19, 21–23, 26–28, 60,
443, 452, 456, 463, 465, 475, 481, 67, 68, 73–76, 78, 80, 81, 91–94, 96,
515–519, 521, 523, 525, 550, 551, 123, 125, 126, 232, 233, 235, 274,
553–555, 629, 634, 649, 692, 738, 294–300, 337–339, 344, 346,
741–743, 745, 748–751, 777, 787, 348–353, 360–367, 383, 384,
798–800, 802, 806, 807, 971 414–417, 419, 421, 428, 431–433,
Magnesium hydroxide, 516, 519, 523, 525, 437–440, 465, 468, 472, 473,
550–558, 738, 741, 745, 751 477–480, 482–484, 490, 494, 495,
Magnet recycling, 561 504, 505, 508
Maintenance strategy, 179–181, 183, 184, 396, Microwave, 105, 159, 676, 912–915
401–403, 926 Mineral carbonation, 55, 56, 515, 550, 798, 800,
Manganese, 162–165, 168, 259, 441, 444, 463, 801, 805, 807
516, 583, 628, 629, 634, 637, Mineral flotation, 826, 860, 861
639–641, 649, 718, 738, 741, 744, Mineral synthesis, 54
745, 748, 750, 751, 753–772, 837, Mineralogy, 55, 517–518, 531–532, 534, 536,
838, 876, 951, 993–1004 537, 539, 543, 639, 682, 700–705,
Manganese silicate, 997–1004 708, 726, 727, 743, 774, 881, 884,
Manganese-silver (Mn-Ag), 837–845 889, 912, 952–953, 967, 969, 983,
Manufacturing processes, 3–8, 11, 55, 65, 383 984, 1000
Maraging steel, 11, 25–28, 335–341 Mining, 119, 171, 172, 174, 259, 538, 543, 561,
Mass balance, 120, 133, 136, 164, 165, 630, 578, 598, 601, 638, 639, 641, 642,
631, 634, 664, 745, 760, 785, 861, 681, 722, 774, 775, 778, 859–861,
862 864–866, 871, 875–877, 879–884,
Materials characterization, 22, 29–36, 54, 74–75, 907–909, 911, 912, 915, 917–935,
78, 87, 92, 232, 237–242, 316, 318, 952, 953, 967–970, 972, 994
478, 489–491, 838, 889 Mining and metallurgy, 879–884
Mean time between failure (MTBF), 403 Mixed hydroxide precipitation, 745, 748–751
Mechanical interlocking, 69 Modelling, 55, 267, 306, 307, 312, 355, 356,
Mechanical joinability, 494 371–374, 427, 428, 430, 432, 498,
Mechanical properties, 10, 11, 13, 15, 19, 21–23, 577–582, 607, 665, 860, 861, 866
26, 27, 49, 51, 74, 80, 91, 95, 96, 203, Modified thermomechanical treatment (MTMT),
271, 296–298, 316, 344, 349, 352, 270, 271, 277
Index 1013
Monazite, 530, 701, 722–724, 726–730, 735, P

