The Dispersion and Selective Flocculation of Hematite The Dispersion and Selective Flocculation of Hematite ORE ORE

Download as pdf or txt
Download as pdf or txt
You are on page 1of 313

Michigan Technological University

Digital Commons @ Michigan Tech

Dissertations, Master's Theses and Master's Dissertations, Master's Theses and Master's
Reports - Open Reports

2015

THE DISPERSION AND SELECTIVE FLOCCULATION OF HEMATITE


ORE
Howard James Haselhuhn III
Michigan Technological University

Follow this and additional works at: https://digitalcommons.mtu.edu/etds

Part of the Chemical Engineering Commons, and the Mining Engineering Commons
Copyright 2015 Howard James Haselhuhn III

Recommended Citation
Haselhuhn, Howard James III, "THE DISPERSION AND SELECTIVE FLOCCULATION OF HEMATITE ORE",
Dissertation, Michigan Technological University, 2015.
https://digitalcommons.mtu.edu/etds/945

Follow this and additional works at: https://digitalcommons.mtu.edu/etds


Part of the Chemical Engineering Commons, and the Mining Engineering Commons
THE DISPERSION AND SELECTIVE
FLOCCULATION OF HEMATITE ORE

By

Howard James Haselhuhn III

A DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

In Chemical Engineering

MICHIGAN TECHNOLOGICAL UNIVERSITY

2015

© 2015 Howard James Haselhuhn III


This dissertation has been approved in partial fulfillment of the requirements for the
Degree of DOCTOR OF PHILOSOPHY in Chemical Engineering

Department of Chemical Engineering

Thesis Advisor: Dr. S. Komar Kawatra

Committee Member: Dr. Tony Rogers

Committee Member: Dr. Caryn Heldt

Committee Member: Dr. Paul Sanders

Department Chair: Dr. S. Komar Kawatra


Table of Contents

List of Figures .................................................................................................................... ix

List of Tables .................................................................................................................. xvii

Preface............................................................................................................................ xviii

Acknowledgements ........................................................................................................... xx

Abstract ............................................................................................................................ xxi

Introduction .................................................................................................................... xxiii

Chapter 1 Introduction ................................................................................................ xxv

Chapter 2 Introduction ................................................................................................ xxv

Chapter 3 Introduction ............................................................................................... xxvi

1 The Role of Surface Chemistry in Iron Ore Beneficiation .......................................... 1

1.1 Iron Ore Beneficiation Processes ........................................................................ 1

1.2 Relationship of Water Chemistry, Surface Chemistry and the Selective

Flocculation and Dispersion Process .............................................................................. 8

1.2.1 Dissolution of Iron Ore Minerals .................................................................... 8

1.2.2 Ionic Contributions from Chemical Reagents................................................. 9

1.2.3 Ionic Contributions from Fresh and Re-use water in Concentrator .............. 12

1.3 Surface Chemistry of Iron Ore .......................................................................... 13

1.3.1 Surface Speciation and Surface Potential ..................................................... 16

1.3.2 Ion Adsorption Phenomena at Mineral Oxide Surfaces ............................... 24

1.4 References ......................................................................................................... 29

v
2 Effects of Water Chemistry on Dispersion and Selective Flocculation of Hematite

Ore ............................................................................................................................. 33

2.1 Water Chemistry Effects on Zeta Potential of Concentrated Hematite Ore ..... 33

2.1.1 Abstract ......................................................................................................... 33

2.1.2 Introduction ................................................................................................... 34

2.1.3 Background ................................................................................................... 35

2.1.4 Materials and Methods .................................................................................. 39

2.1.5 Results and Discussion ................................................................................. 40

2.1.6 Conclusions ................................................................................................... 43

2.1.7 References ..................................................................................................... 44

2.2 The Role of Water Chemistry in the Selective Flocculation and Dispersion of

Iron Ore at a Pilot Scale ................................................................................................ 47

2.2.1 Abstract ......................................................................................................... 47

2.2.2 Introduction ................................................................................................... 48

2.2.3 Materials and Methods .................................................................................. 53

2.2.4 Results and Discussion ................................................................................. 60

2.2.5 Discussion, Analysis and Conclusions ......................................................... 73

2.2.6 References ..................................................................................................... 74

2.3 Measurement of Zeta Potential and Correlation with Performance throughout

an Operating Hematite Beneficiation Plant .................................................................. 79

2.3.1 Executive Summary ...................................................................................... 79

2.3.2 Introduction ................................................................................................... 81

vi
2.3.3 Background ................................................................................................... 82

2.3.4 Methods and Materials .................................................................................. 93

2.3.5 Results and Discussion ............................................................................... 105

2.3.6 Extended Deslime Thickener Analysis ....................................................... 182

2.3.7 References ................................................................................................... 199

3 Effects of Reagent Selection on Dispersion and Selective Flocculation of Hematite

Ore ........................................................................................................................... 203

3.1 Effects of Dispersant Adsorption on Settling Behavior of Iron Ore ............... 203

3.1.1 Abstract ....................................................................................................... 203

3.1.2 Introduction ................................................................................................. 204

3.1.3 Materials and Methods ................................................................................ 212

3.1.4 Results and Discussion ............................................................................... 214

3.1.5 Conclusions ................................................................................................. 224

3.1.6 References ................................................................................................... 226

3.2 Comparison of Dispersants for Selective Flocculation of Hematite Ore at a

Pilot Scale ................................................................................................................... 229

3.2.1 Abstract ....................................................................................................... 229

3.2.2 Introduction ................................................................................................. 229

3.2.3 Materials and Methods ................................................................................ 233

3.2.4 Results and Discussion ............................................................................... 239

3.2.5 Conclusions ................................................................................................. 244

3.2.6 References ................................................................................................... 245

vii
Conclusions ..................................................................................................................... 249

Appendix A – Zeta Potential Analyses from Plant Scale Studies .................................. 251

Appendix B - Thermogravimetric Analysis for Plant Scale Studies .............................. 263

Appendix C – Particle Size Analyses for Plant Scale Studies ........................................ 275

viii
List of Figures

Figure 1.1: Example magnetite benefication process flow sheet ........................................ 3

Figure 1.2: Example hematite beneficiation process flow sheet......................................... 5

Figure 1.3: Dry processing of concentrated iron ores ......................................................... 7

Figure 1.4: Water-solid interface of a hydrated iron oxide. .............................................. 15

Figure 1.5: Distribution of positively, neutrally and negatively charged surface sites..... 22

Figure 1.6: Surface potentials as a function of pH............................................................ 24

Figure 1.7: Inner-sphere complex speciation at particle surfaces ..................................... 26

Figure 1.8: Outer-sphere complex speciation at particle surfaces .................................... 28

Figure 2.1: (a) Un-hydrated and (b) hydrated hematite surface structure......................... 37

Figure 2.2: Adsorption isotherms ..................................................................................... 38

Figure 2.3: Zeta potential of concentrated natural hematite ore ....................................... 41

Figure 2.4: Inner-sphere complexes of magnesium, calcium, strontium and barium ....... 42

Figure 2.5: Zeta potential of silica, synthetic hematite and natural hematite ................... 52

Figure 2.6: X-ray diffraction pattern for iron ore ............................................................. 56

Figure 2.7: Diagram of the continuous pilot scale deslime thickener............................... 58

Figure 2.8: Iron grade versus recovery plot for pH experiments. ..................................... 61

Figure 2.9: Phosphorus grade versus rejection plot for pH modification experiments. ... 62

Figure 2.10: The relationship between zeta potential of feed, underflow and overflow

samples and pH ..................................................................................................... 63

Figure 2.11: Iron grade versus recovery plot for sodium concentration modification

experiments. .......................................................................................................... 64
ix
Figure 2.12: Phosphorus grade versus rejection plot for sodium concentration

modification experiments...................................................................................... 65

Figure 2.13: The relationship between zeta potential for feed, underflow and overflow

samples and sodium concentration. ...................................................................... 66

Figure 2.14: Iron grade versus recovery plot for calcium concentration modification

experiments. .......................................................................................................... 67

Figure 2.15: Phosphorus grade versus rejection plot for calcium concentration

modification experiments...................................................................................... 68

Figure 2.16: The relationship between zeta potential for feed, underflow and overflow

samples and calcium concentration. ..................................................................... 69

Figure 2.17: Iron grade versus recovery plot for magnesium concentration modification

experiments. .......................................................................................................... 70

Figure 2.18: Phosphorus grade versus rejection plot for magnesium concentration

modification experiments...................................................................................... 71

Figure 2.19: The relationship between Zeta potential for feed, underflow and overflow

samples and magnesium concentration. ................................................................ 72

Figure 2.20: Adsorption isotherms ................................................................................... 84

Figure 2.21: Zeta potential of concentrated natural hematite ore ..................................... 85

Figure 2.22: Hematite surface structure frontal view and side view. ............................... 87

Figure 2.23: Distribution of ferric hydroxide complexes as a function of pH .................. 89

Figure 2.24: (a) Molar equilibrium distribution of magnesium species in process water.

(b) Molar equilibrium distribution of calcium species in process water. ............. 91

x
Figure 2.25: Molar distribution of ionic species by valency. ........................................... 93

Figure 2.26: Original sampling locations chosen by Dr. Joshua Carlson in 2007 ............ 94

Figure 2.27: Sampling locations in full process flow diagram ......................................... 95

Figure 2.28: pH of each sample by location on May 5th, 2011 ...................................... 106

Figure 2.29: Conductivity of each sample by location on May 5th, 2011 ...................... 107

Figure 2.30: Calcium hardness and total hardness for each sample by location ............ 108

Figure 2.31: Zeta potential of each sample at process conditions by location ............... 109

Figure 2.32: The Effect of silica content on zeta potential of hematite ore. ................... 110

Figure 2.33: pH of each sample by location on July 13th, 2011 ..................................... 114

Figure 2.34: Conductivity of each sample by location on July 13th, 2011 ..................... 115

Figure 2.35: Calcium hardness and total hardness for each sample by location ............ 117

Figure 2.36: Zeta potential of each sample at process conditions by location ............... 118

Figure 2.37: pH of each sample by location on November 30th, 2011 ........................... 122

Figure 2.38: Conductivity of each sample by location on November 30th, 2011 ........... 123

Figure 2.39: Sodium and chloride concentrations of each sample by location .............. 124

Figure 2.40: Correlation between sodium and conductivity ........................................... 125

Figure 2.41: Magnesium, calcium and total hardness concentrations of each sample by

location ................................................................................................................ 126

Figure 2.42: Zeta potential of each sample by location on November 30th, 2011 .......... 128

Figure 2.43: Calcium and magnesium content of the solids of each sample by location 130

Figure 2.44: Iron and acid insolubles content of the solids of each sample by location 131

Figure 2.45: pH of each sample before and after caustic addition.................................. 136

xi
Figure 2.46: Conductivity of each sample before and after caustic addition.................. 137

Figure 2.47: Sodium concentration of each sample before and after caustic addition ... 138

Figure 2.48: Correlation between sodium and conductivity ........................................... 139

Figure 2.49: Potassium concentration of each sample before and after caustic addition 140

Figure 2.50: Calcium concentration of each sample before and after caustic addition .. 142

Figure 2.51: Magnesium concentration of each sample before and after caustic addition

............................................................................................................................. 143

Figure 2.52: Total Hardness of each sample before and after caustic addition .............. 144

Figure 2.53: Soluble iron concentration of each sample before and after caustic addition

............................................................................................................................. 146

Figure 2.54: Chloride concentration of each sample before and after caustic addition .. 147

Figure 2.55: Nitrate concentration of each sample before and after caustic addition..... 148

Figure 2.56: Total carbonate concentration of each sample before and after caustic

addition ............................................................................................................... 149

Figure 2.57: Sulfate concentration of each sample before and after caustic addition .... 150

Figure 2.58: Correlation between soluble iron and sulfate ............................................. 151

Figure 2.59: Total dissolved solids (TDS) of each sample before and after caustic

addition ............................................................................................................... 152

Figure 2.60: Zeta potential of each sample before and after caustic addition ................ 153

Figure 2.61: Calcium concentration in solids of each sample before and after caustic

addition ............................................................................................................... 154

xii
Figure 2.62: Magnesium concentration in solids of each sample before and after caustic

addition ............................................................................................................... 155

Figure 2.63: Iron concentration in solids of each sample before and after caustic addition

............................................................................................................................. 156

Figure 2.64: Acid insoluble concentration in solids of each sample before and after

caustic addition ................................................................................................... 157

Figure 2.65: Effects of Glass-H on Zeta Potential .......................................................... 159

Figure 2.66: Cyclone overflow starch effects on zeta potential...................................... 160

Figure 2.67: Starch effectiveness as a function of calcium hardness at cyclone overflow

............................................................................................................................. 161

Figure 2.68: Dependence of iron concentration on zeta potential .................................. 162

Figure 2.69: Correlation between zeta potential at the deslime feed and the iron recovery

within the deslime thickener. .............................................................................. 163

Figure 2.70: Correlations between calcium hardness and process performance. ........... 164

Figure 2.71: Correlation between deslime iron recovery and conductivity of deslime feed.

............................................................................................................................. 165

Figure 2.72: Flotation feed starch effects on zeta potential ............................................ 166

Figure 2.73: Starch effectiveness as a function of calcium hardness at flotation feed ... 167

Figure 2.74: Zeta potential throughout flotation cells during each sampling campaign 168

Figure 2.75: Effect of zeta potential of flotation feed on iron recovery. ........................ 169

Figure 2.76: Effect of calcium hardness in flotation on the iron recovery. .................... 170

Figure 2.77: Effect of conductivity in flotation on the iron recovery. ............................ 171

xiii
Figure 2.78: The effect of deslime weight rejection on iron loss during flotation. ........ 172

Figure 2.79: The effect of water hardness on deslime, flotation and overall recovery... 173

Figure 2.80: Carbonate equilibrium distribution changes during dewatering. ............... 174

Figure 2.81: Calcium equilibrium distribution changes during dewatering. .................. 175

Figure 2.82: Magnesium equilibrium distribution changes during dewatering. ............. 176

Figure 2.83: Zeta potential of sample during dewatering steps ...................................... 177

Figure 2.84: Grade versus recovery curve for the deslime thickener ............................. 185

Figure 2.85: Enrichment ratio versus recovery curve for the deslime thickener ............ 187

Figure 2.86: Enrichment ratio and grade versus recovery curve for the deslime thickener

............................................................................................................................. 188

Figure 2.87: Iron grade and recovery during selective flocculation and dispersion ....... 192

Figure 3.1: Surface complex formation at the metal-oxide/water interface. .................. 207

Figure 3.2: pH dependance of cation (a) and anion (b) adsorption ................................ 209

Figure 3.3: Illustrations of the 3 theories behind dispersant action ................................ 211

Figure 3.4: XRD patterns of iron ore used for dispersant study ..................................... 213

Figure 3.5: Sodium silicate effectiveness as a dispersant. .............................................. 216

Figure 3.6: Sodium tripolyphosphate effectiveness as a dispersant. .............................. 218

Figure 3.7: Sodium hexametaphosphate effectiveness as a dispersant. .......................... 220

Figure 3.8: Sodium henicosapolyphosphate effectiveness as a dispersant. .................... 222

Figure 3.9: Ethylenediaminetetraacetic acid effectiveness as a dispersant..................... 224

Figure 3.10: Scaled diagram of the continuous laboratory scale deslime thickener ....... 234

Figure 3.11: Iron grade/recovery using inorganic dispersants ........................................ 240

xiv
Figure 3.12: Iron grade/recovery using polyacrylic acid dispersants ............................. 241

Figure 3.13: Phosphorus rejection versus iron recovery for inorganic dispersants ........ 243

Figure 3.14: Phosphorus rejection versus iron recovery for polyacrylic acid dispersants

............................................................................................................................. 244

Figure A.1: Zeta potential as a function of pH for screen underflow ............................. 251

Figure A.2: Zeta potential as a function of pH for screen underflow with Glass-H ....... 252

Figure A.3: Zeta potential as a function of pH for cyclone overflow ............................. 253

Figure A.4: Zeta potential as a function of pH for cyclone overflow with starch .......... 254

Figure A.5: Zeta potential as a function of pH for deslime thickener overflow ............. 255

Figure A.6: Zeta potential as a function of pH for deslime thickener underflow ........... 256

Figure A.7: Zeta potential as a function of pH for flotation feed ................................... 257

Figure A.8: Zeta potential as a function of pH for scavenger returns............................. 258

Figure A.9: Zeta potential as a function of pH for scavenger tailings ............................ 259

Figure A.10: Zeta potential as a function of pH for rougher concentrate ....................... 260

Figure A.11: Zeta potential as a function of pH for concentrate thickener underflow ... 261

Figure A.12: Zeta potential as a function of pH for filter feed ....................................... 262

Figure B.1: Thermolysis curves for screen underflow.................................................... 263

Figure B.2: Thermolysis curves for screen underflow with Glass-H ............................. 264

Figure B.3: Thermolysis curves for cyclone overflow ................................................... 265

Figure B.4: Thermolysis curves for cyclone overflow with starch ................................. 266

Figure B.5: Thermolysis curves for deslime thickener overflow ................................... 267

Figure B.6: Thermolysis curves for deslime thickener underflow ................................. 268

xv
Figure B.7: Thermolysis curves for flotation feed .......................................................... 269

Figure B.8: Thermolysis curves for scavenger returns ................................................... 270

Figure B.9: Thermolysis curves for scavenger tailings .................................................. 271

Figure B.10: Thermolysis curves for rougher concentrate ............................................. 272

Figure B.11: Thermolysis curves for concentrate thickener underflow ......................... 273

Figure B.12: Thermolysis curves for filter feed.............................................................. 274

Figure C.1: Particle size distribution of screen underflow ............................................. 275

Figure C.2: Particle size distribution of screen underflow with Glass-H ....................... 276

Figure C.3: Particle size distribution of cyclone overflow ............................................. 277

Figure C.4: Particle size distribution of cyclone overflow with starch........................... 278

Figure C.5: Particle size distribution of deslime overflow ............................................. 279

Figure C.6: Particle size distribution of deslime underflow ........................................... 280

Figure C.7: Particle size distribution of flotation feed .................................................... 281

Figure C.8: Particle size distribution of scavenger returns ............................................. 282

Figure C.9: Particle size distribution of scavenger tailings ............................................ 283

Figure C.10: Particle size distribution of rougher concentrate ....................................... 284

Figure C.11: Particle size distribution of concentrate thickener underflow ................... 285

Figure C.12: Particle size distribution of Filter Feed ...................................................... 286

xvi
List of Tables

Table 1.1: Process reagents that contribute cations and anions to the process water. ...... 11

Table 1.2: Chemical analyses of process water ................................................................ 13

Table 1.3: Surface reactions of the constituents of iron ore in water ............................... 21

Table 2.1: Water quality parameters for each experiment ................................................ 57

Table 2.2: Sample location descriptions ........................................................................... 96

Table 2.3: Process data for May 5th, 2011. ..................................................................... 111

Table 2.4: Process Data on July 13th, 2011 ..................................................................... 120

Table 2.5: Process data for November 30th, 2011 ........................................................... 129

Table 2.6: Process data for sections 7-9 on April 2nd, 2012 ......................................... 135

Table 2.7: Operational parameters found to be important .............................................. 190

Table 2.8: Operational parameters found to be important .............................................. 195

Table 2.9: Operational parameters found to be important .............................................. 196

Table 3.1: Chemical names and structures of common dispersants ............................... 206

Table 3.2: Elemental analysis of iron ore used in this study .......................................... 213

Table 3.3: Water chemistry used during laboratory thickening tests.............................. 237

Table 3.4: Chemical structures of dispersants used in this study ................................... 238

xvii
Preface

This dissertation is a compilation of published journal articles, journals articles accepted

for publication, and unpublished project reports developed by Howard Haselhuhn and Dr.

S. Komar Kawatra. The layout of the journal articles have been edited to present a flow

of topics throughout the dissertation. Unpublished project reports were stripped of

intellectual property prior to their inclusion within this dissertation.

All experiments and measurements discussed within this dissertation were performed by

Howard Haselhuhn with the assistance of the undergraduate researchers acknowledged in

the acknowledgement section of this dissertation. The overall project oversight,

experimental design, technical considerations, and presentation of results in both paper

and visual form were prepared by both Howard Haselhuhn and Dr. S. Komar Kawatra.

The following peer reviewed publications have been included within this dissertation

with permission from the Journals they have been published in or are soon to be

published in:

Haselhuhn, H. J., and Kawatra, S. K., 2015, "The Role of Surface Chemistry in Iron Ore

Beneficiation," Minerals & Metallurgical Processing, Accepted.

Haselhuhn, H. J., 2012, "Water Chemistry Effects on the Zeta Potential of Concentrated

Hematite Ore," Minerals & Metallurgical Processing, Vol. 29, No. 2, pp. 135-136.

xviii
Haselhuhn, H. J, and Kawatra, S. K., 2015, "The Role of Water Chemistry in the

Selective Flocculation and Dispersion of Iron Ore," Minerals & Metallurgical

Processing, In Press

Haselhuhn, H. J., 2013, "Dispersant Adsorption and Effects on Settling Behavior of Iron

Ore," Minerals & Metallurgical Processing, Vol. 30, No. 3, pp. 188-189.

Haselhuhn, H. J., Kawatra, S. K., 2015, "Design of a Continuous Pilot Scale Deslime

Thickener," Minerals & Metallurgical Processing, Accepted.

xix
Acknowledgements

I would first like to thank Dr. S. Komar Kawatra for his extensive role in the research

projects that were accomplished for this dissertation. Without his advice, support and

most of all, patience, this work would not have been possible.

I would also like to thank Dr. Timothy Eisele for his time in reading and giving feedback

on all of my work. He has been a great friend throughout the past five years.

I would like to thank Cliffs Natural Resources, and particularly, Dr. Joshua Carlson, Scott

Rasmussen, and Mathew Rea for providing the materials, advice and partial financial

support for the research in this dissertation. I would also like to thank the U.S.

Department of Education for financial support and ASISC for laboratory use and

financial support. A special thanks goes out to all of the undergraduate students who put

in time on the research projects I have been working on. In particular, I would like to

thank Shannon Ennis for her time assisting with pilot thickener studies.

Finally, I would like to thank my wife, Amberlee Haselhuhn, for putting up with me over

the past five years.

xx
Abstract

Iron ore is one of the most important ores in the world. Over the past century, most

mining of iron ore has been focused on magnetite (Fe3O4). As the name suggests,

magnetite is magnetic in nature and is easily separated from gangue (unwanted) minerals

through magnetic separation processes. Unfortunately, the magnetite ore bodies are

diminishing. Because of this, there has been a recent drive to pursue technology that can

economically separate hematite (Fe2O3) from its gangue minerals as hematite is a much

more abundant source of iron. Most hematite ore has a very small liberation size that is

frequently less than 25μm. Beneficiation of any ore with this fine of a liberation size

requires advanced processing methods and is seldom pursued.