736, 774 Pack boriding, 30, 31, 35
Multi-material additive manufacturing, 25 Parisite, 682, 696, 701, 703, 705
Pelletizing, 110, 114, 120, 121, 159, 869, 872,
873, 941, 942
N Petroleum coke, 105, 106, 152, 153
Nanofabrication PH 13-8Mo, 74, 76
Nanomechanic, 10 Piaui, 550
Nanoscale solid state electrolyte, 255–258 Picosecond laser, 84, 87, 89
Nanostructured titanium, 361 Pilot, 106, 159, 162–169, 425, 426, 430, 435,
Naphthenic acids (NAs), 251–253, 265–267, 519, 526, 530–533, 535–537,
509, 742, 764 539–544, 547, 553, 567, 582, 640,
Natural ageing, 494, 504–510 641, 651, 682, 684–688, 694–697,
Negative emissions technology (NET), 53–57 722, 735, 745–751, 766, 767, 771,
Neodymium(III), 591–597 772, 774, 815, 816, 819, 889,
Net-zero, 129, 157–160, 173, 447, 876, 892–893, 898, 927, 930, 985,
917–919, 922, 924, 925, 929, 935 998–1002
Net-zero carbon, 797, 875–877 Pilot testing, 889
Ni-based superalloys, 91, 485 Piloting, 540, 765, 892
Nickel, 8, 171–174, 208, 259, 262, 263, 391, Planetary hydrogen (PH), 119
464, 516, 517, 519, 521–523, 525, Platinum, 178, 245–246, 970, 972
550, 551, 583–587, 628, 629, 632, Polymetallic nodules (PMN), 757–758,
637, 651, 661–672, 738–746, 748, 993–1004
751, 753–772, 777–785, 787–794, Potential sources, 646, 774
797–807, 811, 823, 875, 917, 923, Powder bed fusion (PBF), 9–13, 299, 300, 335
932–934, 940, 943, 947, 949–956, Power consumption, 136, 142, 143, 199
960, 961, 970, 974, 993, 995–998 Precipitates, 11, 16–18, 74, 75, 91, 94, 96, 269,
Nickel laterite, 628, 777, 778, 798 270, 272, 274–277, 350–352, 428,
Nickel sulphate, 672, 787–794 429, 443, 478, 490, 491, 505,
Nickel sulfate separation, 661–672 507–509, 523, 540, 542, 554, 563,
Nitrogen, 28, 105, 196, 212, 213, 215, 219, 220, 564, 647, 649, 692, 693, 695, 696,
222, 223, 225, 294, 317, 449, 481, 700, 718, 722, 725, 726, 738, 740,
649, 754, 944, 1000 741, 745, 746, 748–751, 754, 778,
Non-destructive testing (NDT), 183–190, 848, 798, 813, 815, 819, 947, 959, 996
851 Precipitation, 5, 10, 12, 13, 27, 54, 73, 75, 91,
Non-ferrous metal vessels 94, 96, 231, 260–262, 274, 295, 298,
301, 337, 347, 351–353, 360, 364,
494, 504–510, 516, 519, 522, 523,
O 525, 547, 550, 551, 553, 554, 628,
Ocean alkalinity enhancement, 516 629, 634, 646, 649, 662, 671, 672,
Oil sand process affected water (OSPW), 692, 694, 695, 697, 710, 718, 719,
251–253, 265, 267 724–726, 740, 741, 744–751, 760,
Olivine, 205, 467, 517, 738, 739, 742, 743, 751, 800, 807, 811, 813, 815, 817, 819,
798–802, 807 844, 969, 971
Operating furnaces monitoring, 847–856 Predictive maintenance, 176, 181–184, 855, 856,
Optimization, 92, 162, 253, 326, 367, 522, 523, 926
556, 558, 638, 639, 641, 715, 718, Preventive maintenance (PM), 181, 396,
760, 782, 785, 838, 876 401–403
Ore pretreatment, 161–169 Process flowsheets, 516, 530, 547, 551, 647,
Ore sorting, 534–538, 543, 565–569, 571, 573, 652, 726, 736, 738–742, 788, 940
577, 579–582, 980 Process parameters, 40, 45, 54, 84, 89, 300, 305,
Out of plane bending, 409–411 306, 317, 433, 435, 480
Overaging, 15, 17–19, 272 Processing challenges, 949–961
Oxidising agents, 584–588 Productions and recycling trends, 951
1014 Index
Project development, 518, 525, 526, 540–541, Renewable, 113, 171, 198, 208, 737, 860,
775, 998 907–909, 919, 923–925, 930, 931,
Pyrometallurgy, 584, 798, 881, 932, 997 934, 943
Pyrrhotite rejection update, 953 Renewable energy, 112, 171, 503, 662, 773, 860,
865, 866, 877, 919
Residual stress, 11, 28, 31, 83, 405–407
R Resin, 152–154, 178, 259–263, 356, 585,
Rapid solidification, 429, 504 608–619, 622, 623, 625, 695, 718,
Rare earth elements (REEs), 119, 120, 317, 471, 741, 813, 815–817, 819
534, 543, 547, 576, 591–593, Resistance spot welding, 49
595–597, 621, 622, 681, 682, Resistive switching, 61
690–697, 700, 701, 703–707, 710, Resistivity, 4, 5, 50, 60, 61, 134, 135, 137–143,
714–716, 718, 719, 722–726, 270, 274, 456, 478
728–731, 733, 735, 736, 773–775, Resources, 107, 171, 181, 183, 204, 403, 404,
777, 813, 819, 951 517, 519, 531, 540, 550, 565–567,
Rare-earth recycling, 561–564 571, 583, 591, 598, 601, 625,
Rare earths, 207, 530, 531, 533, 539–543, 547, 675–678, 681, 682, 722, 742, 773,
561, 563, 571, 591, 621, 682, 689, 774, 777, 787, 797, 798, 864, 865,
691, 696, 697, 700, 701, 703, 710, 876, 880, 883, 897–906, 928, 933,
714, 715, 718, 719, 722, 725, 738, 943, 950, 951, 972, 973, 979, 994,
773–775, 818 996
Recycling, 116, 129, 152, 154, 155, 157, 158, Reutilization
175, 251, 252, 259, 261, 265, Rheometallurgy, 552
295–297, 302, 547, 561, 604, 628, Risk, 66, 84, 110, 176, 177, 179, 184, 185, 192,
629, 631, 633, 634, 647, 649, 652, 196, 214, 295, 331, 361, 398, 399,
662, 741, 748, 774, 775, 787, 789, 477, 518, 519, 553, 637, 638, 642,
863, 869–871, 873, 880, 881, 883, 726, 775, 812, 818, 861, 902, 905,
884, 918, 925, 931, 934, 951, 961, 919, 945–948, 1001
967–975 Risk-based decision making, 184
Reduction, 4, 10, 22, 47, 50, 53, 60–62, 65, 109, Roasting, 646, 682, 691, 692, 694, 723, 778,