A single process, known as selective flocculation and dispersion, has been successfully

implemented at a plant scale for the beneficiation of fine liberation size hematite ore.

Very little is known about this process as it was discovered by the U.S. Bureau of Mines

by accident. The process is driven by water chemistry and surface chemistry

modifications that enhance the separation of the hematite from its gangue minerals. This

dissertation focuses on the role of water chemistry and process reagents in this hematite

beneficiation process. It has been shown that certain ions, including calcium and

magnesium, play a significant role in the process. These ions have a significant effect on

the surface chemistry as reported by zeta potential studies. It was shown that magnesium

ions within the process water have a more significant impact on surface chemistry than

calcium ions due to steric hindrance effects at the hematite surface. It has also been

xxi
shown that polyacrylic acid dispersants, if used in the process, can increase product

quality (increase iron content, decrease phosphorus content, decrease silica content)

substantially.

Water, surface and reagent chemistry experiments were performed at a laboratory, pilot,

and full plant scale during the course of this work. Many of the conclusions developed in

the laboratory and pilot scale were found to be true at the full plant scale as well. These

studies are the first published in history to develop theories of water chemistry and

surface chemistry interactions at a full plant scale.

xxii
Introduction

Iron ore is one of the most important ores to the global society. It primarily consists of

two minerals: magnetite (Fe3O4) and hematite (Fe2O3). Currently, magnetite is heavily

mined and concentrated using magnetic separators. Hematite is rarely mined because it

cannot be separated easily from its gangue minerals (silica, clay, etc…). Most of the

world’s iron ore is in the form of hematite; therefore, as the grade of magnetite ore

continues to decline, it is important to develop new and better technology to concentrate

hematite ore.

One process exists that can economically and industrially concentrate hematite ore. It

uses a combination of selective flocculation and dispersion desliming and reverse

cationic froth flotation. This process was developed by the U.S. Bureau of Mines and has

only been implemented at one plant in the world. It is a complicated process relying on

both surface chemistry and water chemistry to separate hematite from other minerals. The

surface chemistry, water chemistry, and reagent selection in this process is the topic of

this dissertation.

Surface chemistry play a very important role in the hematite beneficiation process.

Surface chemistry refers to the interactions of aqueous ions and reagents at the solid iron

ore particle surfaces. These surfaces typically carry an electrical charge that attracts ions

and reagents from the solution. The interaction between ions, reagents and the surface

can result in either physical or chemical bonds that can affect how the particle behaves in

the process. For example, a flocculant can be attracted to charged particle surfaces, form

xxiii
chemical or physical bonds with surfaces on multiple particles, and thereby cause

flocculation of those particles. This same phenomena can happen with aqueous ions. If

particle surfaces are negatively charged and a positively charged cation exists nearby, the

negatively charged surfaces will form either physical or chemical bonds with the cation,

and thereby cause flocculation.

During the first step of the hematite beneficiation process, known as selective

flocculation and dispersion desliming, reagent selection and water chemistry are very

crucial. At this step, all particles, regardless of mineralogy, carry a negatively charged

surface. It is important to remove positively charged ions such as calcium and magnesium

in order to prevent flocculation of particles containing different mineralogies. This is

accomplished using a dispersant and is discussed in depth in Chapter 3. It is also

important to add a selective flocculant that only flocculates a particular mineral, in this

case, hematite. If a dispersant is added to negate the effects of cations and a selective

flocculant is added to flocculate only hematite, hematite flocs will settle while all other

minerals will remain dispersed and can be removed.

Reagents selection, water chemistry and surface chemistry are also very important during

the second step of the hematite beneficiation process, known as reverse cationic froth

flotation. In this beneficiation process, a reagent is added to depress the surface charge of

only hematite. Then a reagent, known as a collector, is added that has a cationic end

group with a hydrophobic chain. Since the surface charge of the hematite has been

depressed, the cationic end of the collector selectively adsorbs to the non-hematite

particle surfaces. The hydrophobic end of the collector is easily attached to air bubbles
xxiv
which are injected to allow the now-hydrophobic particles to float to the top of the vessel,

while the flocculated and depressed hematite particles sink to the bottom and a removed

as the product.

This dissertation discusses the surface phenomena occurring within this beneficiation

process in detail. Experiments were performed to show the effects of water chemistry,

surface chemistry, and reagent selection in the laboratory, in a pilot scale beneficiation

process, and at a full scale hematite beneficiation plant. It is the first compilation of work

on this process that relates surface phenomena on all three process scales.

Chapter 1 Introduction

Chapter 1 of this dissertation is a review of the common iron ore beneficiation processes.

It serves as an in-depth introduction to the remainder of the dissertation. This chapter

explains how the water chemistry and surface chemistry are linked in the process and

discusses how changes in either can affect process performance.

Chapter 2 Introduction

Chapter 2 focuses on the water and surface chemistry required for an effective separation

of hematite from its gangue minerals in hematite ore. Chapter 2 is split into three

sections. The first section (2.1) details a laboratory analysis of the zeta potential of

hematite ore and shows how different common ions in water can affect the zeta potential.

Implications to process performance are also discussed. The second section (2.2) details

the effects of these water chemistry factors on process performance in a pilot scale

selective flocculation and dispersion deslime thickener. This thickener was constructed at

xxv
Michigan Tech and designed with the help of Chris Braun at FLSmidth. During this

study, water chemistry as changed in the process and the resulting grade and recovery of

hematite ore was measured. The third section (2.3) details a study of water chemistry and

surface chemistry at the previously mentioned selective flocculation and dispersion full

scale hematite beneficiation plant. In the first analysis, samples were taken throughout the

plant and relationships between water chemistry, surface chemistry and plant

performance were determined. In the second analysis, samples were taken and analyzed

on site three times per day for five days. Relationships between water chemistry, surface

chemistry, and process performance were used to give recommendations to the project

sponsor.

Chapter 3 Introduction

Chapter 3 contains material concerning dispersant selection for the selective flocculation

and dispersion process. Chapter 3 is split into two sections. The first section (3.1)

discusses dispersant effectiveness at a laboratory scale. During this study, the effects of

dispersant choice, dispersant dose, and water chemistry were determined. The

effectiveness of each dispersant was compared. The second section (3.2) uses many of

the dispersants from the first section (3.1) and applies them at a pilot scale to determine if

the laboratory conclusions scaled to a pilot separation process.

xxvi
1 The Role of Surface Chemistry in Iron Ore Beneficiation1

1.1 Iron Ore Beneficiation Processes

There are two principle types of ore bodies that are processed in the United States:

Magnetite and Hematite. Magnetite (Fe3O4) is a magnetic iron oxide that is easily

concentrated using a combination of magnetic separators and flotation. The reserves of

magnetite ore are continuously diminishing due to its ease of beneficiation making

technology that can process hematite ores extremely important. Hematite (Fe2O3) is a

non-magnetic iron oxide that constitutes the majority of the worldwide iron ore reserves.

Processing hematite ore typically requires advanced separation techniques including

selective flocculation and dispersion, reverse froth flotation, or high-tension electrostatic

separation.

Magnetite beneficiation typically takes place in a series of magnetic separators followed

by reverse froth flotation. A flow sheet for the magnetite beneficiation process at a

Minnesota plant is shown in Figure 1.1. During this process, a ~38% Fe magnetite ore is

ground in a rod mill to ~3% passing at 400 mesh. The first stage of magnetic separators,

known as cobbers, remove large siliceous particles from the process. The cobber

concentrate is then ground to ~51% passing at 400 mesh. This ground material is then

1
The material contained within this chapter has been published in the journal “Minerals
& Metallurgical Processing.” It is included in this dissertation with permission from the
journal.
Citation:
Haselhuhn, H. J., and Kawatra, S. K., 2015, "The Role of Surface Chemistry in Iron Ore
Beneficiation," Minerals & Metallurgical Processing, Accepted.
1
processed by a second stage of magnetic separators known as roughers. The roughers

concentrate the iron oxides to ~59% Fe. The concentrate is passed through a cyclone to

remove fine material (~61% passing at 400 mesh) for further processing. The course

material is recirculated back into the ball mill for further grinding. The cyclone overflow

is processed via thickening where fine silica is removed in the overflow. The concentrate

from thickener typically contains ~67% Fe. The next beneficiation step is a magnetic

separator known as the finisher. This step concentrates the iron up to ~68.5% Fe. (Reid,

1988) The final beneficiation step uses flotation to float the remaining silica from the

magnetite. Once the beneficiation process is complete, the concentrated iron ore is then

filtered and agglomerated into pellets in preparation for dry processing. Similar flow

sheets are used in other Minnesota and Michigan magnetite beneficiation plants (Mular et

al., 2002).

2
Figure 1.1: Example magnetite benefication process flow sheet used in Minnesota

mining operation. Adapted from Mular et al. (2002) and Reid (1988).

3
Hematite beneficiation processes (aside from electrostatic separation) rely much more

heavily on water chemistry and chemical reagents (flocculants and dispersants) to drive

the separation processes. This type of process is typically known as selective flocculation

and dispersion. This name stems from the use of corn starch as a selective flocculant for

iron oxides and the use of inorganic phosphate and silicate general dispersants. An

example of a flow sheet for this type of hematite beneficiation plant is shown in Figure

1.2. The incoming ore is ground with caustic soda (NaOH) in a ball mill (semi-

autogenous) and pebble mill until liberation size is reached (~81% passing at 500 mesh).

The fine material is removed by cyclones and the iron ore is initially concentrated to

~55% Fe in selective thickeners (deslime thickeners) using starch as a selective

flocculant. The deslime thickener underflow is further concentrated via reverse cationic

froth flotation using an etheramine collector. Carbon dioxide is added to the concentrate

to reduce the alkalinity prior to filtration. A calcite/dolomite fluxing agent is also added

at this stage for slag removal during downstream reduction processes. The concentrate is

then filtered and agglomerated into pellets in preparation for dry processing.(Keranen,

1986) This type of beneficiation plant is seldom economical because it is difficult to

control the surface chemistry that affect reagent selectivity (Department of Energy, 2001;

Haselhuhn et al., 2012).

4
Caustic Dispersant Starch Starch Collector CO2 & Flux Binder
Tailings
Cyclone

operation. Adapted from Keranen (1986).


Hematite

5
Ore To Dry
Disc Filter Balling
Flotation Process
Mill Drum
Deslime Circuit
Thickener Tailings
Mill

Figure 1.2: Example hematite beneficiation process flow sheet used in Michigan mining
The dry processing steps are very similar in most iron ore beneficiation plants. It consists

of drying and sintering the pelletized concentrate using either a straight grate or a grate-

kiln-grate furnace. Once dried, the pellets are shipped via railway or waterway to a

smelting plant. Smelting can take place in a variety of ways including the blast furnace,

electric arc furnace (EAF) or on a smaller scale, the rotary hearth furnace (RHF or

Midrex ITmk3®). A flow-sheet for dry processing of iron ore is shown in Figure 1.3.

6
Figure 1.3: Dry processing of concentrated iron ores through a grate-kiln-grate

sintering furnace and a blast furnace for pig iron production.

7
This dissertation focuses on the water chemistry, surface chemistry and reagents used

during the selective flocculation and dispersion wet process described in Figure 1.2.

1.2 Relationship of Water Chemistry, Surface Chemistry and the

Selective Flocculation and Dispersion Process

Ionic equilibrium and surface chemistry play vital roles during the selective flocculation

and dispersion process of hematite beneficiation (Eisele et al., 2005; Engesser, 2003;

Ripke et al., 2003). Ions in the process water are attracted to the solid particle surfaces

during the separation process. These interactions affect the electrostatic charge of the

particle in solution (zeta potential) and the ability of chemical reagents to act on the

surfaces in a way that promotes their separation. To study the effects of these ions in the

process, it is important to understand where these ions come from; i.e. what is dissolving

in the water to produce these ions and what ions are already present in the fresh and

recycled process waters. It is also important to understand how these dissolved ions

interact with the surfaces during processing; i.e. which ions can cause issues during

selective flocculation and dispersion.

1.2.1 Dissolution of Iron Ore Minerals

United States iron ores primarily consist of magnetite (Fe3O4). As the reserves of

magnetite continue to diminish, other ores including hematite (Fe2O3), goethite

(FeO(OH)), pyrite (FeS2), and siderite (FeCO3) will need to be mined to meet the

worldwide iron demand. These minerals behave differently during the concentration

process. Magnetite, hematite and goethite are, for the most part, insoluble iron oxides

8
during the concentration process; whereas, pyrite and siderite dissolve to form sulfate

ions and carbonate ions, respectively:

Pyrite Dissolution in Process Water (Engesser, 2003):

Ͷ‫ܵ݁ܨ‬ଶሺ௦ሻ ൅ ͳͷܱଶ ൅ ͳͶ‫ܪ‬ଶ ܱ ՜ Ͷ‫݁ܨ‬ሺܱ‫ܪ‬ሻଷ ൅  ͳ͸‫ ܪ‬ା ൅ ͺܱܵସିଶ 

Siderite Dissolution in Process Water (Bénézeth et al., 2009):

‫ܱܥ݁ܨ‬ଷǡሺ௦ሻ ՜ ‫ ݁ܨ‬ାଶ ൅ ‫ܱܥ‬ଷିଶ

The sulfuric acid generation from the oxidation reaction of pyrite is neutralized in the

process water by calcite (CaCO3) and dolomite (CaMg(CO3)2) present in the ore. This

neutralization reaction results in an increase in the ionic concentrations of calcium,

magnesium, carbonate and sulfate ions in the process water (Engesser, 2003). Gangue

minerals found in the ore, such as the aforementioned calcite and dolomite, also dissolve

to contribute calcium, magnesium and carbonate to the ionic content of the process water:

Calcite Dissolution in Process Water: Dolomite Dissolution in Process Water

‫ܱܥܽܥ‬ଷǡሺ௦ሻ ՜ ‫ܽܥ‬ାଶ ൅ ‫ܱܥ‬ଷିଶ ǡ ‫ݎ݋‬ ‫݃ܯܽܥ‬ሺ‫ܱܥ‬ଷ ሻଶǡሺ௦ሻ

՜ ‫ܽܥ‬ାଶ ൅ ‫݃ܯ‬ାଶ ൅ ʹ‫ܱܥ‬ଷି ǡ ‫ݎ݋‬

‫ ܪ‬ା ൅ ‫ܱܥܽܥ‬ଷǡሺ௦ሻ ՜ ‫ܽܥ‬ଶା ൅ ‫ܱܥܪ‬ଷି ʹ‫ ܪ‬ା ൅ ‫݃ܯܽܥ‬ሺ‫ܱܥ‬ଷ ሻଶǡሺ௦ሻ

՜ ‫ܽܥ‬ାଶ ൅ ‫݃ܯ‬ାଶ ൅ ʹ‫ܱܥܪ‬ଷି

1.2.2 Ionic Contributions from Chemical Reagents

Most other ions that are contained within the process water are the result of adding

reagents during the concentration process. Many reagents are added during the

9
concentration process that use sodium to solubilize their functional groups including

sodium hydroxide (pH modifier and dispersant), sodium carbonate (pH modifier), sodium

oleate (anionic collector), sodium silicate (dispersant), sodium phosphates (dispersant),

and sodium dodecyl sulphate (anionic collector) (Arol et al., 2003; Haselhuhn et al.,

2012; Iwasaki, 1983; Quast, 2000). There are also reagents that contribute anions to the

process water including sodium silicate, sodium phosphate, sodium carbonate, carbon

dioxide (pH modifier) and dodecylammonium chloride (cationic collector). These

reagents, their functional groups and some typical dosage rates found in literature can be

seen in Table 1.1.

10
Table 1.1: Process reagents that contribute cations and anions to the process water.

Typical

Trade Chemical Dosage


Chemical Name Structure Reagent Use
Name Formula (Units

Specified)

0.30 lb/tona
Water
Sodium Silicate Na2SiO3 Dispersant 0–4
Glass
lb/tonb

Sodium Dispersant/Water 0–1


STPP Na5P3O10
Tripolyphosphate Softener lb/tonb

pH

Sodium Hydroxide Caustic NaOH Modifier/Dispers 2.45 lb/tona

ant

Soda
Sodium Carbonate NaCO3 pH Modifier
Ash

Carbon Dioxide - CO2 pH Modifier

Anionic
Sodium Oleate - NaC18H33O2
Collector

Sodium Dodecyl NaC12H25S Anionic 0.2 – 2.2


-
Sulphate O4 Collector lb/tonc

Dodecyl-
C12H25NH3 Cationic
ammonium -
Cl Collector
Chlorided

References:

a - (Keranen, 1986) b - (Green et al., 1984) c - (Quast, 2000) d - (Choi et al., 1961)

11
1.2.3 Ionic Contributions from Fresh and Re-use water in Concentrator

Depending on the iron ore processing facility, the water used in the process can come

from recirculating process water, fresh water or a combination of both. A recent study

conducted by Haselhuhn et al., (2012) on process water chemistry throughout an iron ore

concentrator showed the principle ions present in the process water at different location

within a hematite ore concentrator. This study concluded that the process water changes

drastically from the initial addition of water during comminution to the removal of water

at filtration. This data is shown in Table 1.2. As shown, the incoming water has

significant amounts of sodium, carbonate and chloride ions present. The sodium

concentration slightly increases from 455 to 495 ppm due to reagent addition. The total

carbonate concentration (bicarbonate plus carbonate) increases from 818 ppm to 1295

ppm due to the addition of a calcite/dolomite fluxing agent and the addition of carbon

dioxide prior to filtration. The addition of the fluxing agent and CO2 also cause an

increase in the calcium and magnesium concentration of the process water because of the

dissolution of the fluxing agent at the decreased pH. The high concentrations of ions seen

at the filter feed can be detrimental to downstream balling, sintering and reduction

processes. (Haselhuhn et al., 2012)

12
Table 1.2: Chemical analyses of process water at an operating hematite ore

concentration plant at water addition and removal operations (Haselhuhn et al., 2012).

Description Post Comminution Process Water Filter Feed Process Water

Calcium (mg/L) 4.47 71

Magnesium (mg/L) 2.67 58

Sodium (mg/L) 455 495

Potassium (mg/L) 11.4 20.5

Iron (mg/L) 16.4 0.5

pH (WAL) 10.59 7.67

Chloride (mg/L) 110 100

Sulfate (mg/L) 90 100

Fluoride (mg/L) 4.1 4.2

Nitrate (mg/L) 16 0.1

Nitrite (mg/L) 0.29 0

Carbonate (mg/L) 724 0

Bicarbonate (mg/L) 94 1295

Hardness (CaCO3 mg/L) 22 416

1.3 Surface Chemistry of Iron Ore

Interactions between process water ions and the surfaces of the slurry particles account

for most variations in processing conditions during flotation and desliming operations in

the selective flocculation and dispersion process. Understanding how specific ions

interact with surfaces is essential to optimize process performance and predict

downstream processing requirements.


13
The pH of the water is the primary factor determining the surface charge of particles in

solution. (Laskowski et al., 1992; Stumm et al., 1992). At a low pH the surface is

positively charged, whereas, at a high pH, the surface is negatively charged as shown in

Figure 1.4. The pH at which the surface bears no charge is referred to as the point of zero

charge (PZC).

14
Surface Inner- Outer-
Shear
Hydroxyl sphere sphere
Plane
Groups Complexes Complexes

H H Surface
O O Charge
pH=11

H
Fe O Ca+

O
H
O
O pH=2
O H H

H
H
H+
H H
Fe OH Cl-

O
H
H
Charge

O H H Bulk Fluid

0
pH§8 O
Fe O H
H H
O O
pH=11
Zeta
H

H
Fe O- Na+
O
O

Potential
H

H
O O
H H

Distance from Surface

Figure 1.4: Water-solid interface of a hydrated iron oxide. As distance from surface

hydroxyl groups increases, the charge with respect to the bulk fluid decreases. The

charge at the shear plane is known as the zeta potential. pH: hydrogen from solution

adsorbs to surface oxygen; depending on the pH of the solution, the hydroxyl surface

can be positively, negatively, or neutrally charged. Complexes: inner sphere complexes

exist at the hydroxyl surface if pH is adequate for formation; outer-sphere complexes

exist due to coulombic forces away from surface. Adapted from Dzombak et al. (1990),

Laskowski et al. (1992), Sparks et al. (1998), and Stumm et al. (1992).

15
The surface charge determines the charge of ions in solution that will be attracted to the

surface. The ions that are attracted to the surface either form inner-sphere or outer-sphere

complexes at the surface. An inner-sphere complex consists of a primarily single covalent

bond between the ion and the surface. An outer sphere complex occurs when the ion

approaches the surface through coulombic forces but is separated from the surface by one

or more water molecules, see Figure 1.4 (Stumm et al., 1992).

The coulombic forces diminish with distance from the surface. As the coulombic forces

diminish, the outer-sphere complexes can no longer be held to the surface and are merely

present as diffuse ions in the bulk fluid. The distance at which the coulombic forces are

too weak to hold outer-sphere complexes to the surface is referred to as the shear plane,

see Figure 1.4. The potential difference between the shear plane (sometimes referred to as

the slipping plane) and the bulk solution is referred to as the zeta potential. The zeta

potential of particle surfaces is primarily a function of water chemistry and the

underlying surface charge of the particle.

1.3.1 Surface Speciation and Surface Potential

Each oxide mineral in the iron ore has a unique relationship with the surrounding aqueous

environment. This relationship creates an electrical charge at the surface of the particle.

The magnitude and sign (positive or negative) of the surface charge is first and foremost

a function of pH. For an oxide mineral, the 2 pKa model has been extensively used to

define the surface charge of particles. It consists of chemical equations relating the

concentration of negatively, neutrally, and positively charged surface sites to an

16
equilibrium constant. This model consists of the following equilibrium reactions and

equations (Equations 1.1 and 1.2):

ೌ೛೛
௄ೌభ ௔௣௣ ሾ‫ܪܱܯ ؠ‬ሿሼ‫ ܪ‬ା ሽ
‫ܪܱܯ ؠ‬ଶା ሯልሰ ‫ ܪܱܯ ؠ‬൅ ‫ ܪ‬ା ‫ܭ‬௔ଵ ൌ Equation 1.1
ሾ‫ܪܱܯ ؠ‬ଶା ሿ

ೌ೛೛
௄ೌమ ௔௣௣ ሾ‫ ିܱܯ ؠ‬ሿሼ‫ ܪ‬ା ሽ
‫ ܪܱܯ ؠ‬ሯልሰ ‫ ܱܯ ؠ‬൅ ‫ܪ‬ ି ା ‫ܭ‬௔ଶ ൌ Equation 1.2
ሾ‫ܪܱܯ ؠ‬ሿ

ܹ݄݁‫݁ݎ‬ǡ

‫݂݁ܿܽݎݑݏ݄ܾ݈ܽ݀݊݅݁ܽ݅ݎ݁ݐ݈ܽ݉݀݅݋ݏܽݏ݁ݐ݋݊݁݀ ؠ‬

‫݈ܽݎ݁݊݅݉݁݀݅ݔ݋݊ܽ݉ݎ݋݂݋ݐ݁݀݅ݔ݋݋ݐ݀݁݀݊݋ܾ݉݋ݐܽ݊ܽݏ݁ݐ݋݊݁݀ܯ‬

ሺ݁‫ݔ‬ǣ‫ ܯ‬ൌ ‫݁ܨݎ݋݂݁ܨ‬ଶ ܱଷ ሻ


௔௣௣
‫ܭ‬௔௫ ൌ ‫݊݋݅ݐܿܽ݁ݎ݊݁ݒ݄݅݃݁ݐݎ݋݂ݐ݊ܽݐݏ݊݋ܿ݉ݑ݅ݎܾ݈݅݅ݑݍ݈݁݀݁݅݌݌ܣ‬

݉‫݈݋‬
ሾ‫ܪܱܯ ؠ‬ଶା ሿ ൌ ‫ ݏ݁ݐ݅ݏ݂݁ܿܽݎݑݏ݀݁݃ݎ݄ܽܿݕ݈݁ݒ݅ݐ݅ݏ݋݌݂݋݊݋݅ݐܽݎݐ݊݁ܿ݊݋ܥ‬൬ ൰
‫ܮ‬

݉‫݈݋‬
ሾ‫ܪܱܯ ؠ‬ሿ ൌ ‫ ݏ݁ݐ݅ݏ݂݁ܿܽݎݑݏ݀݁݃ݎ݄ܽܿݕ݈݈ܽݎݐݑ݂݁݊݋݊݋݅ݐܽݎݐ݊݁ܿ݊݋ܥ‬൬ ൰
‫ܮ‬

݉‫݈݋‬
ሾ‫ ିܱܯ ؠ‬ሿ ൌ ‫ ݏ݁ݐ݅ݏ݂݁ܿܽݎݑݏ݀݁݃ݎ݄ܽܿݕ݈݁ݒ݅ݐ݂ܽ݃݁݊݋݊݋݅ݐܽݎݐ݊݁ܿ݊݋ܥ‬൬ ൰
‫ܮ‬

݉‫݈݋‬
ሼ‫ ܪ‬ା ሽ ൌ ‫ ݎ݁ݐܽݓݏݏ݁ܿ݋ݎ݌݄݁ݐ݊݅݊݁݃݋ݎ݀ݕ݄݂݋ݕݐ݅ݒ݅ݐܿܣ‬൬ ൰
‫ܮ‬

௔௣௣
The applied equilibrium constant, ‫ܭ‬௔௫ , is a function of the concentration of the reactive

surface sites, the activity of the ions present in the process water, and the electrical

potential between the surface and the bulk fluid surrounding the surface, denoted surface

potential (߰). It can be correlated to published values of intrinsic equilibrium constants

using the following empirical relation (Equation 1.3):


17
௓ήிήట
௔௣௣ ௜௡௧ Equation 1.3
‫ܭ‬௔௫ ൌ ‫ܭ‬௔௫ ή݁ ோή்

ܹ݄݁‫݁ݎ‬ǡ

௜௡௧
‫ܭ‬௔௫ ൌ ‫݊݋݅ݐܿܽ݁ݎ݊݁ݒ݄݅݃݁ݐݎ݋݂ݐ݊ܽݐݏ݊݋ܿ݉ݑ݅ݎܾ݈݅݅ݑݍ݁ܿ݅ݏ݊݅ݎݐ݊ܫ‬

ܼ ൌ ‫݊݋݅ݐܿܽ݁ݎ݊݁ݒ݄݅݃݁ݐݎ݋݂݁݃ݎ݄݂ܽܿ݁ܿܽݎݑݏ݄݊݅݁݃݊ܽܥ‬

‫ܥ‬
‫ ܨ‬ൌ ‫ ݐ݊ܽݐݏ݊݋ܥݕܽ݀ܽݎݎܽܨ‬൬ͻ͸ͶͺͷǤ͵Ͷ ൰
݉‫݈݋‬

߰ ൌ ܵ‫݈ܽ݅ݐ݊݁ݐ݋݂ܲ݁ܿܽݎݑ‬ሺܸሻ

݇݃ ή ݉ଶ
ܴ ൌ ‫ ݐ݊ܽݐݏ݊݋ܥݏܽܩ݈ܽ݁݀ܫ‬ቆͺǤ͵ͳͶ ଶ ቇ
‫ ݏ‬ή ݉‫ ݈݋‬ή ‫ܭ‬

ܶ ൌ ܶ݁݉‫݁ݎݑݐܽݎ݁݌‬ሺ‫ܭ‬ሻ

To use these equations, both the intrinsic equilibrium constant and the surface potential at

a given pH must be known. Literature values for the intrinsic equilibrium constant for

common constituents of iron ore are shown in Table 1.3. The surface potential, can be

calculated via the Gouy-Chapman theory (Equation 1.4) using the total ionic

concentration of the slurry, c, the temperature of the slurry, T, the dielectric constant of

the slurry, İ, and the valency of the ions in the slurry, Z and the charge of the surface, ı

(Stumm et al., 1992):

ξʹߪ
ʹܴܶ ή ܽ‫ ݄݊݅ݏ‬ቆ ቇ
Ͷඥܴܶߝߝ଴ ܿ Equation 1.4
߰ൌ
‫ܨ‬ή‫ݖ‬

ܹ݄݁‫݁ݎ‬ǡ

݇݃ ή ݉ଶ
ܴ ൌ ‫ ݐ݊ܽݐݏ݊݋ܥݏܽܩ݈ܽ݁݀ܫ‬ቆͺǤ͵ͳͶ ቇ
‫ ݏ‬ଶ ή ݉‫ ݈݋‬ή ‫ܭ‬
18
ܶ ൌ ܶ݁݉‫݁ݎݑݐܽݎ݁݌‬ሺ‫ܭ‬ሻ

ߝ ൌ ‫ݐ݊ܽݐݏ݊݋ܥܿ݅ݎݐ݈ܿ݁݁݅ܦ‬ሺ͹ͺǤͷ݂‫ʹݐܽݎ݁ݐܽݓݎ݋‬ͷι‫ܥ‬ሻ

‫ܥ‬
ߝ଴ ൌ ܲ݁‫ ݁ܿܽ݌ܵ݁݁ݎܨ݂݋ݕݐ݅ݒ݅ݐݐ݅݉ݎ‬൬ͺǤͺͷͶ ή ͳͲିଵଶ ൰
ܸή‫ܯ‬

݉‫݈݋‬
ܿ ൌ ܶ‫ ݕݎݎݑ݈݂ܵ݋݊݋݅ݐܽݎݐ݊݁ܿ݊݋ܥܿ݅݊݋ܫ݈ܽݐ݋‬൬ ൰
݉ଷ

‫ ݖ‬ൌ ܸ݈ܽ݁݊ܿ‫ݕݎݎݑ݈ݏ݄݁ݐ݊݅ݏ݊݋݂݅݋ݕ‬

‫ܥ‬
ߪ ൌ ܵ‫ ݁݃ݎ݄ܽܥ݂݁ܿܽݎݑ‬൬ ൰
݉ଶ

The unknown variable in Equation 1.4 is the surface charge, ı. This electric charge

attracts counter-ions in the solution forming a layer of charged ions surrounding the

particle. This layer of ions can reduce or strengthen the effects of process reagents added

to the slurry and thus have a large impact on mineral separation operations. The surface

charge must be solved for iteratively using the following equation (Equation 1.5) in

conjunction with Equation 1.1-1.4 and a mole balance on surface sites:

‫ܨ‬
ߪൌ ሺሾ‫ܪܱܯ ؠ‬ଶା ሿ െ ሾ‫ ିܱܯ ؠ‬ሿሻ Equation 1.5
ܽή‫ݏ‬

ܹ݄݁‫݁ݎ‬ǡ

݇݃
ܽ ൌ ‫ ݏ݈݀݅݋݂ܵ݋݊݋݅ݐܽݎݐ݊݁ܿ݊݋ܥ‬൬ ൰
‫ܮ‬

݉ଶ
‫ ݏ‬ൌ ܵ‫ ܽ݁ݎܣ݂݁ܿܽݎݑ݂ܵܿ݅݅ܿ݁݌‬ቆ ቇ
݇݃

19
Specific surface area can be measured for each case; however, in absence of technology

to measure surface area, literature values are available for ranges of surface areas of each

of the main constituents in iron ore (Cornell et al., 1996; Demchuk et al., 2005).

20
Table 1.3: Surface reactions of the constituents of iron ore in water and their associated

equilibrium constants (Davis et al., 1986; Fuerstenau et al., 2007).

Equilibrium Constants for Iron Ore


Surface Acidity Reaction & Equation
Minerals

ೌ೛೛
௄ೌభ ࢒࢕ࢍ൫ࡷ࢏࢔࢚ ࢏࢔࢚
ࢇ૚ ൯ ࢒࢕ࢍ൫ࡷࢇ૛ ൯ Mineral
‫ܪܱ݁ܯ ؠ‬ଶା ሯልሰ ‫ ܪܱ݁ܯ ؠ‬൅ ‫ ܪ‬ା

-7.09 -11.11 Goethite (FeO(OH))

௓ήிήట
௔௣௣ ௜௡௧ -5.8 -11.10 Hematite (Fe2O3)
‫ܭ‬௔ଵ ൌ ‫ܭ‬௔ଵ ή݁ ோή்

ೌ೛೛
௄ೌమ -4.30 -8.30 Magnetite (Fe3O4)
‫ ܪܱ݁ܯ ؠ‬ሯልሰ ‫ ିܱ݁ܯ ؠ‬൅ ‫ ܪ‬ା

1.30 -5.70 Quartz (SiO2)


௓ήிήట
௔௣௣ ௜௡௧
‫ܭ‬௔ଶ ൌ ‫ܭ‬௔ଶ ή ݁ ோή்

pH is the most important factor in the surface chemistry of slurried iron ore. pH defines

the surface acidity of the mineral constituents in the slurry. The surface site distributions

for these oxide minerals show that the surfaces carry either a positive, negative or neutral

charge depending on the pH of the aqueous medium. When the number of positively

charged surface sites and the number of negatively charged surface sites are equal, the

corresponding pH value is known as the point of zero charge (PZC). When the pH is

above the PZC, the surface has a net negative surface charge. When the pH is below the

PZC, the surface has a net positive surface charge.

21
‫ܪܱܯ ؠ‬ଶା ‫ܪܱܯ ؠ‬ ‫ିܱܯ ؠ‬

Figure 1.5: Distribution of positively, neutrally and negatively charged surface sites as

a function of pH for Goethite, Hematite, Magnetite and Silica. Note: M denotes an atom

in the crystal lattice such as iron or silicon.


22
From Figure 1.5 it can be seen that each different mineral surface has a different

distribution of positively, negatively and neutrally charged surface sites at different pHs.

Inherently, this causes each surface to carry a different surface potential, ȥ, at each pH.

Figure 1.6 shows the surface potential of each mineral as a function of pH. The surface

potential of the slurry particles attract or repel ions in the process water and is responsible

for effective reagent adsorption that drives separation processes such as flotation. The

information presented in Figure 1.6 helps explain why certain collectors and surfactants

have a narrow effective pH range and why cations and anions are attracted to and bond to

surfaces in specific pH ranges.

23
Figure 1.6: Surface potentials as a function of pH for Silica, Magnetite, Hematite and

Goethite. Slurry particle characteristics measured from filter feed at hematite ore

concentration plant. Site density for mineral surface sites found in literature. (Cornell et

al., 1996; Demchuk et al., 2005)

1.3.2 Ion Adsorption Phenomena at Mineral Oxide Surfaces

Like the hydrogen ion, other ions can form physical or chemical bonds with a charged

surface and cause changes in the surface potential. Ions that form chemical bonds with

the surface, or specifically adsorbing ions, create inner-sphere complexes with the

surface. Ions that form physical bonds with the surface via electrostatic interaction create

outer-sphere complexes with the surface. Both inner-sphere and outer-sphere complexes

have an effect on the selective flocculation and dispersion process.

24
Inner-sphere complexation can be modeled similarly to the equations discussed above

except that equilibrium expressions for ionic interactions with each other must be

considered as well. Much data has been compiled on ionic interactions and surface

adsorption to iron oxides and silicon oxides in literature (Berka et al., 2001; Dzombak et

al., 1990; Kinniburgh et al., 1982; Kinniburgh et al., 1976; Kitamura et al., 1999; Smith

et al., 2004). Robust software, such as PHREEQCI, was used to iteratively model ion

adsorption and inner-sphere complexation at mineral surfaces while taking ionic

equilibrium into account. To use this software, a comprehensive database was created

including all surface speciation reactions to be modeled. Some intrinsic adsorption

constants for ion adsorption had to be estimated based on those of similar minerals. Using

the process water constituents and slurry solid characteristics at an operating hematite

concentrator, the speciation of inner-sphere complexes at slurry particle surface sites as a

function of pH was developed. This speciation map is shown in Figure 1.7.

25
ܶ‫݁ܿܽݎ‬
‫ܽܥܱܯ ؠ‬ା 
‫݃ܯܱܯ ؠ‬ା 

‫ܪܯ ؠ‬௫ ‫ܱܥ‬ଷ 

‫ܪܯ ؠ‬௫ ܱܲସ 
‫ ିܱܯ ؠ‬
‫ܱܯ ؠ‬଴ 
‫ܪܱܯ ؠ‬ଶା 

Figure 1.7: Inner-sphere complex speciation at particle surfaces as a function of pH. At

a given pH, the percent of a surface covered by a particular surface complex is shaded

accordingly. For example, at a pH of 7, most of the surface is covered in carbonate

surface complexes. Note: x denotes number of atoms in complex, y denotes charge of



complex, trace includes ‫ࡲࡹ ؠ‬૙ and ‫ࡻࡿ ࢞ࡴࡹ ؠ‬૝ .

Figure 1.7 shows, from a theoretical basis, what inner-sphere complexes dominate the

surfaces of the iron ore particles. Inner-sphere complexes within Figure 1.7 were grouped

according to the ion type; for instance,

ሾŁMHx POy4 ሿ=ൣŁMH2 PO04 ൧+ሾŁMHPO-4 ሿ+ൣŁMPO-2


4 ൧. These inner-sphere complexes can be

displaced by or hinder adsorption of process reagents depending on the thermodynamics


26
and chemical equilibrium of the reagent adsorption. These inner-sphere complexes also

travel with the surfaces during the process and contribute to the ionic constituents within

the final product.

Negatively and neutrally charged phosphate inner-sphere complexes dominate the surface

of iron ore particles below a pH of 5, as shown in Figure 1.7. Above pH 5, the surface is

dominated by negatively and neutrally charged carbonate inner-sphere complexes until a

pH of 8. At a pH of 8 and above, the surface is dominated by positively charged

magnesium inner-sphere complexes and un-complexed hydroxyl sites with a negative

charge.

Outer-sphere complexes form surrounding the particle near the particle surface and are

held to the surface by electrostatic forces. Outer-sphere complexes surrounding particle

surfaces remain with the particles during the process due to these electrostatic forces. The

total molar quantity of each ion in the diffuse layer can be modeled using PHREEQCI as

well. A similar mapping of diffuse layer content as a function of pH is shown in Figure

1.8. These calculations were made assuming the layer of outer-sphere complexes spans to

one Debye length beyond the surface using calculations found in Dzombak and Morel

(1990).

In acidic conditions, sulfate, chloride and carbonate outer-sphere complexes dominate the

surface of hematite ore particles, as shown in Figure 1.8. As the pH approaches neutral,

the carbonate outer-sphere complexes diminish allowing for sodium outer-sphere

27
complexes to dominate. In alkaline conditions, the surface is dominated by sodium and

carbonate outer-sphere complexes.

ܶ‫݁ܿܽݎ‬

‫ܪ̱ܱܯ ؠ‬௫ ܱܵସ
‫̱ܱܽܰܯ ؠ‬ା 
‫ ି ݈ܥ̱ܱܯ ؠ‬

‫ܪ̱ܱܯ ؠ‬௫ ‫ܱܥ‬ଷ

Figure 1.8: Outer-sphere complex speciation at particle surfaces as a function of pH. At

a given pH, the percent of a surface covered by a particular outer-sphere complex is

shaded accordingly. For example, at a pH of 7, most of the surface is covered in sodium

outer-sphere complexes. Note: x denotes number of atoms in complex, y denotes charge

of complex, trace includes ‫ࢇ࡯̱ࡻࡹ ؠ‬ା૛ , ‫ ିࡲ̱ࡻࡹ ؠ‬, ‫ ࢟ࢋࡲ̱ࡻࡹ ؠ‬, ‫ࡷ̱ࡻࡹ ؠ‬ା , ‫ؠ‬
࢟ ࢟
ࡹࡻ̱ࡹࢍା૛ , ‫ ࢞ࡻࡺ̱ࡻࡹ ؠ‬, and ‫ࡻࡼ ࢞ࡴ̱ࡻࡹ ؠ‬૝ .

28
1.4 References

Arol, A. I., and Iwasaki, I., 2003, "Effect of sodium silicate on flocculation of hematite

with starch in the presence of calcium," Separation Science and Technology, Vol. 38, No.

3, pp. 647-659.

Bénézeth, P., Dandurand, J. L., and Harrichoury, J. C., 2009, "Solubility product of

siderite (FeCO3) as a function of temperature (25–250 °C)," Chemical Geology, Vol. 265,

No. 1–2, pp. 3-12.

Berka, M., and Bányai, I., 2001, "Surface Complexation Modeling of K+, NO3í, SO42í,

Ca2+, Fí, Co2+, and Cr3+ Ion Adsorption on Silica Gel," Journal of Colloid and Interface

Science, Vol. 233, No. 1, pp. 131-135.

Choi, H., Cooke, S., and Iwasaki, I., 1961, "Flotation Characteristics of Hematite,

Goethite and Activated Quartz with 18-Carbon Aliphatic Acids and Related Compunds,"

AIME Transactions, Vol. 217, pp. 237-244.

Cornell, R. M., and Schwertmann, U., 1996, The Iron Oxides, 2nd Edition, John Wiley &

Sons, 703 pp.

Davis, J. A., and Hayes, K. F., 1986, Geochemical Processes at Mineral Surfaces,

American Chemical Society, 683 pp.

Demchuk, E. and Murashov, V. V., 2005, "Surface Sites and Unrelaxed Surface Energies

of Tetrahedral Silica Polymorphs and Silicate," Surface Science, Vol. 595, pp. 6-19.

29
Department of Energy, 2001, “Selective Flocculation of Fine Mineral Particles,” Office

of Industrial Technologies Energy Efficiency and Renewable Energy, U.S. Department of

Energy.

Dzombak, D. A., and Morel, F., 1990, Surface complexation modeling : hydrous ferric

oxide, Wiley, 393 pp.

Eisele, T. C., Kawatra, S. K., and Ripke, S. J., 2005, "Water Chemistry Effects In Iron

Ore Concentrate Agglomeration Feed," Mineral Processing and Extractive Metallurgy

Review, Vol. 26, No. 3, pp. 295-305.

Engesser, J., 2003, "Effect of water chemistry, water treatment and Blaine on magnetite

filtering and magnetite agglomeration with bentonite clay," Minerals & Metallurgical

Processing, Vol. 20, No. 3, pp. 125-134.

Fuerstenau, M. C., Jameson, G., and Yoon, R.H., 2007, Froth Flotation: A Century of

Innovation, Society for Mining, Metallurgy, and Exploration, 891 pp.

Green, R. E., and Colombo, A. F., 1984, “Dispersion-Selective Flocculation-Desliming

Characteristics of Oxidized Taconites,” US Bureau of Mines, Report of Investigations, RI

8867, 24 pp.

Haselhuhn, H. J., Carlson, J. J., and Kawatra, S. K., 2012, "Water chemistry analysis of

an industrial selective flocculation dispersion hematite ore concentrator plant,"

International Journal of Mineral Processing, Vol. 102–103, pp. 99-106.

30
Iwasaki, I., 1983, "Iron Ore Flotation, Theory and Practice," Mining Engineering, Vol.

35, pp. 622-632.

Keranen, C. U., 1986, “Reagent Preparation, Distribution and Feeding Systems at the

Tilden Mine,” Design and Installation of Concentration and Dewatering Circuits, A. L.

Mular and M. A. Anderson eds., Society for Mining, Metallurgy and Exploration, pp.

308-319.

Kinniburgh, D. G., and Jackson, M. L., 1982, "Concentration and pH dependence of

calcium and zinc absorption by iron hydrous oxide gel," Soil Science Society of America

Journal, Vol. 46, No. 1, pp. 56-61.

Kinniburgh, D. G., Jackson, M. L., and Syers, J. K., 1976, "Adsorption of Alkaline Earth,

Transition, and Heavy Metal Cations by Hydrous Oxide Gels of Iron and Aluminum,"

Soil Science Society of America Journal, Vol. 40, No. 5, pp. 796-799.

Kitamura, A., Fujiwara, K., Yamamoto, T., Nishikawa, S., and Moriyama, H., 1999,

"Analysis of Adsorption Behavior of Cations onto Quartz Surface by Electrical Double-

layer Model," Journal of Nuclear Science and Technology, Vol. 36, No. 12, pp. 1167-

1175.

Laskowski, J. S., and Ralston, J., 1992, Colloid Chemistry in Mineral Processing,

Elsevier Science, 428 pp.

Mular, A. L., Halbe, D. N., and Barratt, D. J., 2002, Mineral Processing Plant Design,

Practice, and Control, Society for Mining, Metallurgy, and Exploration, 2500 pp.

31
Quast, K. B., 2000, "A review of hematite flotation using 12-carbon chain collectors,"

Minerals Engineering, Vol. 13, No. 13, pp. 1361-1376.

Reid, K. L., 1988, “Application of QEM*SEM for Beneficiation Studies of Minnesota

Taconite,” Process mineralogy VIII, D. J. T. Carson Ed., TMS, 391 pp.

Ripke, S. J., and Kawatra, S. K., 2003, "Effect of Cations on Unfired Magnetite Pellet

Strength," Minerals & Metallurgical Processing Journal, Vol. 20, No. 3, pp. 153-159.

Smith, R. M., Martell, A. E., and Motekaitis, R. J., 2004, “NIST critically selected

stability constants of metal complexes database,” Version 8.0, National Institute of

Standards and Technology, U.S. Department of Commerce.

Sparks, D. L., and Grundl, T. J., 1998, Mineral-Water Interfacial Reactions, American

Chemical Society, 438 pp.