110, 112, 113, 115–117, 120–122, 780, 781
124, 126, 127, 130, 131, 145, Rotation, 92, 371, 555
152–155, 157–159, 164–167, 169, Run-out, 176–179, 183, 847
175, 182, 199–201, 212–214,
216–220, 222, 231, 232, 234, 246,
270, 271, 315, 350, 357, 373, 376, S
377, 379, 384, 385, 387, 395, 396, Salt anodization, 67, 70
437, 477, 478, 482–484, 503, 504, Scale up, 155, 889, 893, 902
507, 508, 521, 525, 534, 539, 555, Scandium, 121, 621, 622, 777–785, 811–812,
579, 584, 619, 692, 718, 724, 744, 876
754–756, 758, 760, 762–765, 771, Scandium recovery, 811–820
778–782, 784–785, 807, 840, 869, Scanning electron microscopy (SEM), 10, 16,
871, 877, 913, 914, 918, 924–931, 31, 66–68, 74, 75, 79, 92, 93, 121,
933, 934, 940, 943, 948, 951, 954, 123–125, 232, 233, 239, 240, 316,
961, 997, 998, 1001 318, 319, 322, 347, 349, 351, 362,
Reduction pre-treatment, 777–785 363, 365–367, 414, 416, 417, 419,
Refractories, 106, 117, 175–192, 295, 306, 430, 438, 440–442, 449, 489, 531, 534,
438, 439, 463–465, 467–469, 517, 535, 701, 705, 780, 999
690, 847, 848, 850, 852–856, 955 Scanning flow cell, 246
Refractory waste, 175 Scratch test, 74, 75
Reliability, 180, 184, 196, 197, 398, 498, 895, Scrubbing, 397, 516, 546, 550, 551, 790, 819,
913 982
Reliability centered maintenance (RCM), 396, Seabed nodules, 753–772
401–404 Seasonal variation, 969, 972
Remote mining, 860, 866 Segregation roasting, 207, 208
Index 1015
Selective nickel and cobalt extraction, 800, 801, Solvent extraction, 260, 547, 591, 592, 607, 662,
803 678, 718, 755, 770, 787–794, 811,
Selective reduction, 778 932, 949, 951
Selectivity, 260, 262, 584, 592, 593, 595–597, Solvometallurgy, 678
603, 604, 622, 663, 664, 666, 669, Spodumene concentrate, 646, 979–992
672, 675, 676, 678, 708, 709, 719, Stability, 66, 91, 208, 246, 255, 306, 311, 313,
741, 751, 798, 803, 824, 887, 954, 347, 348, 370, 391, 419, 432, 433,
957, 958, 970, 972, 980 477, 552, 553, 685, 748, 771, 798,
Self-clusters, 506–510 801, 899, 905, 947, 959, 970, 971
Self-piercing riveting (SPR), 494–500 Stainless steel, 31, 73, 76, 135, 370, 371, 662,
Severe plastic deformation (SPD), 5, 361 800, 933, 934, 950, 951, 961
Silica, 93, 106, 107, 123, 336, 453, 456, 466, Steel, 9–13, 21, 22, 26, 30, 31, 47–51, 55, 109,
468, 518, 530, 531, 539, 543, 738, 110, 112–117, 129, 130, 154, 155,
740, 742, 744, 745, 751, 778, 779, 157–160, 171–173, 193–201, 223,
782–785, 813–815, 819, 997, 998, 224, 231, 270, 280, 297, 305, 337,
1001 340, 356, 360, 369, 370, 383, 392,
Silicate minerals, 53–55, 57, 738, 739, 798, 800, 425, 426, 430, 438, 452–454, 456,
802, 807, 960 457, 459, 461, 463, 464, 475, 481,
Silicon, 30, 32, 117, 133, 134, 293, 344, 345, 483, 484, 493, 495, 496, 498, 504,
347–353, 391, 439, 441, 442, 444, 534, 869–873, 877, 897–899, 901,
463, 479, 483, 718, 726, 742, 780, 902, 904–905, 924–929, 933, 934,
813, 1000, 1001 995, 996, 999–1001
Silicon production, 134 Steelmaking, 109, 110, 113, 115, 117, 130,
Silicon-manganese alloy (SiMn), 993–1004 157–160, 207, 459, 869–874, 925,
Silver extraction, 837–845 926, 999, 1004
Sintering, 54, 102, 104, 105, 110, 120, 121, Stellite alloy, 25–28
123–126, 255, 257, 299, 869, 870, Stereolithography (SLA), 355–357
872, 873 Strategic mine planning algorithm, 637–642
Size distribution, 92, 94, 113, 154, 253, 257, Strength, 5–7, 15, 17, 18, 21–23, 26, 51, 66–70,
317–319, 369, 451, 452, 454, 455, 83, 117, 130, 180, 231, 235, 269–277,
458, 460, 465, 692, 701, 838, 873, 295, 298, 300, 302, 315, 343, 355,
881, 888–893, 912 359, 360, 362–367, 369, 384, 391,
Slag, 55, 110, 115–117, 120, 129, 151–155, 159, 414, 417, 418, 420, 421, 432, 465,
163–166, 168, 169, 175, 177, 466, 468, 475–477, 481–485, 494,
203–206, 220, 222, 295, 370, 496, 497, 504–510, 619, 671, 723,
464–468, 631, 869, 873, 897–906, 724, 740, 819, 850, 870, 873, 902,
933, 941, 955, 997–1004 904
Slag corrosion, 464–466 Strengthening mechanisms, 5, 270
Slag valorization, 152 Stress corrosion cracking (SCC), 15, 237–242
SO2 pretreatment, 837–845 Structural adhesive, 65, 66
Sodium hydroxide, 554, 631, 692, 741, 745, Sulphide concentrate, 787–794
982 Supported liquid membranes (SLMs), 346, 592,
Solid fraction curve, 332 595, 596, 662–664, 670–672
Solid wastes, 645, 646, 651, 738, 870, 873, 880, Surface profile, 83–89
998 Surface roughness, 31, 32, 39–41, 66–70, 74, 86,
Solidification, 4, 41, 74, 92, 299, 300, 329, 331, 87, 89, 93, 94, 96, 427, 428
332, 336, 337, 340, 341, 344–353, Surfactants, 555, 907–909
426–428, 438, 471, 472, 479, 480, Sustainability, 261, 437, 476, 485, 847, 870, 876,
483, 504, 506, 508 880, 908, 911, 912, 949–961, 968
Solid-state, 54, 120, 130, 162, 316, 318, 330, Sustainable operation
340, 591 Sustainable processing, 675
Solid-state battery, 255–258 System dynamics, 859–867
Solvent, 255, 262, 562–564, 584, 586, 592, 650, System flexibility, 859–867
651, 676, 678, 770 Synchysite, 701, 703, 705, 722, 726–728, 736
1016 Index
T Valorization, 119, 120, 151–155, 880, 881, 884,

Tailings, 251, 265, 279, 515–519, 547, 621, 687, 899–901
689, 692, 738, 739, 742–744, 751,
774, 811, 812, 816–819, 826, 837,
838, 845, 861, 863, 888–889, 896, W
911, 912, 914, 958, 959, 967, 969, Waste, 92, 116, 152, 159, 175–192, 197,
974, 980, 982, 985–987 203–205, 252, 279–289, 343, 535,
Talc, 516, 798, 823–834, 955, 959–961 547, 561, 564, 568–569, 583, 587,
Techno-economic analysis (TEA), 629, 662, 672 603, 604, 621, 622, 675, 738, 774,
Tensile behavior, 23 775, 778, 787, 812–817, 819, 847,
Tensile properties, 15, 17, 96, 477, 504–508 861, 869–874, 877, 879–884, 888,
Texture, 21–23, 26, 68, 69, 300, 335–341, 348, 900, 906, 911, 933, 944, 981, 983,
351, 383, 384, 388, 389, 489, 574, 996, 997
639, 894 Waste solids, 812, 871, 873
Thermal analysis, 329–332 Waste treatment, 280, 880, 881
Thermal conductivity, 15, 131, 296, 427, 471, Water management, 967
478–481 Water quality, 967–975
Thermal stability, 355, 414, 419, 420, 422 Water recycling, 967
Thermomechanical modeling, 270, 271, 274 Wear, 30, 112, 177, 179, 180, 185–188, 190,
Thin films, 43–45, 491 191, 316, 325, 360, 362, 464, 468,
Thin strips, 305, 425–427, 430, 431, 434, 438 469, 847, 853, 855, 856
Thin strips/sheets, 305, 425–427, 430–432, 434, Wear resistance, 26, 30, 33–34, 316, 355
437–444 Weld thermal cycles, 405, 406
3D printing, 336 Wilfley table, 682, 684, 687–691, 697
Ti 13Nb 13Zr (TNZ), 359–367 Wire arc additive manufacturing (WAAM), 73,
Titanium, 12, 13, 79, 119, 207, 293–297, 74, 76
299–302, 343–347, 353, 359–362,
461, 463, 811
Titanium alloys, 29, 79, 80, 293–301, 343–353, X
359–360, 367 Xenotime, 530, 531, 533–535, 537–539,
Titanium ore processing, 293–302 543–545, 547, 566, 574, 577, 579,
Tool steel, 30, 39–41 580, 582, 701–703, 705, 706, 710,
Transformative technology, 738 712, 719, 774
Turbulence shear, 380
Z
U Zero carbon emissions, 797
Uranium, 542, 543, 546, 547, 575, 607, 610, Zero curve, 329–331
611, 615–619, 692, 718, 719, 774, Zero waste flowsheet, 998
811, 880 Zinc, 154, 172, 173, 262, 263, 452, 463, 583,
Urban mining, 675, 678 765–767, 770, 771, 873, 875,
Utilization, 112, 129–131, 151, 155, 162, 181, 972–974
183, 185, 197, 198, 203–205, 608, Zones, 32, 33, 45, 47–51, 84, 89, 135, 138, 161,
615, 619, 620, 905, 918, 923, 950 162, 165, 187–189, 242, 270, 272,
274, 275, 308–312, 331, 430, 452,
455, 457, 473, 507, 508, 553, 651,
V 681, 690, 692, 694, 696, 753, 774,
777, 893–895, 952, 959, 993–995