Stumm, W., Sigg, L., and Sulzberger, B., 1992, Chemistry of the solid-water interface,

Wiley, 448 pp.

32
2 Effects of Water Chemistry on Dispersion and Selective

Flocculation of Hematite Ore

2.1 Water Chemistry Effects on Zeta Potential of Concentrated

Hematite Ore2

2.1.1 Abstract

The effectiveness of selective flocculation and dispersion processes for the concentration

of hematite ore are strongly dependent on the ionic content of the process water. It has

been noticed that magnesium ions are more detrimental to iron recoveries during

selective flocculation and dispersion and flotation processes than calcium. This

phenomenon was studied by measuring the zeta potential of hematite ore at various

concentrations of sodium, magnesium, calcium, strontium and barium at a pH of 11.

Results show that zeta potential inverts from negative to positive at very low

concentrations of magnesium ions in the solution. It takes a significantly higher

concentration of calcium, ions to achieve the same effect. This difference may be

attributed to the ability of magnesium ions to adsorb to all surface hydroxyl groups,

2
The material contained in this chapter has been published in the journal “Minerals and
Metallurgical Processing.” It is included in this dissertation with permission from the
journal. This material can be found in the following two publications:

Haselhuhn, H. J., 2012, "Water Chemistry Effects on the Zeta Potential of Concentrated
Hematite Ore," Minerals & Metallurgical Processing, Vol. 29, No. 2, pp. 135-136.

Haselhuhn, H. J., and Kawatra, S. K., 2015, "The Role of Surface Chemistry in Iron Ore
Beneficiation," Minerals & Metallurgical Processing, Accepted.
33
whereas calcium, due to their larger size, can only adsorb to every other hydroxyl group.

This hypothesis was confirmed by results similar to those seen with calcium when this

test was repeated with strontium and barium ions. The tendency for calcium ions to

adsorb to every other surface hydroxyl group causes higher concentrations of calcium to

be less detrimental to hematite concentration processes than magnesium.

2.1.2 Introduction

High grade iron ore has been nearly depleted in the United States forcing the mining

industry to turn to beneficiation processes to concentrate lower grade ores. Some of these

processes exploit the magnetic and electrostatic properties of the iron minerals to separate

the valuable iron bearing constituents from impurities. The remainder utilize colloidal

interactions as a driving force for their separation processes. A Lake Superior District

hematite ore mine has been successfully concentrating 8.0 million tons of hematite ore

per year with a process called Selective Flocculation and Dispersion (Cliffs Natural

Resources, 2011). This process relies on many different reagents and the surface

chemistry of fine colloidal particles to separate hematite from siliceous impurities.

The zeta potential of particle surfaces is extremely important in selective flocculation

dispersion processes because it influences particle interactions and the effectiveness of

many separation reagents. The zeta potential of the particle surfaces defines the

dispersive stability of the particles in solution. Retaining a high dispersive stability is

essential for the desliming and flotation processes employed in selective flocculation-

dispersion concentrators.

34
Water softening agents such as sodium polyphosphates, sodium silicate and sodium

hydroxide are typically added as dispersants prior to selective flocculation and dispersion

of hematite from silica (Keranen, 1986). It has been noticed that magnesium ions have a

much more detrimental effect on selective flocculation and dispersion than calcium ions

at a local iron concentrator and the reasons behind this are poorly understood (Carlson,

2011).

2.1.3 Background

2.1.3.1 Zeta Potential and Ion Complexation

The pH of the water is the primary factor determining the surface charge of particles in

solution. This is well described by the two-pKa model as shown in Equation 2.1

(Laskowski et al., 1992; Stumm et al., 1992). At a low pH the surface is positively

charged, whereas, at a high pH, the surface is negatively charged as shown in Figure 1.4.

The pH at which the surface bears no charge is referred to as the point of zero charge

(PZC).

௣௄భ ௣௄మ
‫ܪܱ݁ܨ‬ଶା ርሮ ‫ ܪܱ݁ܨ‬ርሮ ‫ିܱ݁ܨ‬ Equation 2.1

The surface charge determines the charge of ions in solution that will be attracted to the

surface. The ions that are attracted to the surface either form inner-sphere or outer-sphere

complexes at the surface. An inner-sphere complex consists of a primarily covalent bond

between the ion and the surface. An outer sphere complex occurs when the ion

approaches the surface through electrostatic forces but is separated from the surface by

one or more water molecules, see Figure 1.4 (Stumm et al., 1992).
35
The electrostatic forces diminish with distance from the surface. As the electrostatic

forces diminish, the outer-sphere complexes can no longer be held to the surface and are

merely present as diffuse ions in the bulk fluid. The distance at which the electrostatic

forces are too weak to hold outer-sphere complexes to the surface is referred to as the

shear plane, see Figure 1.4. The potential difference between the shear plane and the bulk

solution is referred to as the zeta potential. The zeta potential of particle surfaces is

primarily a function of water chemistry and the underlying surface charge of the particle.

2.1.3.2 Hematite Crystal Structure

The hematite crystal structure is nearly identical to that of corundum; a nearly perfect

hexagonal close pack (HCP) structure, see Figure 2.1. This crystal structure is composed

of layers of iron and oxygen alternating in the [001] direction (Eggleston et al., 1992).

The oxygen layers have sets of triangles that are composed of both tightly bound oxygen

and loosely bound oxygen. The tightly bound oxygen (O-O) spacing is ~270pm; whereas

the loosely bound oxygen spacing is ~300pm, as determined by scanning tunneling

microscopy (Eggleston et al., 1992). When this crystal structure is exposed to an alkaline

aqueous environment, free iron atoms at the cleavage plane are hydrolyzed. The hydroxyl

groups at the surface have been calculated by vector math to be 285pm apart. This value

is in agreement with published literature for iron spacing on the surface of un-hydrated

hematite (Pradip, 1994; Ravishankar et al., 1995).

36
O Fe
O O Fe
O Fe O (a) Fe O
Fe O Fe O O Fe

[100]
O Fe O O Fe
Fe O Fe O Fe O
O Fe O O Fe
O O Fe

O Fe O-
O O Fe O-
O O-
Fe O Fe O
(b)
O-
Fe O O-
Fe O O Fe O-
O Fe O [100] O Fe O-
O-
Fe O O-
Fe O Fe O
O O-
Fe O O Fe O-
O O Fe O-

Figure 2.1: (a) Un-hydrated and (b) hydrated hematite surface structure in alkaline

conditions; frontal view and side view. Drawn using information from Eggleston et al.

(1992), Pradip (1994), and Ravishankar et al. (1995). All dimensions in picometers.

2.1.3.3 Divalent Cation Adsorption

Adsorption of divalent cations to the surface hydroxyl groups of iron oxides is strongly

dependent upon pH (Dzombak et al., 1990). Adsorption isotherms of calcium,


37
magnesium, strontium and barium onto an iron oxide surface are shown in Figure 2.2.

The surface must have negatively charged hydroxyl groups to attract cations and allow

for adsorption. As discussed earlier, a highly alkaline environment maximizes negatively

charge hydroxyl groups.

Figure 2.2: Adsorption isotherms for Magnesium (O'Brien et al., 1996), Calcium

(Kinniburgh et al., 1982), Strontium and Barium (Kinniburgh et al., 1976) onto a

hydrated iron oxide surface.

38
2.1.4 Materials and Methods

Concentrated iron ore was donated by a hematite ore concentration plant. The

concentrated ore was washed with distilled water in a pressure filter in order to remove

ions on the surface of the hematite particles. The washed solids were dried in a

convection oven at 107°C for 24 hours. A small sample (~50g) of the dried hematite

particles were then rehydrated with distilled water in a 300 mL beaker. This slurry was

immediately decanted to remove fine particles. This was repeated with 2 liters of 5%

HCl, 2 liters of 5% NaOH and 2 liters of distilled water to dissolve and wash away any

calcium or magnesium bearing minerals present in the ore that could dissolve during the

zeta potential tests. The coarse particles remaining were dried at 107ºC for 24 hours.

Once dry, the course particles were split into 1 gram samples with a micro-rotary riffle

splitter to ensure a representative sample was obtained for each test. The 1 gram sample

was then weighed and hydrated with distilled water for 24 hours prior to use.

An Anton Paar Electrokinetic Analyzer (EKA) with an automatic titrator and a powder

cell was used to measure the zeta potential of the coarse hematite particles as a function

of titrant dosage. The background solution was one liter of 1mM NaCl brought to a pH of

11 using 1 M NaOH. A pH of 11 was chosen for this study for two reasons. Alkaline

environments promote cation adsorption while inhibiting anion adsorption and the slurry

pH of a local selective flocculation – dispersion iron ore concentrator is held between

10.5 and 11.5 to maximize dispersion of ground particles (Keranen, 1986).

39
Titrants were prepared by adding distilled water to reagent grade MgCl2 • 6H2O, CaCl2 •

2H2O, SrCl2 • 6H2O, and BaCl2 • 2H2O. After the titration was complete, the final

background solutions were analyzed for magnesium, calcium, strontium, and barium

concentrations. Magnesium and calcium concentrations were measured using an atomic

absorption analyzer. Strontium and barium concentrations were measured via

Ethylenediamenetetraacetic acid (EDTA) potentiometric titrations using a water hardness

ion selective electrode. The concentration of ions at each step in the titration was

interpolated from the initial and final concentrations and plotted with their respective zeta

potential values. Each zeta potential measurement was repeated in triplicate to show

statistical significance in the results.

2.1.5 Results and Discussion

Very small concentrations of magnesium caused the zeta potential to invert from a

negative to a positive charge, as shown in Figure 2.3. A much larger concentration of

calcium was required to invert the zeta potential from a negative to a positive charge. To

determine if atomic size was the cause of this phenomena, larger ions, strontium and

barium, were used as titrants in a similar manner.

40
Figure 2.3: Zeta potential of concentrated natural hematite ore as a function of cation

concentration at a pH of 11.

The slope of the zeta potential curves is steeper at lower concentrations than at higher

concentrations of divalent cations. This shows that these cations may initially form inner-

sphere complexes until all available hydroxyl groups have been complexed, as shown in

Figure 2.1. Within inner-sphere complexes, the surface hydroxyl groups (-O-) act as

lewis bases, donating an electron and increasing the electron density of the coordinated

cation (Stumm et al., 1992).

41
O Fe O- Mg+
O O Fe O- Mg+
O O- + O
Mg
Fe Fe O
Mg
Fe
O O -+
Mg
Fe
O O -+
O Fe O- Mg+

[100]
O Fe O O Fe O- Mg+
O- + O
Mg
Fe O- + O
Mg
Fe
(a) Fe O
Ø278pm
O Mg
Fe
O O -+ O Fe O- Mg+
O O Fe O- Mg+
Ø278pm
-
O Fe O
O O Fe O- Ca+
O O-+
Ca
Fe O Fe O

Fe
O -
O Fe
O -
O O Fe O-

[100]
O Fe O O Fe O- Ca+
(b)
Fe O
Ø342pm O-+
Ca
Fe O O-+
Ca
Fe O
O Fe O-
O O-
Fe O
O Fe O- Ca+
O
Ø342pm
-
O Fe O
O O Fe O- Sr+
O O-+
Sr
Fe O Fe O

O-
Fe O O-
Fe O O Fe O-
[100]

O Fe O O Fe O- Sr+
(c)
Fe O
Ø370pm O-+
Sr
Fe O O-+
Sr
Fe O
O Fe O-
O O-
Fe O
O Fe O- Sr+
O
Ø370pm

O Fe O-
O O Fe O- Ba+
O O-+
Ba
Fe O Fe O

O-
Fe O O-
Fe O O Fe O-
[100]

(d)
O Fe O O Fe O- Ba+
Fe O
O-+
Ba
Fe O O-+
Ba
Fe O
Ø392pm
O Fe O-
O O-
Fe O
O Fe O- Ba+
O
Ø392pm

Figure 2.4: Inner-sphere complexes of magnesium, calcium, strontium and barium

formed at surface hydroxyl groups in the [100] plane. All measurements in picometers.
42
At the cleavage plane of the hematite hexagonal close packed crystal structure, the

average hydroxyl group spacing was calculated as 285 pm. This spacing is larger than the

single covalent diameter of magnesium (278 pm) but smaller than the single covalent

diameter of calcium (342 pm), strontium (370 pm), and barium (392 pm) (Pyykkö et al.,

2009); see Figure 2.4. Because of this spacing, magnesium inner-sphere complexes can

form at every hydroxyl group, however, calcium, strontium and barium inner-sphere

complexes can only form at every other hydroxyl group.

2.1.6 Conclusions

In the hematite ore selective flocculation and dispersion process, the zeta potential of the

particles is of particular importance because the particles must remain dispersed for

effective separation. Magnesium ions have shown a much more detrimental effect on

selective flocculation and dispersion than calcium ions at a local hematite concentration

plant and the reasons behind this phenomenon are poorly understood. The dependence of

the zeta potential of natural hematite concentrate on the concentration of cations at a pH

of 11 was studied using an electrokinetic analyzer. Results show that zeta potential

inverts from negative to positive at very low concentrations of magnesium ions in the

solution. It takes a significantly higher concentration of calcium, strontium and barium

ions to achieve the same effect. This difference is attributed to the ability of magnesium

ions to form inner-sphere complexes with all surface hydroxyl groups, whereas calcium,

strontium and barium, due to their larger atomic size, can only complex with every other

hydroxyl group. The tendency for calcium ions to form inner-sphere complexes with

43
every other surface hydroxyl group causes higher concentrations of calcium to be less

detrimental to hematite concentration processes than magnesium.

2.1.7 References

Carlson, J. J., 2011, Effect of Calcium and Magnesium on Deslime Iron Recovery,

Personal Communication.

Cliffs Natural Resources, 2011, Steel Starts Here, Operations Data Fact Sheet, 4 pp.

O'Brien, Robert J., Ajoy Raje, Robert L. Spicer, Liguang Xu, Shiqi Bao, Scott Lambert,

and Burtron H. Davis. "Technology Development for Iron Fischer-Tropsch Catalysts."

Proceedings of the First Joint Power & Fuel Systems Contractors Conference,

Pittsburgh, PA. 1996.

Dzombak, D. A., and Morel, F., 1990, Surface complexation modeling : hydrous ferric

oxide, Wiley, 393 pp.

Eggleston, C. M., and Hochella, M. F., 1992, "The Structure of Hematite (001) Surfaces

by Scanning Tunneling Microscopy - Image Interpretation, Surface Relaxation, and Step

Structure," American Mineralogist, Vol. 77, No. 9-10, pp. 911-922.

Haselhuhn, H. J., 2012, "Water Chemistry Effects on the Zeta Potential of Concentrated

Hematite Ore," Minerals & Metallurgical Processing, Vol. 29, No. 2, pp. 135-136.

Keranen, C. U., 1986, “Reagent Preparation, Distribution and Feeding Systems at the

Tilden Mine,” Design and Installation of Concentration and Dewatering Circuits, A. L.

44
Mular and M. A. Anderson eds., Society for Mining, Metallurgy and Exploration, pp.

308-319.

Kinniburgh, D. G., and Jackson, M. L., 1982, "Concentration and pH dependence of

calcium and zinc absorption by iron hydrous oxide gel," Soil Science Society of America

Journal, Vol. 46, No. 1, pp. 56-61.

Kinniburgh, D. G., Jackson, M. L., and Syers, J. K., 1976, "Adsorption of Alkaline Earth,

Transition, and Heavy Metal Cations by Hydrous Oxide Gels of Iron and Aluminum,"

Soil Science Society of America Journal, Vol. 40, No. 5, pp. 796-799.

Laskowski, J. S., and Ralston, J., 1992, Colloid Chemistry in Mineral Processing,

Elsevier Science, 428 pp.

Pradip, 1994, “Reagents Design and Molecular Recognition at Mineral Surfaces,”

Reagents for Better Metallurgy, Society for Mining Metallurgy, pp. 245-252.

Pyykkö, P., and Atsumi, M., 2009, "Molecular Single-Bond Covalent Radii for Elements

1–118," Chemistry – A European Journal, Vol. 15, No. 1, pp. 186-197.

Ravishankar, S. A., Pradip, and Khosla, N. K., 1995, "Selective flocculation of iron oxide

from its synthetic mixtures with clays: a comparison of polyacrylic acid and starch

polymers," International Journal of Mineral Processing, Vol. 43, No. 3–4, pp. 235-247.

Sparks, D. L., and Grundl, T. J., 1998, Mineral-Water Interfacial Reactions, American

Chemical Society, 438 pp.

45
Stumm, W., Sigg, L., and Sulzberger, B., 1992, Chemistry of the solid-water interface,

Wiley, 448 pp.

46
2.2 The Role of Water Chemistry in the Selective Flocculation and

Dispersion of Iron Ore at a Pilot Scale 3

2.2.1 Abstract

Fine grained hematite ore can be concentrated by the process of selective flocculation and

dispersion. This process relies on proper reagent selection and water chemistry. Many

previous studies have been performed that analyze the effects of different reagents on this

process in a laboratory environment. This paper focuses on the water chemistry within

the process in a pilot scale continuous deslime thickener.

The pH, sodium concentration, calcium concentration, and magnesium concentration

were varied to determine their effects on the iron concentrate grade and recovery, and the

phosphorus concentrate grade and rejection in the pilot scale selective deslime thickener.

It was found that the ideal pH for the iron grade and recovery of the process using a

starch selective flocculant is 10.5. Although, phosphorus rejection was increased at lower

pH values. Minimization of sodium concentration was shown to improve iron grade, iron

recovery and phosphorus rejection. Calcium acted as a nonselective flocculant showing

higher iron recovery, lower iron grade and lower phosphorus rejection with increasing

3
The material contained in this chapter was previously published in the journal
“Minerals and Metallurgical Processing Journal.” It is included in this dissertation with
permission from the journal.
Citation:
Haselhuhn, H. J, and Kawatra, S. K., 2015, "The Role of Water Chemistry in the
Selective Flocculation and Dispersion of Iron Ore," Minerals & Metallurgical
Processing, In Press
47
concentration. Conclusions were unable to be drawn with experiments varying in

magnesium concentration.

The zeta potential of the solid-liquid interface of particles in each sample taken was also

analyzed to show relationships between zeta potential and process performance. In all

cases, a maximization of the magnitude of zeta potential correlated with increased iron

grade and recovery. This supports the hypothesis that a higher level of dispersion

enhances the selective flocculation separation process.

2.2.2 Introduction

The United States produced 53.2 of the worldwide 3,000 million metric tons of

concentrated iron ore in 2012 (Tuck, 2013). With this in mind, there has been a large

increase in iron ore processing research (Bolen, 2014; Carlson et al., 2013; Carlson,

2008; Halt, 2014; Halt et al., 2014a; Halt et al., 2014b; Haselhuhn, 2012, 2013;

Haselhuhn et al., 2012a; Haselhuhn et al., 2012b; Liu et al., 2014; Manouchehri, 2014;

Sandvik et al., 2014; Semberg et al., 2014). Of the 53.2 million metric tons produced,

roughly 8 million metric tons of hematite (Fe2O3) concentrate were produced using a

process known as selective flocculation and dispersion (Cliffs Natural Resources, 2011).

This process is the only economically viable method for concentrating low grade (<40%

Fe) fine grain (<25μm liberation size) hematite ores. It is scarcely used due to difficulty

in controlling the surface chemistry of the ore and high reagent costs (Department of

Energy, 2001; Haselhuhn, 2013; Haselhuhn et al., 2012a). The process is highly

dependent upon both reagent selection and precise control of the water chemistry during

48
separation. This study shows the dependence of process performance of a pilot scale

deslime thickener on the water chemistry in the feed slurry.

The process used to concentrate hematite ore via selective flocculation and dispersion

shown in Figure 1.2 requires two concentration steps. The first and arguably most

important step is a selective thickening process known as selective flocculation desliming

(Siirak et al., 1988). Selective flocculation desliming is the focus of this research. During

this process, a dispersed slurry of liberated ore is treated with a selective flocculant that

acts by bridging hematite particles together. These flocs are settled in a thickening vessel

while the dispersed gangue minerals leave the process through the overflow. This process

requires three reagents to function effectively: Caustic soda, cooked corn starch and

sodium polyphosphate.

Caustic Soda (NaOH) is added during primary autogenous grinding as a pH modifier.

The caustic soda addition rate is adjusted to deliver a pH of between 10.5 and 11 at the

feed of the deslime thickener. This pH is required to both ensure adequate dispersion and

to facilitate maximum starch selectivity and adsorption (Green et al., 1984; Haselhuhn et

al., 2012a; Weissenborn, 1996).

Cooked acid modified corn starch is used as a selective flocculant for hematite ores. It is

added to the finely ground ore immediately following hydrocyclone sizing in the

hydrocyclone overflow conditioning tank. It has been speculated that adding starch

immediately following grinding is required because of mineral aging effects in the

process. Oxidation of freshly cleaved iron oxide surfaces is speculated to diminish starch

49
adsorption efficiency. This has been observed in laboratory studies though definitive

proof of this concept has never been shown.

A 21-member chain sodium polyphosphate is used as a dispersant for hematite ore. This

dispersant was chosen due to its ability to reject phosphorus bearing minerals, namely

apatite (Ca5(PO4)3(F,Cl,OH)), in both selective flocculation desliming and flotation

(Siirak et al., 1988). The sodium polyphosphate is added during secondary grinding to

ensure the dispersion of fine siliceous minerals that can coat the surfaces of the hematite

particles.

The selective flocculation and dispersion process depends upon the surface chemistry of

the particles. Because of this, the water used during the process is one of the most

important parameters. Even minor changes in water chemistry parameters can cause

drastic changes in process performance. In literature, it has been shown that the four

primary water quality parameters in the process water that affect the selective

flocculation and dispersion process are pH, sodium concentration, magnesium

concentration and calcium concentration (Haselhuhn, 2012; Haselhuhn et al., 2012a).

pH control is necessary to maintain dispersion within the process and to promote starch

adsorption and selectivity (Haselhuhn et al., 2012a; Weissenborn, 1996). The dispersive

quality of the mineral solution is a function of the zeta potential of the particles. The

magnitude of the zeta potential of all mineral particles is maximized in highly alkaline

conditions as shown in Figure 2.5. Two particles, regardless of mineralogy, with negative

zeta potentials of significant magnitude will remain dispersed from one another due to

50
electrostatic repulsion. Thus, a sufficiently high pH in the process water during selective

flocculation and dispersion is required to maintain this electrostatic repulsion.

One important aspect to note is that typical studies involving zeta potential, such as those

shown in Figure 2.5, are performed in low conductivity environments with aqueous ion

concentrations drastically lower than those seen during beneficiation. The zeta potential

of solid-liquid interfaces in mineral beneficiation plants typically do not correlate with

those performed in a perfect laboratory environment due to the high concentrations of

aqueous ions present and the complex mineralogy of the ores. This study intends to show

zeta potentials as they would be seen in an operating hematite beneficiation plant.

51
Figure 2.5: Zeta potential of silica, synthetic hematite and natural hematite (Haselhuhn

et al., 2012a)

The pH is also important to ensure starch adsorption and selectivity. Several studies have

shown maximized iron recovery during both flotation and selective flocculation

desliming at a pH of 10.5 (Weissenborn, 1996).

The sodium concentration within a selective flocculation dispersion process is of critical

importance because it has the highest concentration within the process and plays a minor

role in surface chemistry (Haselhuhn et al., 2012a). These high sodium concentrations, in

excess of 500 ppm, typically are the result of caustic use in pH control. Recent efforts to

lower the sodium concentration at a hematite selective flocculation and dispersion facility
52
have shown increases in both iron grade and recovery at the underflow of the deslime

thickener.

Calcium and magnesium concentrations within a selective flocculation dispersion process

are critical at even low doses due to their ability to non-selectively flocculate mineral

particles in alkaline conditions. The ions of these elements form chemical bonds with

oxide surfaces in alkaline conditions leaving a free positively charged surface site

available for adsorption or electrostatic effects. In significant concentration, the

selectivity of the process approaches zero. The polyphosphate dispersants for the process

mitigate the effects of calcium and magnesium by forming precipitates with negatively

charged surface sites. This can block starch adsorption but regains selectivity (Iwasaki,

1989).

Most published research studies regarding selective flocculation and dispersion have been

performed at a laboratory scale in water systems with no competing ions. The water

chemistry in an operating hematite beneficiation plant has been reported and reviewed by

Haselhuhn et al. (2012a). No previous studies on a pilot scale have been published

considering the effects of competing ions in a selective flocculation and dispersion

deslime thickener. The purpose of this study was to addresses this research gap.

2.2.3 Materials and Methods

A continuous pilot scale deslime thickener was built to show the dependence of water

quality on process performance. The key indications of process performance included

iron and phosphorus grade in the deslime thickener underflow, total recovery of iron, and

53
total phosphorus rejection. These parameters were determined via iron analysis,

phosphorus analysis and a mass balance performed at each operating condition.

2.2.3.1 Material Selection and Preparation

Hematite ore was sampled using a cross-flow sampling device at an operating hematite

beneficiation plant. The samples were taken after primary comminution but before

dispersant addition as shown in Figure 1.2. This location was chosen for three reasons:

(1) any location downstream of this location would have a surface active reagent within

the process water, (2) sampling after secondary comminution would allow the fresh

surface sites to oxidize prior to use in the laboratory thereby reducing starch selectivity as

discussed above, (3) a representative ore sample would be very difficult to obtain without

at least minimal comminution in the process. The downfall of sampling at this location

was that the sample had to be ground in the laboratory to liberation size.

The ore samples from the operating beneficiation plant were dewatered and dried using

decantation and a convection oven at 107°C. A particle size analysis of the dried material

showed an 80% passing size of 1.41 mm. The particle size required for an effective

separation is known to be 82% passing at 25 μm (Siirak et al., 1988). This particle size

was achieved through dry grinding in a laboratory ball mill for 130 minutes. To prevent

oxidation of the iron oxide surfaces prior to use, the ground samples were immediately

split using a rotary riffle splitter and inerted under an argon atmosphere. The ore

contained 42.3% iron and 0.0268% phosphorus indicating a higher than average grade

ore during the sampling process. An x-ray diffraction of the iron ore used in the study is

shown in Figure 2.6.


54
The water used in the process was created from distilled water mixed with magnesium

chloride hexahydrate (to adjust magnesium), calcium chloride dihydrate (to adjust

calcium), sodium hydroxide (to adjust pH), potassium chloride (to add potassium),

sodium carbonate (to add carbonate), sodium sulfate (to add sulfate) and sodium chloride

(to adjust sodium). These reagents were added to create a water profile similar to the

actual water chemistry found in the beneficiation process by Haselhuhn et al. (2012a).

The total volume of water for each experiment varied to hold a constant 2% solids by

weight in the feed slurry. This was necessary due to slight variations in ore mass in each

split as a result of the ore splitting process. Throughout the study, the concentration of

carbonate, sulfate and potassium ions were held constant at 400 ppm, 400 ppm, and 10

ppm, respectively. Four sets of experiments were carried out: pH modification, sodium

concentration modification, calcium concentration modification and magnesium

concentration modification. The water quality parameters for these experiments are

shown in Table 2.1.

55
Figure 2.6: X-ray diffraction pattern for iron ore sampled from screen underflow at an

operating hematite beneficiation plant

56
Table 2.1: Water quality parameters for each experiment

Experiment pH Ca (ppm) Mg (ppm) Na (ppm)


1 9 3 2 600
2 9.5 3 2 600
3 10 3 2 600
4 10.5 3 2 600
5 11 3 2 600
6 11 3 2 600
7 11 3 2 700
8 11 3 2 800
9 11 3 2 900
10 11 3 2 1000
11 11 0 2 600
12 11 3 2 600
13 11 6 2 600
14 11 9 2 600
15 11 12 2 600
16 11 3 0 600
17 11 3 2 600
18 11 3 4 600
19 11 3 6 600
20 11 3 8 600

2.2.3.2 Experimental Setup

The simulated process water and ground hematite ore were mixed in the feed tank of a

pilot scale continuous selective deslime thickener. A scaled diagram of the deslime

thickener is shown in Figure 2.7. A small mixer controlled by a variac was used to keep

the slurry agitated in the feed tank. A Masterflex Easy-Load peristaltic pump was used to

pump mixed feed slurry from the feed tank to the feed well of the thickener at a rate of 25

mL/sec. A peristaltic pump, Manostat Veristaltic Pump Jr., was used to remove the

underflow slurry from the thickener at a rate of 5 mL/sec. These flow rates were

optimized to obtain maximum iron grade and recovery at typical process conditions. The

thickener was constructed from 7.62 cm (3 inch) acrylic pipe with a stainless steel conical
57
bottom at a cone angle of 75°. The total height of the thickening unit was 138.43 cm

(54.5 inches). The thickener had a feed-well depth of 21.27 cm (8.375 inches). Under

these conditions, a 15 minute and 45 second underflow residence time and a 49 second

overflow residence time were achieved.

20.96cm.

7.84cm.

17.46cm.
21.27cm.

45.72cm.

126.52cm.
Deslime Feed
Thickener Tank
33.02cm.

53.66cm.

6.35cm.

Feed
U/F 2.54cm. Pump
Pump
11.75cm.

Figure 2.7: Diagram of the continuous pilot scale deslime thickener used in this study

58
To begin the experiment, the simulated process water was added to the feed tank shown

in Figure 2.7. The feed pump was disconnected from the thickener and allowed to

recirculate the process water back into the feed tank. Final adjustments in pH were made

using hydrochloric acid. Sodium polyphosphate (21-member chain) was added to the

process water at a dose of 0.023 g/kg (0.05 lb per metric ton). The ore sample specified

for the experiment was then added to the process water. Immediately after the ore was

added, cooked corn starch was added at a dose of 0.045 g/kg (0.10 lb per metric ton).

This mixture was allowed to recirculate in the feed tank for 5 minutes at which time the

feed pump was reconnected to the feed well of the deslime thickener. The underflow

pump was then activated to prevent an unmovable buildup of solids in the cone of the

thickener. Once the thickener filled and began overflowing, the flow rate of each pump

was adjusted to 5 mL/sec for the underflow and 20 mL/sec for the overflow. The system

was allowed to equilibrate for 30 minutes prior to sampling.

2.2.3.3 Analysis Methodology

The underflow stream, overflow stream, and feed stream of the thickener was sampled

using 500 mL glass sample bottles after the 30 minute equilibration period. These

samples were immediately measured for zeta potential using a Malvern Zetasizer. The

results from the Malvern Zetasizer reported the average zeta potential of the solid-liquid

interfaces within the slurry sample. The overflow and underflow samples were then

filtered. The volume of the filtrate was measured in a 500 mL graduated cylinder. The

filter cake from each sample was dried in a direct (non-convection) drying oven at 107°C.

The dry filter cake was weighed and measured for iron and phosphorus content. The iron

59
content was measured using a UV-VIS spectrometer via an orthophenanthroline

colorimetric reaction. The phosphorus concentration was measured using ASTM

Standard E-1070-11 (2011). The data gathered from the filter cake and filtrate analyses

were used to construct a mass balance and calculate iron recovery and phosphorus

rejection.

2.2.4 Results and Discussion

The results of the experiments confirmed many of the hypotheses found in prior literature

(Haselhuhn, 2012; Haselhuhn et al., 2012a; Pradip, 1994; Weissenborn, 1996). Optimal

pH conditions and water chemistry during the selective flocculation dispersion desliming

process were determined.

2.2.4.1 pH Modification Experiments

The iron grade versus recovery data for the pH modification experiments are shown in

Figure 2.8. This data shows that the optimal pH for enhancing both grade and recovery in

this process is 10.5. This pH has been shown to optimal in starch adsorption studies

performed by Weissenborn (1996). The selectivity of starch is due to the binuclear

complexation of un-aged Fe sites with adjacent oxide groups at the ends of the starch

polymer. The molecular recognition of starch to these surface sites is due to the matching

spacing between Fe surface sites on hematite surfaces (2.852 Å) and oxygen end groups

on starch polymers (2.85 Å) (Pradip, 1994). It is important for future studies to determine

why the pH of 10.5 is optimal. The enhanced adsorption of cooked corn starch in alkaline

conditions is due to the lack of hydrogen available to satisfy anionic oxygen groups at the

end monomers of the starch polymer. The downfall of this high pH is phosphorus
60
rejection. These experiments showed maximum phosphorus rejection at pH values lower

than 10. See Figure 2.9.

Figure 2.8: Iron grade versus recovery plot for pH experiments. Data labels indicate pH.

The best pH for iron grade and recovery was found to be around 10.5.

61
Figure 2.9: Phosphorus grade versus rejection plot for pH modification experiments.

Data labels indicate pH. Lower pH values yielded higher phosphorus rejection.

The zeta potential of the solid-liquid interfaces within the slurry is also a function of pH.

Figure 2.10 shows the zeta potential of the feed, underflow and overflow samples at

varying pH. The maximized magnitude of zeta potential (where zeta potential reaches its

highest negative value) of the solid-liquid interfaces in all slurry samples at a pH of 10.5

indicates the highest level of dispersive stability. This correlates well with the grade

versus recovery plot in Figure 2.8 showing that the grade and recovery of the deslime

thickener is associated with the zeta potential of the solid-water interfaces within the

slurry.

62
Figure 2.10: The relationship between zeta potential of feed, underflow and overflow

samples and pH is shown. A polynomial trend is drawn to show the minimum, most

dispersed, state. The 10-10.5 pH range yielded a maximization in the magnitude of zeta

potential suggesting higher dispersive stability.

2.2.4.2 Sodium Concentration Modification Experiments

The relationship between sodium concentration and iron grade and recovery is shown in

Figure 2.11. These data confirm the plant scale observations that decreased sodium

concentrations increase both the grade and the recovery of iron ore. Like the pH

modification, the increase in iron grade and recovery is accompanied by a decrease in

phosphorus rejection as shown in Figure 2.12. However; the phosphorus grade, under the

best iron recovery conditions, is still adequate to meet production goals (~0.04%/wt

Phosphorus) (Siirak et al., 1988). Sodium ions within the process water cause double

layer compression at the solid-liquid interfaces and ultimately diminish the dispersive
63
stability and zeta potential of solid-liquid interfaces within the slurry. This hinders starch

adsorption and selectivity.

Figure 2.11: Iron grade versus recovery plot for sodium concentration modification

experiments. Decreasing sodium yielded both an increase in iron grade and recovery.

64
High Na

Low Na

Figure 2.12: Phosphorus grade versus rejection plot for sodium concentration

modification experiments. Increasing sodium yielded a phosphorus higher rejection.

The zeta potential of the solid-liquid interfaces within the slurry is dependent not only on

the pH of the slurry but also on the ions within the slurry. A plot of the zeta potential of

the feed, underflow and overflow samples are shown in Figure 2.13 with a relation to the

sodium concentration within the slurry water. As expected, the sample with the highest

negative value of zeta potential also had the lowest sodium concentration. The

relationship between the zeta potential of solid-liquid interfaces and process performance

that was found during pH modification experiments can also be seen in the sodium

concentration modification experiments.

65
Figure 2.13: The relationship between zeta potential for feed, underflow and overflow

samples and sodium concentration. Increases in sodium yielded lower magnitudes of

zeta potential suggesting a loss in dispersive stability.

2.2.4.3 Calcium Concentration Modification Experiments

Calcium concentration plays a vital role in the selective flocculation and dispersion

desliming process. A plot of iron grade versus recovery at varying calcium concentrations

is shown in Figure 2.14. Calcium acts as a non-selective flocculant thereby diminishing

the selectivity of the process. As calcium concentration increases, more material at a

lower grade will fall to the underflow of the thickener. As expected, a higher calcium

concentration also causes lower phosphorus rejection as shown in Figure 2.15. This is

also due to non-selective flocculation caused by calcium adsorption at the solid-liquid

interface. Equation 2.2 and 2.3 show the reactions taking place at the iron oxide and

66
silicon oxide surface with calcium ions as reported by Dzombak and Morel (1990) and

Kitamura et al. (1999). Adsorbed calcium creates a positively charged surface site

thereby attracting negatively charged particles non-selectively.

High Ca

Low Ca

Figure 2.14: Iron grade versus recovery plot for calcium concentration modification

experiments. Increasing calcium increased the iron recovery at a lower grade.

Slight changes in calcium concentrations, with no competing ions present, can cause

drastic changes in the zeta potential of solid-liquid interfaces (Haselhuhn, 2012).

However; this water system has both sulfate and carbonate present thereby diminishing

the effects due to aqueous complexation. Ionic competition is occurring between these

ions and the charged surfaces as shown in Equation 2.2 - 2.7(Dzombak et al., 1990;

Kitamura et al., 1999; Parkhurst et al., 2013).

67
Low Ca

High Ca

Figure 2.15: Phosphorus grade versus rejection plot for calcium concentration

modification experiments. Decreasing calcium yielded an increase in phosphorus

rejection.

୪୭୥ሺ௞೔೙೟ ሻୀିହǤ଼ହ
‫ ܪܱ݁ܨ ؠ‬൅ ‫ܽܥ‬ଶା ርۛۛۛۛۛۛۛۛۛۛሮ ‫ܽܥܱ݁ܨ ؠ‬ା ൅ ‫ ܪ‬ା Equation 2.2

୪୭୥ሺ௞೔೙೟ ሻୀିଵǤହସ
‫ ܪܱ݅ܵ ؠ‬൅ ‫ܽܥ‬ଶା ርۛۛۛۛۛۛۛۛۛۛሮ ‫ܽܥܱ݅ܵ ؠ‬ା ൅ ‫ ܪ‬ା Equation 2.3

୪୭୥ሺ௞ሻୀଷǤଶଶସ
‫ܽܥ‬ଶା ൅ ‫ܱܥ‬ଷଶି ርۛۛۛۛۛۛۛሮ ‫ܱܥܽܥ‬ଷ Equation 2.4

୪୭୥ሺ௞ሻୀଵଵǤସଷହ
‫ܽܥ‬ଶା ൅ ‫ ܪ‬ା ൅ ‫ܱܥ‬ଷଶି ርۛۛۛۛۛۛۛۛሮ ‫ܱܥܪܽܥ‬ଷା Equation 2.5

୪୭୥ሺ௞ሻୀଶǤଷ଴଴
‫ܽܥ‬ଶା ൅ ܱܵସଶି ርۛۛۛۛۛۛۛሮ ‫ܱܵܽܥ‬ସ Equation 2.6

୪୭୥ሺ௞ሻୀଵǤ଴଼
‫ܽܥ‬ଶା ൅ ‫ܱܵܪ‬ସି ርۛۛۛۛۛۛሮ ‫ܱܵܪܽܥ‬ସା Equation 2.7

68
A plot of the zeta potential of the feed, underflow and overflow samples against calcium

concentration is shown in Figure 2.16. A slight increasing trend in zeta potential is

observed with increasing calcium concentration in all samples.

Figure 2.16: The relationship between zeta potential for feed, underflow and overflow

samples and calcium concentration. Increases in calcium caused a decrease in the

magnitude of the zeta potential.

2.2.4.4 Magnesium Concentration Modification Experiments

Magnesium, like calcium, has been speculated during other studies to play a role in the

selective flocculation and dispersion and flotation processes (Haselhuhn, 2012; Iwasaki,

1989). This trend was not observed during this study. The plot of grade and recovery of

iron is shown in Figure 2.17. Similar to the iron grade and recovery, the phosphorus

grade and rejection data showed no trends; see Figure 2.18.

69
Figure 2.17: Iron grade versus recovery plot for magnesium concentration modification

experiments. No conclusions could be drawn from these data.

70
Figure 2.18: Phosphorus grade versus rejection plot for magnesium concentration

modification experiments. No conclusions could be drawn from these data.

Changes in magnesium concentration with no competing ions present in the water have

shown large changes in zeta potential in other studies (Haselhuhn, 2012). Like calcium,

magnesium also participates in aqueous complexation reactions in the presence of sulfate

and carbonate ions. Ionic competition is occurring between these ions and the charged

surfaces as shown in Equation 2.8-2.12 (Dzombak et al., 1990; Kitamura et al., 1999;

Parkhurst et al., 2013).

୪୭୥ሺ௞೔೙೟ ሻୀିସǤ଺ Equation 2.8


‫ ܪܱ݁ܨ ؠ‬൅ ‫݃ܯ‬ଶା ርۛۛۛۛۛۛۛۛۛሮ ‫݃ܯܱ݁ܨ ؠ‬ା ൅ ‫ ܪ‬ା 

୪୭୥ሺ௞೔೙೟ ሻୀିଶǤ଴଺ Equation 2.9


‫ ܪܱ݅ܵ ؠ‬൅ ‫݃ܯ‬ଶା ርۛۛۛۛۛۛۛۛۛۛሮ ‫݃ܯܱ݅ܵ ؠ‬ା ൅ ‫ ܪ‬ା

71
୪୭୥ሺ௞ሻୀଶǤଽ଼ Equation 2.10
‫݃ܯ‬ଶା ൅ ‫ܱܥ‬ଷଶି ርۛۛۛۛۛۛሮ ‫ܱܥ݃ܯ‬ଷ

୪୭୥ሺ௞ሻୀଵଵǤଷଽଽ Equation 2.11


‫݃ܯ‬ଶା ൅ ‫ ܪ‬ା ൅ ‫ܱܥ‬ଷଶି ርۛۛۛۛۛۛۛۛሮ ‫ܱܥܪ݃ܯ‬ଷା

୪୭୥ሺ௞ሻୀଶǤଷ଻଴ Equation 2.12


‫݃ܯ‬ଶା ൅ ܱܵସଶି ርۛۛۛۛۛۛۛሮ ‫ܱܵ݃ܯ‬ସ

This study showed a slight decrease in the magnitude of zeta potential with increasing

magnesium concentration. This is due to adsorption of magnesium at the solid-liquid

interface of both iron oxide and gangue mineral particles causing charge reversal as seen

in Equation 2.8 and 2.9. The plot of zeta potential in relation to magnesium concentration

for the feed, underflow and overflow samples is shown in Figure 2.19.

Figure 2.19: The relationship between Zeta potential for feed, underflow and overflow

samples and magnesium concentration. Increases in magnesium caused a decrease in

the magnitude of the zeta potential.


72
2.2.5 Discussion, Analysis and Conclusions

This study was performed to show the effects of water chemistry on the selective

flocculation and dispersion desliming process for hematite ore. It was conducted in a pilot

scale continuous deslime thickener. Finely ground hematite ore was suspended in a

solution of simulated process water at varying pH, sodium concentration, calcium

concentration and magnesium concentration. Cooked acid modified corn starch was used

as a selective hematite flocculant and a 21-member sodium polyphosphate was used as a

dispersant.

During the pH modification experiments, pH was varied between 9 and 11. It was found

that a pH of 10.5 maximizes iron grade and recovery; though, higher phosphorus

rejection was found at lower pH values. The high iron grade and recovery was

accompanied by a maximization in the magnitude of zeta potential indicating that the iron

grade and recovery is a function of the level of dispersion in the thickener.

Sodium concentration was shown to have a profound effect on the process. The sodium

concentration was varied between 600 and 1000 ppm. Unlike traditional grade and

recovery curves, in which an increase in grade is associated with a decrease in recovery,

decreasing the sodium concentration within the deslime thickener caused an increase in

both grade and recovery. Similar to the conclusions found in the pH modification studies,

the samples with higher iron grade and recovery also showed a maximization in the

magnitude of zeta potential. This suggests that high sodium concentrations can lower the

dispersive stability of the slurry yielding lower grades in the product.

73
Calcium and magnesium have been previously shown to have an effect on the selective

flocculation and dispersion process. Both of these ions can adsorb to the oxide particle

surfaces causing a charge reversal and non-selective flocculation (Haselhuhn, 2012;

Iwasaki, 1989). During this study, calcium was varied from 0 to 12 ppm. The effect of

calcium concentration follows a standard grade recovery curve, in which higher grades

are accompanied by lower recoveries. As predicted by other studies, the iron recovery

increased with increasing calcium; however, the iron grade in the deslime underflow

decreased. This suggests non-selective flocculation. Increasing calcium also resulted in a

decrease in phosphorus rejection showing that the phosphorus bearing minerals

participate in the non-selective flocculation. Magnesium was also studied between 0 and

8 ppm. Changes in the magnesium concentration did not provide any significant

correlations to iron recovery or phosphorus rejection. The magnitude of zeta potential

during these studies was maximized at lower concentrations of calcium and magnesium.

This agrees well with prior studies showing highly negative zeta potentials at low

concentrations of calcium and magnesium (Haselhuhn, 2012).

In future studies, the results of this work will be applied directly to a full scale hematite

selective flocculation and dispersion beneficiation plant.

2.2.6 References

ASTM, 2011, E1070-11 Standard Test Method for Determination of Phosphorus in Iron

Ores by Phospho-Molybdenum-Blue Spectrophotometry, ASTM International,

Conshohocken, PA.

74
Bolen, J., 2014, "Modern Air Pollution Control for Iron Ore Induration," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 103-114.

Carlson, J. J., and Kawatra, S. K., 2013, "Factors Affecting Zeta Potential of Iron

Oxides," Mineral Processing and Extractive Metallurgy Review, Vol. 34, No. 5, pp. 269-

303.

Carlson, J. J. and Kawatra, S. K., 2008, "Effect of Particle Shape on the Filtration Rate in

an Industrial Iron Ore Processing Plant," Minerals & Metallurgical Processing, Vol. 25,

No. 3, pp. 165-168.

Cliffs Natural Resources, 2011, Steel Starts Here, Operations Data Fact Sheet, 4 pp.