COM 2022 Proceedings

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

COM 2022 Proceedings

Uploaded by

Copyright:

Available Formats

Proceedings

ISBN 978-3-031-17424-7 ISBN 978-3-031-17425-4 (eBook)

Janice Zinck, Geoscience and Mines, Government of Nova Scotia

Roki Fukuzawa, Hatch

Sidney Omelon, McGill University

“Big Idea” Transformational R&D Plenary Chair

Sumanth Shankar, McMaster University

Advances in Materials Manufacturing VI – Existing and Emerging Materials

Javad Gholipour Baradari, National Research Council Canada

John Goode, J.R. Goode and Associates

Part I Advances in Materials Manufacturing VI – Existing and Emerging

Al Alloys and Manufacturing Processes for Lightweight Applications

Laser Powder Bed Fusion of M789 Steel on Wrought N709

Fabrication of Thin Rotary Tools Through a Combination

Heat Treatment of Multi-Material Additively Manufactured

Microstructural Characterization of Pack-Borided H13 Tool Steel

Synthesis of Fast-Weathering Nutrient-Doped Silicate Minerals

Direct Laser Writing of Copper/Copper Oxide Patterns for Emerging

Open Pore Effect on Structural Adhesive Joining Using

Tribological Performance of PH 13-8Mo Maraging Stainless Steel

High-Speed Nanoindentation Mapping of Additive Manufactured

Part II Deep Decarbonization Pathways for Pyrometallurgical Processes:

Electrically Heated Fluid Bed as a Low-Carbon Option

High-Temperature Hydrogen Reduction of Bauxite Residue

 Hydrogen the Silver Bullet for Decarbonization?������������������������������������������ 129

Electrical Resistivity of Charge Materials in the Si Process

Utilization of Carbon Fibers (CF) as an Alternative Reduction Agent

Deep Decarbonization and Pathways to Net Zero Emissions:

Technology Developments and the Metals Industry

Predictive Furnace Refractory Maintenance Procedures

Stabilizing Chromium in Ferrochromium Slag for Utilization

Iron Segregation Roasting as a Potential Decarbonization Pathway

Part III Electrochemical Degradation of Multi-­Component Materials

Studying Stress Corrosion Cracking Mechanisms Using Novel

On-Line ICP-MS in Electrocatalysis Research: Platinum

Part IV Energy and Environmental Materials

Adsorption of Naphthenic Acids from Oil Sand Process-Affected

Crystallization of Nanoscale Garnet-Type Li-Ion Conductors

Study of Adsorption Thermodynamics of Model Naphthenic Acids

Enhanced Strength and Electrical Conductivity of the Al-Mg-Si

Part V Light Metals for the Transportation Industry

Titanium: From Ore Extraction and Processing to Its Applications

HSBC Back-Meniscus Stability for a Double-­Impingement Liquid

Development of Fe-Based Metallic Glass Coating on Magnesium

 pplications of Advanced Thermal Analysis for Aluminum Foundries ���������� 329

Microstructure and Texture in Additively Manufactured Maraging

Aluminum Auxetic Structures Produced Using Combined

A Novel Experimental Set-up for Generating Microbubbles

Castability of Al-Fe-Ni Alloys������������������������������������������������������������������������������� 391

Maintenance Strategy for Reliable Instruments in Gas Treatment

Experimental and Numerical Study of 3003-H14 Aluminum

Out-of-Plane Bending Assessment of Heat-Treated Aluminum

A Study on Aluminum-Based Lightweight Entropic Alloys

The Development of the HSBC (Horizontal Single-­Belt Casting)

Scanning Electron Microscopy Analyses of Horizontal Single Belt

Development of LiMCA (Liquid Metal Cleanliness Analyzer)

Microstructural Study of Containerless Solidification

Processing, Properties, and Applications of Lightweight Al

Self-Piercing Riveting of High-Ductility Al–Zn–Mg Casting Alloy

Natural Ageing of Al–Fe(Zn, Mg) Alloy Castings for Structural

Part VI Processing of Critical Materials

Integrating Carbon Capture in Mining Through Metallurgy. Part 1:

Developing a Commercial Heavy Rare Earth Processing Facility

Integrating Carbon Capture in Mining Through Metallurgy. Part 2:

Part I Advances in Materials Manufacturing VI – Existing and Emerging

Al Alloys and Manufacturing Processes for Lightweight Applications

Laser Powder Bed Fusion of M789 Steel on Wrought N709

Fabrication of Thin Rotary Tools Through a Combination

Heat Treatment of Multi-Material Additively Manufactured

Microstructural Characterization of Pack-Borided H13 Tool Steel

Synthesis of Fast-Weathering Nutrient-Doped Silicate Minerals

Direct Laser Writing of Copper/Copper Oxide Patterns for Emerging

Open Pore Effect on Structural Adhesive Joining Using

Tribological Performance of PH 13-8Mo Maraging Stainless Steel

High-Speed Nanoindentation Mapping of Additive Manufactured

Part II Deep Decarbonization Pathways for Pyrometallurgical Processes:

Electrically Heated Fluid Bed as a Low-Carbon Option

High-Temperature Hydrogen Reduction of Bauxite Residue

Hydrogen the Silver Bullet for Decarbonization?�� 129

Electrical Resistivity of Charge Materials in the Si Process

Utilization of Carbon Fibers (CF) as an Alternative Reduction Agent

Deep Decarbonization and Pathways to Net Zero Emissions:

Technology Developments and the Metals Industry

Predictive Furnace Refractory Maintenance Procedures

Stabilizing Chromium in Ferrochromium Slag for Utilization

Iron Segregation Roasting as a Potential Decarbonization Pathway

Part III Electrochemical Degradation of Multi-Component Materials

Studying Stress Corrosion Cracking Mechanisms Using Novel

On-Line ICP-MS in Electrocatalysis Research: Platinum

Adsorption of Naphthenic Acids from Oil Sand Process-Affected

Crystallization of Nanoscale Garnet-Type Li-Ion Conductors

Study of Adsorption Thermodynamics of Model Naphthenic Acids

Enhanced Strength and Electrical Conductivity of the Al-Mg-Si

Titanium: From Ore Extraction and Processing to Its Applications

HSBC Back-Meniscus Stability for a Double-Impingement Liquid

Development of Fe-Based Metallic Glass Coating on Magnesium

pplications of Advanced Thermal Analysis for Aluminum Foundries �� 329

Microstructure and Texture in Additively Manufactured Maraging

Aluminum Auxetic Structures Produced Using Combined

A Novel Experimental Set-up for Generating Microbubbles

Castability of Al-Fe-Ni Alloys�� 391

Maintenance Strategy for Reliable Instruments in Gas Treatment

Experimental and Numerical Study of 3003-H14 Aluminum

Out-of-Plane Bending Assessment of Heat-Treated Aluminum

A Study on Aluminum-Based Lightweight Entropic Alloys

The Development of the HSBC (Horizontal Single-Belt Casting)

Scanning Electron Microscopy Analyses of Horizontal Single Belt

Development of LiMCA (Liquid Metal Cleanliness Analyzer)

Microstructural Study of Containerless Solidification

Processing, Properties, and Applications of Lightweight Al

Self-Piercing Riveting of High-Ductility Al–Zn–Mg Casting Alloy

Natural Ageing of Al–Fe(Zn, Mg) Alloy Castings for Structural

Integrating Carbon Capture in Mining Through Metallurgy. Part 1:

Developing a Commercial Heavy Rare Earth Processing Facility

Integrating Carbon Capture in Mining Through Metallurgy. Part 2:

Deep Eutectic Solvents for the Recovery of Metals: Electrodissolution

Neodymium(III) and Lanthanum(III) Separation by a Flat-Sheet

Process Development for Selective Recovery of Lithium from Black

Processing Black Mass into a Mixed Nickel/Cobalt Hydroxide

Detailing of Mineral Processing Operations Within Strategic

Potential Processes for Producing High-Purity Lithium Hydroxide:

Cobalt–Nickel Separations Using Supported Liquid Membranes

On the Solvometallurgical Extraction of Lithium and Cobalt

The Beneficiation and Metallurgical Process for the Production

The Beneficiation and Hydrometallurgical Process Development

Hydrometallurgical Flowsheet Options for Treatment

Magnet Rare Earth Production Must Double in Ten Years.

Extraction of Nickel, Cobalt, and Scandium from Low-Grade

Innovative Solvent Extraction to Produce High-Purity Cobalt

The Extraction of Nickel and Cobalt from Laterite Ores

Selective Depression of Talc in Cuprite Sulfidation Flotation

Alternative Processing for Silver Extraction from High

Dynamic Renewable Energy-Driven Framework Development

Development of a Smart Database for Wastes from the Canadian

Decanter Centrifuges and Dewatering Technology Selection:

Using Bio-Based Surfactants as Frothers in Froth Flotation

Decarbonization at the Anglo-American Barro Alto Smelter

Ni/Co–Cu Sulphide Ores, Key Contributors to Sustainability:

Water Recycling and Seasonal Water Quality Effects

Production of Spodumene Concentrate from the North Carolina

Using Polymetallic Nodules to Innovate Manganese Markets:

S So Sgs Sss Sdis Sp , (1)

2 Materials and Methods

3 Results and Discussion