Department of Energy, 2001. Selective Flocculation of Fine Mineral Particles. Office of

Industrial Technologies Energy Efficiency and Renewable Energy - U.S. Department of

Energy, Washington D.C.

Dzombak, D. A., and Morel, F., 1990, Surface complexation modeling : hydrous ferric

oxide, Wiley, 393 pp.

Green, R. E., and Colombo, A. F., 1984, “Dispersion-Selective Flocculation-Desliming

Characteristics of Oxidized Taconites,” US Bureau of Mines, Report of Investigations, RI

8867, 24 pp.

Halt, J. A., 2014, "Increasing the preheat strength of cornstarch-bonded pellets," Minerals

& Metallurgical Processing, Vol. 31, No. 3, pp. 179.

75
Halt, J. A., and Kawatra, S. K., 2014a, "Review of organic binders for iron ore

concentrate agglomeration," Minerals & Metallurgical Processing, Vol. 31, No. 2, pp.

73-94.

Halt, J. A., Nitz, M. C., Kawatra, S. K., and Dube, M., 2014b. "Iron Ore Pellet Dustiness

Part I: Factors Affecting Dust Generation," Mineral Processing and Extractive

Metallurgy Review, Vol. 36, No. 4, pp. 258-266.

Haselhuhn, H. J., 2012, "Water Chemistry Effects on the Zeta Potential of Concentrated

Hematite Ore," Minerals & Metallurgical Processing, Vol. 29, No. 2, pp. 135-136.

Haselhuhn, H. J., 2013, "Dispersant Adsorption and Effects on Settling Behavior of Iron

Ore," Minerals & Metallurgical Processing, Vol. 30, No. 3, pp. 188-189.

Haselhuhn, H. J., Carlson, J. J., and Kawatra, S. K., 2012a, "Water chemistry analysis of

an industrial selective flocculation dispersion hematite ore concentrator plant,"

International Journal of Mineral Processing, Vol. 102–103, pp. 99-106.

Haselhuhn, H. J., Swanson, K. P., and Kawatra, S. K., 2012b, "The effect of CO2

sparging on the flocculation and filtration rate of concentrated hematite slurries,"

International Journal of Mineral Processing, Vol. 112-113, pp. 107-109.

Iwasaki, I., 1989, “Bridging Theory and Practice in Iron Ore Flotation,” Advances in

Coal and Mineral Processing Using Flotation, Society for Mining, Metallurgy and

Exploration, pp. 177-190.

76
Kitamura, A., Fujiwara, K., Yamamoto, T., Nishikawa, S., and Moriyama, H., 1999,

"Analysis of Adsorption Behavior of Cations onto Quartz Surface by Electrical Double-

layer Model," Journal of Nuclear Science and Technology, Vol. 36, No. 12, pp. 1167-

1175.

Liu, S., Zhau, Y., Wang, W., and Wen, S., 2014, "Beneficiation of a Low-grade,

Hematite-Magnetite Ore in China," Minerals & Metallurgical Processing, Vol. 31, No. 2,

pp. 136-142.

Manouchehri, H. R., 2014, "Pyrrhotite Flotation and its Selectivity against Pentlandite in

the Beneficiation of Nickeliferous Ores: An Electrochemistry Perspective," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 115-125.

Parkhurst, D. L., and Appelo, C. A. J., 2013, “Description of input and examples for

PHREEQC version 3—A computer program for speciation, batch-reaction, one-

dimensional transport, and inverse geochemical calculations: U.S. Geological Survey

Techniques and Methods,” US Geological Survey.

Pradip, 1994, “Reagents Design and Molecular Recognition at Mineral Surfaces,”

Reagents for Better Metallurgy, Society for Mining Metallurgy, pp. 245-252.

Sandvik, K. L., and Larsen, E., 2014, "Iron Ore Flotation with Environmentally Friendly

Reagents," Minerals & Metallurgical Processing, Vol. 31, No. 2, pp. 95-102.

77
Semberg, P., Andersson, C., and Bjorkman, B., 2014, "Interaction between Iron Oxides

and Olivine in Magnetite Pellets during Reduction at 500°-1,300°C," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 126-135.

Siirak, J., and Hancock, B. A., 1988, “Progress in Developing a Flotation Phosphorus

Reduction Process at the Tilden Iron Ore Mine,” Proceedings of the XVI International

Mineral Processing Congress, Stockholm, Sweden, pp. 1393-1404.

Tuck, C. A., 2013, “Mineral Commodity Studies - Iron Ore,” US Geological Survey.

Weissenborn, P. K., 1996, "Behaviour of amylopectin and amylose components of starch

in the selective flocculation of ultrafine iron ore," International Journal of Mineral

Processing, Vol. 47, No. 3-4, pp. 197-211.

78
2.3 Measurement of Zeta Potential and Correlation with Performance

throughout an Operating Hematite Beneficiation Plant4

2.3.1 Executive Summary

The major separation processes in the operating hematite selective flocculation and

dispersion beneficiation plant (desliming, froth flotation, thickening, and filtration) are all

dependent on interactions of particles, which are largely controlled by the Zeta potential

of the particles. Measurement and control of the Zeta potential is therefore key to

controlling the plant operations and preventing process upsets. The objectives of this

work are therefore to measure how Zeta potential throughout the plant varies over time,

determine how these Zeta potential changes relate to plant performance, and develop

methods for estimating Zeta potential from parameters that can be quickly and

inexpensively measured on-site at the plant.

During this phase of the project, samples were collected 5 times at 12 locations

throughout section 7-9 of the concentrator. Sample locations were strategically chosen to

determine zeta potential effects at reagent addition points and zeta potential effects on

separation process performance. The following conclusions have been drawn from this

study:

4
Some of the material contained in this chapter was previously submitted to Cliffs
Natural Resources as a final report for the project titled “Measurement of Zeta Potential
and Correlation with Performance Phase 1”
79
1. Zeta potential is largely controlled by pH, ionic content of the process water,

conductivity of the process water, silica content, and surface active reagents (starch,

Glass-H (sodium henicosapolyphosphate))

2. Conductivity is related to the ionic content of the process water; however, it is most

closely associated with the sodium concentration and pH

3. Calcium and magnesium present as minerals in the ore are rejected by the deslime

thickener but not by the froth flotation cells.

4. pH has a significant effect on both zeta potential and deslime performance

4.1. Higher pH causes the surface charge to become more negative.

4.2. pH has a direct influence on the amount of iron in the deslime underflow.

4.3. The optimum pH for most calcium and magnesium conditions is near 11. This

pH was shown to yield the highest iron recoveries.

4.4. pH has a significant effect on conductivity.

5. Particle size control is important for influencing deslime weight rejection

6. Water hardness at the cyclone overflow has a large influence on deslime

performance. Higher water hardness causes non-selective flocculation of particles;

however, it allows for increased starch adsorption onto hematite surfaces

7. Starch does noticeably and reproducibly suppress the surface charge of hematite in

both deslime and flotation. The effect of the starch is consistently more prominent in

flotation than deslime due to higher iron concentration in particles

8. The addition of CO2 during dewatering has a significant effect on the pH,

conductivity, water hardness, and zeta potential.

80
9. Flux addition causes the zeta potential to equilibrate at ~-20 mV

10. Deslime recovery is maximized at a lower conductivity, lower water hardness and

higher magnitude of zeta potential

11. Flotation recovery is maximized at the opposite conditions: higher conductivity,

higher water hardness, and lower magnitude of zeta potential, though flotation

recovery is largely dependent on deslime iron recovery and the iron content of the

material fed into flotation

Based on these results, it has been found that monitoring the water chemistry and zeta

potential at the deslime thickener offers the most potential for increasing iron recovery. It

is theorized that pH and water hardness at the deslime thickener can be optimized to yield

a higher overall recovery. Another way that could potentially increase overall iron

recovery is increasing the water hardness in the flotation feed distributors by adding

slaked lime.

2.3.2 Introduction

The major separation processes in the plant (desliming, froth flotation, thickening, and

filtration) are all dependent on the interactions of particles, which are largely controlled

by the Zeta potential of the particles. Measurement and control of the Zeta potential is

therefore key to controlling the plant operations and preventing process upsets. Zeta

potential can be controlled by adjusting the water quality, adding flocculants or

dispersants, or by adjusting the incoming feed ore.

The objectives of this work were to measure how Zeta potential throughout the plant

varies over time, determine how these Zeta potential changes relate to plant performance,
81
and develop methods for estimating Zeta potential from parameters that can be quickly

and inexpensively measured on-site at the plant.

During phase 1 of this project, samples were collected 5 times at 12 locations throughout

section 7-9 of the concentrator. Sample locations were strategically chosen to determine

zeta potential effects at reagent addition points and zeta potential effects on separation

process performance. Samples were tested for water quality parameters including pH,

conductivity, water hardness, calcium, magnesium, iron, sodium, potassium, sulfate,

carbonate, nitrate and oxidation-reduction potential. The solid content of each sample

was analyzed for zeta potential (at process conditions), elemental composition,

thermogravimetric degradation, x-ray diffraction, x-ray fluorescence and particle size

distribution.

2.3.3 Background

The zeta potential of particle surfaces is extremely important in selective flocculation

dispersion processes because it influences particle interactions and the effectiveness of

many separation reagents. The zeta potential of the particle surfaces defines the

dispersive stability of the particles in solution. Retaining a high dispersive stability is

essential for the gravity separation and flotation processes employed in selective

flocculation-dispersion concentrators.

2.3.3.1 Zeta Potential and Ionic surface Complexation

The pH of the water is the primary factor determining the surface charge of particles in

solution. This is well described by the two-pKa model as shown in Equation 2.13

82
(Laskowski et al., 1992; Stumm et al., 1992). At a low pH the surface is positively

charged, whereas, at a high pH, the surface is negatively charged as shown in Figure 1.4.

The pH at which the surface bears no charge is referred to as the point of zero charge

(PZC).

௣௄భ ௣௄మ
‫ ݁ܨ‬െ ܱ‫ܪ‬ଶା ርሮ ‫ ݁ܨ‬െ ܱ‫ ܪ‬ርሮ ‫ ݁ܨ‬െ ܱି Equation 2.13

The surface charge determines the charge of ions in solution that will be attracted to the

surface. The ions that are attracted to the surface either form inner-sphere or outer-sphere

complexes at the surface. An inner-sphere complex consists of a primarily single covalent

bond between the ion and the surface. An outer sphere complex occurs when the ion

approaches the surface through coulombic forces but is separated from the surface by one

or more water molecules, see Figure 1.4 (Stumm et al., 1992).

The coulombic forces diminish with distance from the surface. As the coulombic forces

diminish, the outer-sphere complexes can no longer be held to the surface and are merely

present as diffuse ions in the bulk fluid. The distance at which the coulombic forces are

too weak to hold outer-sphere complexes to the surface is referred to as the shear plane,

see Figure 1.4. The potential difference between the shear plane and the bulk solution is

referred to as the zeta potential. The zeta potential of particle surfaces is primarily a

function of water chemistry and the underlying surface charge of the particle.

2.3.3.2 Calcium and Magnesium Adsorption

Sorption of both calcium and magnesium ions to the surface hydroxyl groups of iron

oxides is strongly dependent upon pH (Dzombak et al., 1990). Sorption isotherms of

83
calcium and magnesium onto an iron oxide surface are shown in Figure 2.20. The surface

must have negatively charged hydroxyl groups to attract cations and allow for adsorption.

As discussed earlier, a highly alkaline environment maximizes negatively charge

hydroxyl groups thereby maximizing the adsorption capability of the iron oxide.

Figure 2.20: Adsorption isotherms for Magnesium (O'Brien et al., 1996), Calcium

(Kinniburgh et al., 1982), Strontium and Barium (Kinniburgh et al., 1976) onto a

hydrated iron oxide surface.

2.3.3.3 Cation Adsorption and Zeta Potential

Sodium ions slightly decreased the magnitude of zeta potential as concentration increased

as shown in Figure 2.21. This is caused by outer-sphere complexation causing the zeta
84
potential to become more positive. Very small concentrations of magnesium caused the

zeta potential to invert from a negative to a positive charge. A much larger concentration

of calcium was required to invert the zeta potential from a negative to a positive charge.

The slope of the zeta potential curves is steeper at lower concentrations than at higher

concentrations of divalent cations. This shows that these cations initially form inner-

sphere complexes until all available hydroxyl groups have been complexed.

Figure 2.21: Zeta potential of concentrated natural hematite ore as a function of cation

concentration at a pH of 11.

Within inner-sphere complexes, the surface hydroxyl groups (-O-) act as lewis bases,

donating an electron and increasing the electron density of the coordinated cation. This
85
causes a single covalent bond to form between the hydroxyl group and the coordinated

cation, be it calcium (Ca2+) or magnesium (Mg2+) (Stumm et al., 1992).

The reason for the difference between how Calcium and Magnesium affect zeta potential

is unknown; however, it has been suspected that it is due to the size of the ions,

magnesium being smaller than calcium. At the cleavage plane of the hematite hexagonal

close packed crystal structure, the average hydroxyl group spacing has been reported as

being between 285.2 and 291.3 pm (Eggleston et al., 1992; Pradip, 1994). This spacing is

larger than the single covalent diameter of magnesium (278 pm) but smaller than the

single covalent diameter of calcium (342 pm) (Pyykkö et al., 2009). Because of this

spacing, magnesium inner-sphere complexes can form at every hydroxyl group, however,

calcium inner-sphere complexes can only form at every third hydroxyl group as shown in

Figure 2.22.

86
O Fe O- Mg+
O O Fe O- Mg+
O O- + O
Mg
Fe Fe O
Mg
Fe
O O -+
Mg
Fe
O O -+
O Fe O- Mg+

[100]
O Fe O O Fe O- Mg+
O- + O
Mg
Fe O- + O
Mg
Fe
(a) Fe O
Ø278pm
O Mg
Fe
O O -+ O Fe O- Mg+
O O Fe O- Mg+
Ø278pm
-
O Fe O
O O Fe O- Ca+
O O-+
Ca
Fe O Fe O

O-
Fe O Fe
O -
O O Fe O-

[100]
O Fe O O Fe O- Ca+
(b)
Fe O
Ø342pm O-+
Ca
Fe O O-+
Ca
Fe O
O Fe O-
O O-
Fe O
O Fe O- Ca+
O
Ø342pm

Figure 2.22: Hematite surface structure frontal view and side view. (a) Inner-sphere

magnesium complexes formed at surface hydroxyl groups. (b) Inner-sphere calcium

complexes formed at surface hydroxyl groups. All dimensions in picometers.

2.3.3.4 Ionic Equilibrium and Water Chemistry

2.3.3.4.1 Ion Pairing

Ions in solution can carry a positive, negative or neutral charge. Positively charged ions,

such as sodium (Na+), are attracted to negatively charged ions, such as carbonate (CO32-).

Ion pairing can result when electrostatic interactions between cations and anions is large

enough to cause the two ions to become one ionic entity. An example of an ion pair is

shown in Equation 2.14 below:

87
௞ೌೞೞ೚೎ ୀଵ଴భǤమళ
ା ଶି ି
ܰܽሺ௔௤ሻ ൅ ‫ܱܥ‬ଷǡሺ௔௤ሻ ርۛۛۛۛۛۛۛۛሮ ܰܽ‫ܱܥ‬ଷǡሺ௔௤ሻ Equation 2.14

The tendency for an ion pair to form can be determined by its association constant, kassoc.

The association constant for monovalent anionic sodium carbonate (NaCO3-) is defined in

Equation 2.15:

ି
ൣܰܽ‫ܱܥ‬ଷǡሺ௔௤ሻ ൧
݇௔௦௦௢௖ ൌ ቆ ା ଶି

ൣܰܽሺ௔௤ሻ ൧ ή ൣ‫ܱܥ‬ଷǡሺ௔௤ሻ ൧ Equation 2.15
௘௤

As can be seen, a higher association constant means the ions are more likely to associate

with each other than dissociate from one another.

2.3.3.4.2 Aqueous Complex Formation

Complexes are very similar to ion pairs, however, they involve chemical interactions

between the ions and not just electrostatic interactions. A common interaction seen in the

concentrator is between iron (Fe3+) and water as shown in Equations 2.16-2.19:

௞భ ୀଵ଴షమǤభవ
ଷା ଶା ା
‫݁ܨ‬ሺ௔௤ሻ ൅ ‫ܪ‬ଶ ܱ ርۛۛۛۛۛۛሮ ‫ܪܱ݁ܨ‬ሺ௔௤ሻ ൅ ‫ܪ‬ሺ௔௤ሻ Equation 2.16

௞మ ୀଵ଴షఱǤలళ
ଷା
‫݁ܨ‬ሺ௔௤ሻ ൅ ʹ‫ܪ‬ଶ ܱ ርۛۛۛۛۛۛሮ ‫݁ܨ‬ሺܱ‫ܪ‬ሻା ା
ଶǡሺ௔௤ሻ ൅ ʹ‫ܪ‬ሺ௔௤ሻ Equation 2.17

௞య ୀଵ଴షభమǤఱల
ଷା ା
‫݁ܨ‬ሺ௔௤ሻ ൅ ͵‫ܪ‬ଶ ܱ ርۛۛۛۛۛۛۛሮ ‫݁ܨ‬ሺܱ‫ܪ‬ሻଷǡሺ௔௤ሻ ൅ ͵‫ܪ‬ሺ௔௤ሻ Equation 2.18

88
௞ర ୀଵ଴షమభǤల
ଷା ା
‫݁ܨ‬ሺ௔௤ሻ ൅ Ͷ‫ܪ‬ଶ ܱ ርۛۛۛۛۛۛሮ ‫݁ܨ‬ሺܱ‫ܪ‬ሻି
ସǡሺ௔௤ሻ ൅ Ͷ‫ܪ‬ሺ௔௤ሻ Equation 2.19

Based on the equilibrium constants, it can be seen that these reactions become less and

less favored as the complex increases in hydroxide groups. The progression of these

reactions is highly pH dependent; at a lower pH (higher concentration of H+), the reaction

proceeds backwards towards Fe3+, however at a higher pH (lower concentration of H+),

the reaction proceeds forwards towards Fe(OH)4-. A graphical distribution of ferric

hydroxide (Fe3+(OH-)n) complexes as a function of pH is shown in Figure 2.23.

‫ ݁ܨ‬ାଷ
‫ ܪܱ݁ܨ‬ାଶ
‫݁ܨ‬ሺܱ‫ܪ‬ሻାଶ
‫݁ܨ‬ሺܱ‫ܪ‬ሻଷ
‫݁ܨ‬ሺܱ‫ܪ‬ሻିସ

Figure 2.23: Distribution of ferric hydroxide complexes as a function of pH

89
2.3.3.4.3 Sparingly Soluble Salts

Some salts such as calcium carbonate, CaCO3 (Calcite) do not fully dissolve in water and

are called sparingly soluble salts. Several of the reagents added to the process are

sparingly soluble salts including calcite and dolomite. Calcite and dolomite can dissolve

and release calcium and magnesium ions into the solution. The amount that a sparingly

soluble salt dissolves in water is determined by its solubility product:

௞ೞ೛ ୀସǤ଼ήଵ଴షవ
ଶା
‫ܱܥܽܥ‬ଷǡሺ௦ሻ ርۛۛۛۛۛۛۛሮ ‫ܽܥ‬ሺ௔௤ሻ ଶି
൅ ‫ܱܥ‬ଷǡሺ௔௤ሻ Equation 2.20

ଶା ଶି
݇௦௣ ൌ ൣ‫ܽܥ‬ሺ௔௤ሻ ൧ ή ൣ‫ܱܥ‬ଷǡሺ௔௤ሻ ൧ Equation 2.21

An increase in the concentration of either calcium or carbonate ions in solution causes the

formation of aqueous calcium carbonate which can form solid calcite, especially if solid

calcite particles are already present for aqueous calcite to precipitate on.

2.3.3.4.4 Calcium and Magnesium Carbonate/Sulfate/Hydroxide Equilibrium

In a solution of calcium, magnesium and carbonate ions, the pH of the solution will

dictate which ionic pairs, ionic complexes and salt precipitates will form. An example of

this would be the formation of aqueous magnesium carbonate, and magnesium sulfate

followed by monovalent magnesium hydroxide (MgOH+) as pH increases as shown in

Figure 2.24 (a). A similar equilibrium distribution is seen with calcium; see Figure 2.24

(b).

90
‫݃ܯ‬ାଶ ‫ܱܵ݃ܯ‬ସ ‫ܱܥ݃ܯ‬ଷ ‫ି ܪܱ݃ܯ‬

‫ܽܥ‬ାଶ ‫ܱܵܽܥ‬ସ ‫ܱܥܽܥ‬ଷ ‫ି ܪܱܽܥ‬

Figure 2.24: (a) Molar equilibrium distribution of magnesium species in process water.

(b) Molar equilibrium distribution of calcium species in process water. Cyclone overflow

water chemistry data from April 2nd, 2012 was used for this plot.

91
As discussed previously and shown in Figure 2.21, both calcium and magnesium cations

significantly affect zeta potential because they are chemically adsorbed to the surface of

particles. These ions cause non-selective flocculation. From the equilibrium distributions

of calcium and magnesium shown in Figure 2.24, it can be seen that pH control is

necessary to minimize the amount of divalent and monovalent cations present in the

process water. The total amount of neutral, monovalent and divalent cations from the

calcium and magnesium distributions is shown in Figure 2.25. Maximizing the amount of

neutral ions in the water minimizes detrimental ions in the solution. The optimum pH in

these process water conditions was found to be 10.99. At this optimum pH, dispersion of

particles is maximized allowing for better separation in the deslime thickener.

92
Figure 2.25: Molar distribution of ionic species by valency. Divalent ions include Mg2+

and Ca2+. Monovalent ions include CaHCO3+, CaOH+, MgHCO3+ and MgOH+. Neutral

ions include CaSO4, CaCO3, MgSO4, and MgCO3.

2.3.4 Methods and Materials

2.3.4.1 Sample Collection and Process Locations

The original sampling locations were chosen by Josh Carlson in 2007 based on reagent

additions and the process flow sheet. The sampling locations that were originally chosen

are shown in Figure 2.26 and Figure 2.27 below and described in Table 2.2. The only

change to these locations that has been made was the removal of the cyclone feed sample

(location 3). Samples at locations 1 through 7 were taken on sections 7-9 at the
93
concentrator. Starting at the flotation feed distributor, samples at locations 8 through 11

were taken on sections 7-12 because the float feed from sections 7-12 are mixed at this

point. Samples at location 12 and 13 were taken from the concentrate underflow (12) and

the polymer addition point (13).

Figure 2.26: Original sampling locations chosen by Dr. Joshua Carlson in 2007

94
Figure 2.27: Sampling locations in full process flow diagram. Adapted from Keranen

(1986).

95
Table 2.2: Sample location descriptions

Location # Acronym Description


1 SUNGH Screen underflow before dispersant addition
2 SUGH Screen underflow after dispersant addition
3 - Cyclone feed
4 CO1 Cyclone overflow before starch addition
5 CO2 Cyclone overflow after starch addition
6 DTO Deslime thickener overflow
7 DTU Deslime underflow before starch addition
8 FlF Flotation circuit feed after starch addition
9 SR Scavenger return
10 ST Scavenger tailings
11 RC Rougher concentrate
12 CTU Concentrate thickener underflow
13 FiF Fluxed filter feed

Once the samples were taken, they were immediately transported back to Michigan Tech

for analysis. Testing priority was given to the zeta potential analysis of the starched

samples because the starch used in the process can degrade over time. When all sample

testing was complete, the remaining solids were sent to an external laboratory for

XRD/XRF analysis.

2.3.4.2 Solids Analysis Procedure

2.3.4.2.1 Zeta Potential Analysis

Slurry samples were analyzed for zeta potential using an Anton Paar® Electrokinetic

Analyzer™ (EKA). To analyze the zeta potential of a slurry, the solid particles and liquid

96
supernatant must be separated. Settled solids were taken from the bottom of the sample

buckets and deslimed in a beaker with distilled water so that only the coarse particles (~ >

25 μm) would be analyzed. This size fraction was chosen due to the particle size

limitation of the EKA. The supernatant liquid of the slurry was centrifuged to remove

suspended solids prior to use as a background electrolyte in the EKA.

First, the zeta potential of the slurry was analyzed at the process conditions. Then the zeta

potential was analyzed as a function of pH by sequential titration using 200 mM HNO3

down to a pH of < 3.0 in increments of 0.5 pH units. The data from these titrations are

plotted in Appendix A.

Samples were not tested in succession by location number but rather by expected

concentration of starch from highest to lowest due to the potential degradation of the

starch over time. The flotation feed sample (location 8) was the first to be analyzed

followed by the deslime thickener feed sample (location 5) as these samples were taken

immediately after the starch was added. The rest of the samples were run as expediently

as possible following those two samples with the last samples to be tested being locations

1 through 4 because starch had not been introduced at those locations.

2.3.4.2.2 Dry Solids Preparation

Settled solids were re-suspended in distilled water and filtered using a 9 cm Whatman

541 filter medium. The solids were dried in a convection oven at 105°C for 24 hours. The

dried solids were then transferred to sealed plastic bags for analysis. Dry solids were

analyzed for percent iron, percent calcium, percent magnesium, percent acid insoluble

97
material (silica), particle size, and XRD/XRF. A thermogravimetric analysis (TGA) was

also conducted on each sample.

2.3.4.2.3 Elemental Analysis (%Fe, %Ca, %Mg, % Acid Insolubles)

Dried solid samples were split using a micro rotary riffle splitter down to a ~1 g sample.

The split sample was then digested in a beaker according to ASTM standards using 50

mL of concentrated HCL on a stir plate at 200°C for 20 minutes (ASTM, 2006). The

digested solution was diluted to 200 mL with distilled water in a class-A volumetric

flask, shaken, and allowed to settle for 2 hours. This solution was then diluted into the

appropriate background solutions for iron, calcium and magnesium analysis by atomic

absorption spectroscopy (AAS). AAS results were used to back-calculate the percent

iron, percent calcium, and percent magnesium in the samples. Percent aluminum was

attempted; however, the data was not consistent due to overwhelming AAS interference

by the iron in solution.

The remaining digested solution was filtered using a weighed Gooch crucible and

Whatman 42 filter media. The Gooch crucible was dried at 105°C for 24 hours and

weighed to determine the percent acid insoluble material in the digested solution.

2.3.4.2.4 Thermogravimetric Analysis

Dried solid samples were split by the four-corners grab sample method and analyzed in a

LECO TGA701 thermogravimetric analyzer. The method ramped temperature up to

107°C at 8°C/min and held until constant weight was achieved. The final mass at

constancy was reported as the moisture content of the sample. The next step was a ramp

98
from 107°C to 207°C at a rate of 8°C/min and held until constant weight was achieved.

The final mass at constancy was reported as the volatile organics content of the sample.

The final step was a ramp from 207°C to 1000°C at a ramp rate of 1°C per minute.

During this time period, many decompositions occur in each sample depending on the

mineralogy. The decompositions most noted are as follows:

ଶଶହିଷଶହι஼
ʹ‫ܱ݁ܨ‬ሺܱ‫ܪ‬ሻ ሱۛۛۛۛۛۛሮ ‫݁ܨ‬ଶ ܱଷ ൅ ‫ܪ‬ଶ ܱ Equation 2.22

ଷଶହିସଵହι஼
Ͷ‫݁ܨ‬ଷ ܱସ ൅ܱଶ ሱۛۛۛۛۛۛሮ ͸‫݁ܨ‬ଶ ܱଷ Equation 2.23

ଷଶହିଵ଴଴଴ι஼
‫ܱܥ݁ܨ‬ଷ ሱۛۛۛۛۛۛۛሮ ‫ ܱ݁ܨ‬൅ ‫ܱܥ‬ଶ Equation 2.24

ଷଶହିଵ଴଴଴ι஼
‫ܱܥܽܥ‬ଷ ሱۛۛۛۛۛۛۛሮ ‫ ܱܽܥ‬൅ ‫ܱܥ‬ଶ Equation 2.25

ଷଶହିଵ଴଴଴ι஼
‫݃ܯܽܥ‬ሺ‫ܱܥ‬ଷ ሻଶ ሱۛۛۛۛۛۛۛሮ ‫ ܱ݃ܯ‬൅ ‫ ܱܽܥ‬൅ ‫ܱܥ‬ଶ Equation 2.26

To distinguish between the decomposition of siderite (FeCO3), calcite (CaCO3), and

dolomite (CaMg(CO3)2), the TGA data was compared to the elemental analysis to isolate

decompositions due to calcium, and magnesium. The remainder of the carbonate

decomposition was assumed to be due to iron in siderite. All thermolysis graphs can be

found in Appendix B.

99
2.3.4.2.5 Particle Size Analysis

Dried solid samples were split by the four-corners grab sample method and analyzed for

particle size using a MicroTrac™ SRA9200 laser scattering particle size analyzer. The

raw particle size data for all samples can be found in Appendix C.

2.3.4.2.6 XRD/XRF

X-ray Diffraction and X-ray Fluorescence was conducted on the dried solid samples by

Thomas Campbell at Cliffs Natural Resources.

2.3.4.3 Water Chemistry Analysis Procedure

2.3.4.3.1 Sample Preparation

The liquid samples were prepared by centrifugation followed by filtration using 9 cm

Whatman 42® 2 μm filter media. Samples were then diluted with a background solution

for analysis by either atomic absorption spectroscopy (AAS), direct measurement via an

ion selective electrode (ISE), or gravimetric precipitation followed by thermogravimetric

analysis (TGA). Background solutions were prepared according to ASTM standards and

recommendations found in the Agilent Technologies AAS Analytical Methods Manual.

2.3.4.3.2 pH and Conductivity

pH and conductivity were measured both at the mine during sample collection and in the

laboratory at Michigan Technological University. Laboratory pH and conductivity are

reported here because the portable measurement devices can be less accurate than the

stationary laboratory devices. pH and conductivity were measured in the laboratory using

100
the Electrokinetic Analyzer (EKA). The laboratory pH and conductivity meters were

calibrated within 24 hours of sample collection.

2.3.4.3.3 Calcium and Total Water Hardness

Calcium and total hardness were tested by the mine on each sample. The test entails a

colorimetric EDTA (ethylene-diamine-tetraacetic acid) titration. The indicator for this

type of titration is typically Eriochrome Black T which turns red in the presence of metal

ions. The addition of EDTA (a stronger complexing reagent) causes the indicator to turn

blue when all of the metal ions have been complexed.

2.3.4.3.4 Sodium Concentration

Sodium concentration was determined on the 11/30/2011 samples and the two sets of

4/2/2012 samples. The 11/30/2011 samples were analyzed directly using a calibrated

Sodium ISE. The results from this electrode are given as the activity of the ion, which,

for sodium is very close to the total concentration. Upon further development of reliable

laboratory procedures, the total concentration of sodium was measured using AAS with

an air-acetylene flame for the 4/2/2012 samples. The background solution was (1+99)

HCL + 2000 ppm KCl.

2.3.4.3.5 Potassium Concentration

Potassium concentrations were measured on the final sample set to verify that they were

consistent with levels previously reported (Haselhuhn et al., 2012a). Potassium is also a

small contributor to conductivity. Total potassium concentration was measured by AAS

101
with an air-acetylene flame for the 4/2/2012 samples (ASTM, 2010d). The background

solution was (1+99) HCl + 2000 ppm NaCl.

2.3.4.3.6 Calcium Concentration

Calcium concentration was inferred from calcium water hardness for all samples except

the 4/2/2012 samples. The calcium concentrations of these samples were measure using

AAS with a nitrous oxide-acetylene flame (ASTM, 2010a). The background solution was

(1+99) HCl + 2000 ppm KCl. The AAS results were comparable to the calcium hardness

results reported by the mine.

2.3.4.3.7 Magnesium Concentration

Magnesium concentration was measured by AAS using an air-acetylene flame on the

11/30/2011 and the 4/2/2012 samples (ASTM, 2010a). The background solution was

(1+99) HCl + 2000 ppm KCl. Magnesium concentrations did not match well with the

total hardness measurements reported by the mine.

2.3.4.3.8 Soluble Iron Concentration

Soluble iron concentration was measured by AAS using an air-acetylene flame on the

4/2/2012 samples (ASTM, 2010c). The background solution was (3+99) HCl. Iron was

measured to determine if trends in iron levels were consistent with those found in a

previous study (Haselhuhn et al., 2012a).

102
2.3.4.3.9 Chloride Concentration

Chloride is a small contributor to conductivity, which can be easily be measured online

and may aid in increasing iron recovery. Chloride concentration was measured using a

chloride ISE on the 11/30/2011 and 4/2/2012 samples (ASTM, 2010b).

2.3.4.3.10 Nitrate Concentration

Nitrate concentration, though very minimal, is a small contributor to conductivity as well.

Nitrate was measured using a nitrate ISE on the 4/2/2012 samples mostly to compare

with trends found in a previous study (Haselhuhn et al., 2012a).

2.3.4.3.11 Carbonate and Sulfate Concentration

Total carbonate and sulfate concentration was measured simultaneously by gravimetric

precipitation with strontium chloride followed by thermogravimetric analysis. Aqueous

strontium precipitates readily with both sulfate and carbonate in solution to form a white

particle suspension by the following reactions (Knovel, 2003).

௄ೞ೛ ୀଵ଴షఴǤఴ
ଶା ଶି
ܵ‫ݎ‬ሺ௔௤ሻ ൅ ‫ܱܥ‬ଷǡሺ௔௤ሻ ሱۛۛۛۛۛۛሮ ܵ‫ܱܥݎ‬ଷǡሺ௦ሻ Equation 2.27

௄ೞ೛ ୀଵ଴షలǤఱఱ
ଶା ଶି
ܵ‫ݎ‬ሺ௔௤ሻ ൅ ܱܵସǡሺ௔௤ሻ ൅ ʹǤ͹‫ܪ‬ଶ ܱ ሱۛۛۛۛۛۛۛሮ ܵ‫ܱܵݎ‬ସ ή ʹǤ͹‫ܪ‬ଶ ܱሺ௦ሻ
Equation 2.28

After precipitation, the solids were filtered and weighed to determine the total amount of

SrSO4·2.7H2O and SrCO3 present. The amount of CO32- and SO42- was determined

103
quantitatively by the amount of thermal decomposition during thermogravimetric

analysis. Thermal decomposition of a mixture of SrSO4·2.7H2O and SrCO3 proceeds as

follows.

ଶ଺଴ିସ଴଴ι஼
ܵ‫ܱܵݎ‬ସ ή ʹǤ͹‫ܪ‬ଶ ܱሺ௦ሻ ൅ ܵ‫ܱܥݎ‬ଷǡሺ௦ሻ ሱۛۛۛۛۛۛሮ ܵ‫ܱܵݎ‬ସǡሺ௦ሻ ൅ ܵ‫ܱܥݎ‬ଷǡሺ௦ሻ ൅ ʹǤ͹‫ܪ‬ଶ ܱሺ௚ሻ Equation
2.29

଼ହ଴ିଵ଴଴଴ι஼
ܵ‫ܱܵݎ‬ସǡሺ௦ሻ ൅ ܵ‫ܱܥݎ‬ଷǡሺ௦ሻ ሱۛۛۛۛۛۛۛሮ ܵ‫ܱܵݎ‬ସǡሺ௦ሻ ൅ ܵ‫ܱݎ‬ሺ௦ሻ ൅ ‫ܱܥ‬ଶǡሺ௚ሻ Equation
2.30

This method has not been reported in literature, however, has been proven to be accurate

to ±0.87% using synthetic carbonate and sulfate reagents.

2.3.4.3.12 Oxidation-Reduction Potential

Oxidation-Reduction Potential (ORP) was measured electrochemically with an ORP

electrode. ORP was measured to give an indication of whether the process water was an

oxidizing environment or a reducing environment. This is an important factor for surface

chemistry as it predicts the propensity of free iron sites to oxidize.

2.3.4.3.13 Total Dissolved Solids

Total dissolved solids (TDS) were measured on the April 2nd, 2012 sample sets by an

external laboratory. 500 mL samples were prepared by centrifugation and filtration.

2.3.4.3.14 Equilibrium Distribution Modeling

The equilibrium distribution of ionic species in complex solutions can be modeled using

many different software packages such as PHREEQC, which is freeware created and
104
distributed by USGS. Equilibrium distributions for this report have been assembled using

this software.

2.3.5 Results and Discussion

2.3.5.1 Overfrothing – May 5th, 2011

2.3.5.1.1 Introduction

The goal of this work was to determine what water chemistry factors play a significant

role in the magnitude of the zeta potential and if there is an inexpensive method for

monitoring zeta potential (or inferring it from other measurements) at any point within

the plant. The May 5th, 2011 sample date was chosen due to poor operating conditions at

the concentrator. The flotation tailings was said to be “overfrothing”. This phenomenon

was attributed to humic acid buildup in the reuse water due to melting snow and runoff.

Although this hypothesis could not be confirmed, this sample set was successful in

delivering a base set of data that the rest of the project could build on.

2.3.5.1.2 pH

pH was recorded both at the plant and during the zeta potential measurements as shown

in Figure 2.28. The pH of the plant water slowly decreased throughout the concentrator

from ~11 post-comminution down to ~10.7 at the concentrate thickener. The pH did not

drift much between the time when the samples were taken and when they were analyzed

at Michigan Technological University except for the concentrate thickener underflow

sample at location 12, see Figure 2.28. The concentrate thickener underflow sample rose

105
in pH from 8.85 to 9.45. This was probably due to the slow dissolution of flux in the

recycled concentrate added to the concentrate thickener.

Figure 2.28: pH of each sample by location on May 5th, 2011

2.3.5.1.3 Conductivity

Conductivity was recorded at the plant using a portable conductivity meter and during the

EKA zeta potential measurements. This was done to ensure that any conductivity drift

between sampling and testing was recorded. The results of conductivity testing are shown

in Figure 2.29. Conductivity slowly decreased throughout the plant from ~3000 μS/cm

post-comminution down to ~240 μS/cm at the rougher concentrate.

106
Figure 2.29: Conductivity of each sample by location on May 5th, 2011

2.3.5.1.4 Calcium and Total Hardness

Calcium hardness and total hardness tests were conducted by the plant personnel and the

results are shown in Figure 2.30. The water hardness was at or below 30 ppm except for

at the deslime underflow (location 7), the scavenger returns (location 9), and the

concentrate underflow (location 12). The reason for the high calcium and total hardness

seen at the deslime underflow and scavenger returns was unknown but could be a result

of sampling error or testing error. The concentrate underflow water hardness was strongly

affected by the addition of recycled fluxed concentrate. As the flux dissolved at the lower

pH, the water hardness was increased substantially.

107
Figure 2.30: Calcium hardness and total hardness for each sample by location on May

5th, 2011

2.3.5.1.5 Zeta Potential

The zeta potential at process conditions (pH, conductivity) are shown in Figure 2.31. It

was very apparent that the starch suppressed the surface charge of the hematite particles

by comparing sample locations 4 and 5 (cyclone overflow before and after starch

addition) and sample locations 7 and 8 (deslime underflow before and after starch

addition). Zeta potential versus pH plots of these samples are included in Appendix A.

The starch addition at the cyclone overflow and deslime underflow showed depressions

in surface charge by 6 mV and 14 mV, respectively.

108
Figure 2.31: Zeta potential of each sample at process conditions by location on May 5th,

2011.

Other notable findings with zeta potential included a more negative deslime overflow

zeta potential than deslime underflow zeta potential. This suggested that silica was being

removed in the deslime thickener because silica has much more negative zeta potential

than hematite at the pH of the deslime thickener (pH=~10.7). The elemental analysis

confirmed this observation by showing that the deslime overflow silica content (67.94%

acid insol.) was much higher than the deslime underflow silica content (19.21% acid

insol.). This observation was also seen at the scavenger tails sample. The zeta potential of

109
the scavenger tails was the most negative showing that the abundance of silica (64.88%

acid insol.) in the sample has a direct effect on the zeta potential.

A plot of zeta potential versus silica content can be seen in Figure 2.32. The general trend

was that higher silica content yielded higher zeta potentials. The trend line was not a

perfect match due to the use of many surface active reagents in the process including

starch (hematite depressant/selective flocculant), glass-H (dispersant), etheramine

(cationic collector and frother), and recycled fluxed hematite concentrate (general

flocculant).

Figure 2.32: The Effect of silica content on zeta potential of hematite ore. Most outliers

are due to the addition of surface altering flocculants throughout the process.

110
2.3.5.1.6 Process Data

Process data for May 5th, 2011 can be seen in Table 2.3. The sample set taken on May 5th,

had calcium hardness values between 4 and 18 ppm (as CaCO3) throughout deslime and

flotation accompanied by a 67.6% overall recovery (deslime overflow (DTO) at 16.2%

Fe and scavenger tails (ST) at 22.3% Fe). The high iron content of the scavenger tailings

sample on May 5th was accompanied by high water hardness values at the flotation feed.

Table 2.3: Process data for May 5th, 2011.

Date: 5/5/11 Time: 7:00 AM


Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.9 100 34.9 100.0 11.0 lbs/ton CYC O/F 35.3 100 Crude lbs/ton
DTU 48.6 57.7 28.1 80.4 pH Plant 0.175 37.7 67.6 DTU 47.8 60.5 1.85
DTO 16.2 42.3 6.8 19.6 11.0 DTO 16.3 39.5
Flot Tail 22.3 22.2
%Passing at 500 Mesh 83.2 Flot Conc 62.5 38.3
Date: 5/5/11 Time: 1:00 PM
Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.9 100 34.9 100.0 10.8 lbs/ton CYC O/F 35.3 100 Crude lbs/ton
DTU 51.5 53.8 27.7 79.3 pH Plant 0.175 39.1 70.0 DTU 49.3 59.6 1.85
DTO 15.6 46.2 7.2 20.7 11.0 DTO 14.5 40.4
Flot Tail 22.1 19.5
%Passing at 500 Mesh 83.4 Flot Conc 62.5 40.1

2.3.5.1.7 Conclusions

The zeta potential of the slurry particles seemed to be the product of several process

conditions. pH and conductivity are always factors in the zeta potential of any colloidal

system. Other likely factors in the determination of zeta potential are charge suppressants

such as starch, ionic collectors and flocculants, and the concentration of siliceous particle

species.

111
Several conclusions were made about the zeta potential at the concentrator including:

1. Starch did noticeably and reproducibly suppress the surface charge of hematite in

both deslime and flotation

2. The effect of the starch was more prominent in flotation

3. The addition of CO2 had a significant effect on the pH, conductivity, water hardness,

and zeta potential. These additions resulted in a

3.1. Decrease in pH

3.2. Increase in water hardness

3.3. Increase in zeta potential

2.3.5.2 Normal Operations – July 13th, 2011

2.3.5.2.1 Introduction

The goal of this work was to determine what water chemistry factors play a significant

role in the magnitude of the zeta potential and if there is an inexpensive method for

monitoring zeta potential (or inferring it from other measurements) at any point within

the plant. The July 13th, 2011 sample set was taken when sections 7-9 were running well.

This was the first set of data taken on a normal processing day and provided a basis that

can be used for comparing data from abnormal processing days.

2.3.5.2.2 pH

The pH of the process on May 5th, 2011 remained between 10.5 and 10.8 until

dewatering as shown in Figure 2.28. On July 13th, the pH was held above 11 throughout

much of the process as shown in Figure 2.33. A higher pH typically caused a higher

112
magnitude of zeta potential of the iron and silica particles resulting in greater dispersion.

In both sample sets, the pH dropped considerably upon the addition of CO2. This was to

be expected as CO2 dissolves in water as carbonic acid (H2CO3) and immediately loses

protons (H+) to free hydroxide ions (OH-) thereby lowering the pH. Depending on the

equilibrium pH, the carbonate ions in solution will either be in the form of bicarbonate

(HCO3-) or carbonate (CO32-) ions:


௣ு̱଺Ǥଶ ௣ு̱ଵ଴Ǥହ
ି ା ଶି ା
‫ܱܥ‬ଶሺ௚ሻ ൅ ‫ܪ‬ଶ ܱሺ௔௤ሻ ՜ ‫ܪ‬ଶ ‫ܱܥ‬ଷሺ௔௤ሻ ሱۛۛۛሮ ‫ܱܥܪ‬ଷሺ௔௤ሻ ൅ ‫ܪ‬ሺ௔௤ሻ ሱۛۛۛۛሮ ‫ܱܥ‬ଷሺ௔௤ሻ ൅ ʹ‫ܪ‬ሺ௔௤ሻ

Equation 2.31

The filter feed sample (location 13) of the May 5th sample set was taken too close to the

polymer mixing point. This caused the sample to be extremely viscous and testing was

stopped for fear of equipment damage. The July 13th filter feed sample was taken when

the process was stabilized and the results showed an increase in pH. The dissolution of

calcium and magnesium from flux (calcium/magnesium carbonates) can neutralize some

of the carbonic acid resulting in the increase in pH shown at the filter feed.

For the July 13th sample set, the deslimed solids from the deslime thickener overflow

were extremely fine (80% passing at 18.45 μm) and could not be measured with the

electrokinetic analyzer (EKA). The pH readings in Figure 2.33 were taken using the EKA

explaining the loss of the location 6 data point in Figure 2.33. This can be considered a

good thing, in that, the deslime thickener was removing slimes and fine particles with

little loss of larger hematite particles that day.

113
Figure 2.33: pH of each sample by location on July 13th, 2011

2.3.5.2.3 Conductivity

The conductivity of the May and July sample sets exhibited the same decreasing trend

throughout the process, however, the July post-comminution samples were significantly

higher in conductivity than the May samples. See Figure 2.34. This indicated a difference

in ore mineralogy, makeup water contents, and/or caustic/dispersant (Glass-H) addition

rates. In both cases, the conductivity stabilized in the mid 2000’s after deslime

thickening.

114
Figure 2.34: Conductivity of each sample by location on July 13th, 2011

2.3.5.2.4 Calcium and Total Hardness

The water hardness tests were somewhat erratic between sampling dates as shown in

Figure 2.35.There are several commonalities between the two sample sets: the deslime

starch addition at locations 4 and 5 did not significantly affect hardness, and the addition

of CO2 drastically increased the water hardness. Based on the idea that starch does not

affect hardness, the dramatic decrease present in the May 5th samples at the amine and

starch addition (locations 7-8) must be attributed to an error in sampling/testing, an effect

caused by the amine, or the reason behind the over-frothing in the flotation cells seen that

115
day. These fluctuations may be attributed to the higher calcium content of the ore during

the May 5th samples (0.49% Ca on May 5th, 0.24% Ca on July 13th).

Another notable phenomenon occurred in the concentrate thickener underflow upon the

addition of CO2. The hardness data suggest that the addition of CO2 dissolves the flux in

the recycled fluxed concentrate. This is most likely due to the dramatic increase in

solubility of calcite and dolomite at the decreased pH.

Increasing the hydrogen content of the water (lowering the pH) causes calcite (and

dolomite) to dissolve because the hydrogen will displace calcium in calcite and form a

bicarbonate ion (or carbonic acid at low pH) and a free calcium ion. This is the source for

increased water hardness upon addition of CO2 to the slurry.

The addition of flux was accompanied with a decrease in water hardness as well as an

increase in pH. This showed that a new equilibrium was being established with the

increased concentration of flux in the slurry.

116
Figure 2.35: Calcium hardness and total hardness for each sample by location on July

13th, 2011

2.3.5.2.5 Zeta Potential

The zeta potentials at the process conditions (pH, conductivity) are shown in Figure 2.36.

The first noticeable trend was the increase in zeta potential with the addition of Glass-H.

This appeared on both sampling dates. This phenomenon was postulated to be a result of

phosphate ion adsorption onto un-hydrated hematite and silica surfaces.

It was very apparent that the starch suppresses the surface charge of the hematite particles

by comparing samples 4 and 5 (cyclone overs before and after starch addition) and

samples 7 and 8 (flotation feed before and after starch addition). This observation was
117
consistent in both sample sets with a more pronounced effect at the flotation starch

addition. Zeta potential versus pH plots of these samples are included in Appendix A.

Both sample sets also showed a consistent 5mV increase in zeta potential upon the

addition of CO2. This is to be expected with the pH change and has been shown to

increase filtration rates (Haselhuhn et al., 2012b). In the July sample, the flux addition

caused a decrease in zeta potential which may have been negating the effects of the CO2

on filtration.

Figure 2.36: Zeta potential of each sample at process conditions by location on July 13th,

2011.
118
Other notable findings with zeta potential included a more negative scavenger tails zeta

potential than flotation feed zeta potential. A similar trend between silica content and zeta

potential was observed for this sample set as well. This suggested that silica is being

effectively removed during flotation because silica has a much more negative zeta

potential than hematite at the pH of flotation (pH=~10.7).

2.3.5.2.6 Process Data

Iron recovery, metallurgical balance, and reagent addition rates can be found in Table

2.4. The recovery on May 5th was 4.4% lower than on July 13th. 1.3% of this decrease in

recovery is due to extra losses in deslime. Water chemistry differences between these two

days included a lower pH, lower water hardness and a diminished effectiveness of starch

adsorption on May 5th. The latter water chemistry properties may play a significant role

in the selectivity and effectiveness of the starch. A diminished selectivity of starch would

cause losses of iron to the deslime overflow.

The sample set taken on July 13th had calcium hardness values between 15 and 24 ppm

(as CaCO3) throughout deslime and flotation accompanied by a maximum overall

recovery of 72.0% (deslime overflow (DTO) at 14.9% Fe and scavenger tails (ST) at

18.4% Fe). The low flotation tailings iron content on July 13th was accompanied by low

water hardness at the flotation feed. This consistency shows a need to monitor water

hardness and pH levels at the float feed distributor as there may be a connection between

iron rejection, water hardness, and pH in flotation.

119
Table 2.4: Process Data on July 13th, 2011

Date: 7/13/2011 Time: 7:00 AM


Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.7 100.0 34.7 100.0 11.1 lbs/ton CYC O/F 34.5 100.0 Crude lbs/ton
DTU 47.0 61.7 29.0 83.5 pH Plant 0.16 39.9 72 DTU 45.6 65.5 1.85
DTO 14.9 38.3 5.7 16.5 11.1 DTO 13.6 34.5
Flot Tail 18.4 25.2
%Passing at 500 Mesh 78.9 Flot Conc 62.6 40.3
Date: 7/13/2011 Time: 1:00 PM
Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.2 100.0 34.2 100.0 11.2 lbs/ton CYC O/F 34.1 100.0 Crude lbs/ton
DTU 47.6 59.1 28.2 82.3 pH Plant 0.14 38.5 70.5 DTU 46.4 62.2 1.85
DTO 14.8 40.9 6.0 17.7 11.1 DTO 13.9 37.8
Flot Tail 19.6 23.4
%Passing at 500 Mesh 80.9 Flot Conc 62.6 38.8

2.3.5.2.7 Conclusions

The July 13th sample set confirmed several theories made about how the zeta potential is

affected by process conditions. The following are the conclusions drawn from this sample

set:

1. Starch did noticeably and reproducibly suppress the surface charge of hematite in

both deslime and flotation

2. The effect of the starch was consistently more prominent in flotation

3. Starch did not significantly change the water hardness

4. The addition of CO2 had a significant effect on the pH, conductivity, water hardness,

and zeta potential. These additions resulted in a

4.1. Decrease in pH

4.2. Increase in conductivity

4.3. Increase in water hardness


120
4.4. Increase in zeta potential

2.3.5.3 Crushers Shut Down – November 30th, 2011

2.3.5.3.1 Introduction

The goal of this work was to determine what water chemistry factors play a significant

role in the magnitude of the zeta potential and if there is an method for monitoring zeta

potential (or inferring it from other measurements) at any point within the plant. The

November sample set was taken when the deslime and flotation processes had a very high

iron rejection and the crushers were shut down for maintenance.

Several process conditions were noticeably different on this day:

1. Flotation and deslime had a very high weight rejection with high iron

concentration in tailings

2. Water hardness values were extraordinarily low throughout the plant (<10 ppm)

3. Flotation overflow appeared stagnant in some areas

4. Crushers were shut down for maintenance

5. Plant flow-rates were turned down

6. Low flow rates caused inadequate distribution of slurry between roughers

2.3.5.3.2 pH

The pH of the process water was reported at 11.0 from the plant labs. This matched well

with the pH at the screen underflow sample however it was not representative of the

entire process during this sampling campaign as shown in Figure 2.37. The pH tended to

decrease throughout the process and hit a low point at 10.28 at the flotation feed. This can
121
cause the zeta potential to increase therefore hindering collector performance and overall

particle dispersion. On May 5th, the same trend in pH occurred, but not to such an extent.

This was also accompanied with higher than normal loss of iron in flotation.

The pH decreases substantially upon the addition of CO2 at the concentrate underflow

followed by a slight increase when flux is added. This is to be expected because dolomite

dissolves and establishes equilibrium at a pH between 8 and 9. This effect at the

dewatering stage has been seen in all sample sets taken.

Figure 2.37: pH of each sample by location on November 30th, 2011

122
2.3.5.3.3 Conductivity

The conductivity throughout the process followed the same trends as in previous sample

sets: starts high and decreases throughout until CO2 and dolomite are added; see Figure

2.38.

Figure 2.38: Conductivity of each sample by location on November 30th, 2011

2.3.5.3.4 Sodium and Chloride

The sodium and chloride content of the process water, shown in Figure 2.39, were

measured using ion selective electrodes (ISEs). The chloride concentration remained

fairly steady between 375 and 475 ppm throughout. The sodium concentration throughout

the plant followed very closely with the conductivity of the process water, varying
123
between 750 and 1050 ppm. In this sample set, the conductivity correlated linearly very

well with the sodium concentration of each sample with a correlation coefficient of 0.986

as shown in Figure 2.40. This is a very good finding because sodium ions decrease the

magnitude of zeta potential as shown in Figure 2.21.

Figure 2.39: Sodium and chloride concentrations of each sample by location on

November 30th, 2011

124
Figure 2.40: Correlation between sodium and conductivity of the November 30th, 2011

sample set

Sodium ions enter the solution via one of five routes. Some sodium is naturally brought

in from the ore and re-use water. Sodium is also added to the process from caustic

(NaOH), Glass-H ((NaPO3)21*Na2O), and sodium silicate (Na2SiO3).

2.3.5.3.5 Calcium, Magnesium and Water Hardness

Common alkaline earth metals (Magnesium and Calcium) are the primary contributors to

water hardness. Control of Mg2+ and Ca2+ ions is essential to maintaining high iron

recoveries in deslime and flotation. Previous studies and ongoing research have shown a

strong link between the concentrations of calcium, magnesium, water hardness, zeta
125
potential and iron recovery. The calcium, magnesium, and total hardness concentrations

of the each location in the process are shown in Figure 2.41.

Figure 2.41: Magnesium, calcium and total hardness concentrations of each sample by

location on November 30th, 2011

This sample set had very low concentrations of calcium and magnesium content

compared to previous sample sets. These values were accompanied by the lowest total

iron recoveries so far at 59.1% (7:00 AM) - 59.2% (1:00 PM); see Table 2.5. Soon after

the calcium concentrations were determined (~ 10 AM), the Glass-H (dispersant) rate

was lowered from 0.06 to 0.03 lb/ton which yielded an immediate improvement in

126
deslime recovery. The deslime overflow iron content was 16.2% Fe at 7:00 AM and was

lowered to 11.1% Fe by 1:00 PM.

The flotation iron rejection increased between these two times. There were several factors

affecting this:

1. Overgrinding causing overabundance of fines that are easily rejected by flotation

2. Change in Glass-H rate did not affect 1:00PM mass balance at flotation because

of high residence time in pebble mill and deslime

The large concentration of magnesium (1.5 and 3.1) between deslime and flotation

caused magnesium to specifically adsorb to surface sites on hematite particles, lowering

zeta potential, see Figure 2.21. This high magnesium concentration was most likely due

to the low pH (10.3) at the flotation feed.

2.3.5.3.6 Zeta Potential

The zeta potential at process conditions (pH, conductivity) are shown in Figure 2.42. The

first noticeable trend was the minimal change in zeta potential with the addition of Glass-

H (dispersant). In previous sample sets Glass-H increased the zeta potential more

substantially, as seen in Figure 2.31 and Figure 2.36.

It is very apparent that the starch suppressed the surface charge of the hematite particles

by comparing samples 4 and 5 (cyclone overflow before and after starch addition) and

samples 7 and 8 (flotation feed before and after starch addition). This observation was

127
consistent in all sample sets with a more pronounced effect at the flotation starch

addition. Zeta potential curves of these samples are included in the Appendix A.

All sample sets have shown an increase by 5 – 9 mV in zeta potential upon the addition

of CO2. The addition of flux decreases the zeta potential to ~20 mV. This may cause

negative effects on filtration performance.

Figure 2.42: Zeta potential of each sample by location on November 30th, 2011

2.3.5.3.7 Process Data

Iron recovery, metallurgical balance, and reagent addition rates can be found in Table

2.5. The total wt% recovery on section 7-9 was 32.1% and 31.8% at 7:00 AM and 1:00

128
PM, respectively, with a target recovery of 42.5 wt%. This was significantly lower than

on any previous sample dates due to high weight rejection in the deslime thickener. This

high rejection is caused by abundance of fines due to overgrinding and lower than normal

water hardness. Flotation had a high iron rejection due to the lower than normal pH and

high magnesium content in the float feed. Overgrinding on section 7-9 caused an 81.7 %

and 86.3% passing at 500 mesh at 7:00 AM and 1:00 PM, respectively. This was higher

than any other section.

Table 2.5: Process data for November 30th, 2011

Date: 11/30/11 Time: 7:00 AM


Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34 100 34.0 100.0 11.0 lbs/ton CYC O/F 34 100 Crude lbs/ton
DTU 46.8 58.2 27.2 80.1 pH Plant 0.06 32.1 59.1 DTU 46.8 58.2 1.87
DTO 16.2 41.8 6.8 19.9 11.0 DTO 16.2 41.8 Conc lbs/ton
Flot Tail 27.3 26.3 5.27
%Passing at 500 Mesh 81.7 Flot Conc 62.7 32.1
Date: 11/30/11 Time: 1:00 PM
Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 33.6 100 33.6 100.0 10.9 lbs/ton CYC O/F 34 100 Crude lbs/ton
DTU 47 62.7 29.5 87.7 pH Plant 0.03 31.8 59.2 DTU 46.1 66.1 1.96
DTO 11.1 37.3 4.1 12.3 11.0 DTO 10.4 33.9 Conc lbs/ton
Flot Tail 30.9 34.4 5.45
%Passing at 500 Mesh 86.3 Flot Conc 62.7 31.7

2.3.5.3.8 Elemental Analysis

The concentration of calcium, magnesium, and iron (Fe2+ and Fe3+) was measured by acid

leaching followed by atomic absorption spectroscopy (ASTM, 2006); see Figure 2.43 and

Figure 2.44. The calcium and magnesium content of the solids were very similar to each

other throughout the process signifying some dolomite present in the ore. Upon flux

addition, the concentration of calcium rose more than magnesium suggesting the use of a
129
calcite-rich flux. Some notable trends include calcite/dolomite rejection during deslime

followed by a high level calcium/magnesium concentration at the flotation feed. The

source of this higher than expected concentration of calcium and magnesium prior to

flotation was most likely due to the merging of section 7-9 and section 10-12 in the float

feed distributor. Section 10-12 may have contained more calcium and magnesium than

section 7-9.

Figure 2.43: Calcium and magnesium content of the solids of each sample by location

on November 30th, 2011

The iron content and acid insoluble material content of the slurry solids are shown in

Figure 2.44. Acid insolubles are typically silica and any phosphate minerals that are not
130
dissolved in hot concentrated hydrochloric acid. As expected, there was a large acid

insolubles rejection during deslime and flotation where silica and other gangue minerals

were removed. The iron was concentrated by both the deslime and flotation, however,

there were high iron losses in the tailings (14.85% in deslime and 31.6% in flotation).

The inconsistency between the deslime underflow and the flotation feed was caused by

the merging of process sections 10-12 with sections 7-9 at that point. There seems to have

been a higher silica concentration in the other sections.

Figure 2.44: Iron and acid insolubles content of the solids of each sample by location on

November 30th, 2011

131
2.3.5.3.9 Conclusions

This ‘emergency’ sampling was performed because of low iron recoveries; especially on

sections 7-9. Several process conditions were noticeably different on this day:

1. Flotation and deslime had a very high weight rejection with high iron

concentration in tailings

2. Water hardness values were extraordinarily low throughout the plant (<10 ppm)

3. Flotation overflow appeared stagnant in some areas

4. Crushers were shut down for maintenance

5. Plant flow-rates were turned down

6. Low flow rates caused inadequate distribution of slurry between roughers

Overgrinding due to inadequate crushing of critical size material yielded an abundance of

fine material that contributed to the high weight rejection in both deslime and flotation.

Low water hardness at deslime caused over-dispersion of the abundant fines and

contributed to the high weight rejection in the deslime thickener.

The pH of the process water was initially ~11.17 at the screen underflow. The pH

decreased progressively until it reached a low point of about 10.3 at the flotation feed.

This caused an overabundance of dissolved magnesium possibly causing poor starch

selectivity, poor etheramine collector selectivity, and decreased zeta potential. The

combination of these three factors led to the high iron loss in the flotation tailings.

The conductivity of the process water was found to be directly proportional to the sodium

concentration in this sample set. This important finding will allow for online

132
determination of monovalent cations based on conductivity measurements from the new

online conductivity meter that became operational in early April, 2012.

This dataset also solidified theories made from the May 5th and July 13th datasets. The

following are the conclusions drawn from this sample set and from the project thus far:

1. Starch did noticeably and reproducibly suppress the surface charge of hematite in

both deslime and flotation

2. The effect of the starch was consistently more prominent in flotation

3. Starch did not significantly change the water hardness

4. The addition of CO2 had a significant effect on the pH, conductivity, water hardness,

and zeta potential. These additions resulted in a

4.1. Decrease in pH

4.2. Increase in conductivity

4.3. Increase in water hardness

4.4. Increase in zeta potential

2.3.5.4 Caustic Adjustment Study – April 2nd, 2012

2.3.5.4.1 Introduction

The goal of this work is to determine what water chemistry factors play a significant role

in the magnitude of the zeta potential and if there is an method for monitoring zeta

potential (or inferring it from other measurements) at any point within the plant. The

April 2nd, 2012 sample sets were taken to determine the effects of a pH change on the

water chemistry, zeta potential and iron recovery. The first sample set was gathered

133
between 8 AM and 12 PM on 4/2/2012 when the pH was 10.9 with a caustic rate of 1.48

lb/ton (crude). After this sample set was taken, the caustic rate was increased to 2.03

lb/ton and the process was given time to allow this change to take effect. The second

sample set was taken between 8 PM and 12 PM on 4/2/2012 when the pH had reached

11.0.

There were several observations on April 2nd that also had an effect on process:

1. Very low water hardness

2. Low glass-H rates compared to previous sample sets (0.04 – 0.07 lb/ton)

3. The plant pH meter was giving pH values ~ 0.4 pH units higher than the

calibrated laboratory pH meter

2.3.5.4.2 Process Data

The process data for sections 7-9 on April 2nd, 2012 is shown in Table 2.6. The process

had been running consistently well for several days prior to these sample sets. At the start

of the first sample set, the pH was at 10.9 with a fairly low caustic rate and Glass-H

(sodium henicosapolyphosphate) rate (1.48 and 0.04 lbs/ton, respectively). The weight %

recovery was average compared to other samples (39.0%) with a higher than average iron

recovery (72.3%). After the first sample set, the caustic rate was increased to 2.03 lbs/ton

causing a pH increase up to 11.0. This caused the deslime thickener to have a lower

weight rejection. The lower weight rejection was accompanied by a smaller amount of

iron lost in the deslime overflow (Fe distribution dropped from 11.2 down to 7.0 in the

DTO). Though there was less iron lost in deslime, the grade of the ore at the deslime

134
underflow was lower forcing more of the separation requirement onto flotation. Overall,

the increase in caustic rate led to less iron loss in deslime, more iron loss in flotation, and

a lower overall recovery.

Table 2.6: Process data for sections 7-9 on April 2nd, 2012

Date: 4/2/12 Time: 7:00 AM


Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.4 100 34.3 100.0 10.8 lbs/ton CYC O/F 34.9 100 Crude lbs/ton
DTU 48.1 63.3 30.4 88.8 pH Plant 0.04 39 72.3 DTU 48 63.4 1.48
DTO 10.5 36.7 3.9 11.2 10.9 DTO 12.1 36.6 Conc lbs/ton
Flot Tail 23.2 24.5 3.86
%Passing at 500 Mesh 83.5 Flot Conc 63.6 38.9
Date: 4/2/12 Time: 7:00 PM
Section 7-9 Tilden 2 Average
Total Recovery
%Fe %Wt. Fe Units Fe Dist. pH Lab Glass-H Rate Weight Iron Unit %Fe %Wt. Caustic Rate
CYC O/F 34.4 100 34.7 100.0 10.9 lbs/ton CYC O/F 34.6 100 Crude lbs/ton
DTU 45 69.3 32.3 93.0 pH Plant 0.07 37.4 69.2 DTU 45.4 70.1 2.03
DTO 10.5 30.7 2.4 7.0 11.0 DTO 9.2 29.9 Conc lbs/ton
Flot Tail 23.1 31.5 5.38
%Passing at 500 Mesh 82.8 Flot Conc 63.7 38.6

2.3.5.4.3 pH

The pH of each sample before and after caustic addition is shown in Figure 2.45. As

stated earlier, the pH readings taken in the lab were consistently lower than the pH given

by the online meter. The pH was consistently higher at every sample location after the

addition of caustic except for the concentrate thickener underflow sample. It was unlikely

that the change in caustic would have affected the concentrate thickener because of its

large residence time. The average increase in pH was 0.18 pH units with the most

significant changes seen at the screen underflow.

135
Figure 2.45: pH of each sample before and after caustic addition on April 2nd, 2012

2.3.5.4.4 Conductivity

The conductivity of each sample before and after caustic addition is shown in Figure

2.46. For each sample set, the same decreasing trend that was seen in previous samples

occurred throughout the process. At the lower pH, the conductivity started at 2932 μS/cm

and stabilized at around 2500 μS/cm. After caustic addition, the initial conductivity was

very high at 3692 μS/cm and diminished to around 2600 μS/cm. The addition of caustic

increased the overall conductivity by an average of 200 μS/cm with a larger difference

seen closer to the screen underflow.

136
Figure 2.46: Conductivity of each sample before and after caustic addition on April 2nd,

2012

2.3.5.4.5 Sodium and Potassium (monovalent cations)

The sodium concentration before and after the increase in caustic rate are shown in

Figure 2.47. Upon increase in caustic rate, the sodium levels immediately increased from

238 to 280 ppm. This was to be expected because caustic (NaOH) will dissociate into

sodium and hydroxide ions when dissolved. The sodium concentration did not remain

higher at the increased pH however. A possible explanation for this is that the increased

pH caused a decrease in the surface potential of the particles in solution causing an

137
increased attraction of cations to the particle surfaces. This would cause the sodium to

decrease as it did.

The overall sodium levels were ~750 ppm lower on April 2nd, 2012 than in the November

30th, 2011 sample set. This could be due to the change in the incoming water supply or

due to a limited use of caustic over the winter months. In these sample sets, the

conductivity related well to the sodium concentration as seen in the November 30th,

sample set; see Figure 2.48. The trend-line was shifted upwards after the addition of

caustic suggesting that pH has a large influence on the conductivity as well.

Figure 2.47: Sodium concentration of each sample before and after caustic addition on

April 2nd, 2012

138
Figure 2.48: Correlation between sodium and conductivity of the April 2nd, 2012 sample

sets

The potassium levels were measured in this sample set primarily to compare to a previous

water chemistry study in which samples were taken at these addition points and sent to a

certified laboratory for analysis (Haselhuhn et al., 2012a). The potassium concentrations

before and after the caustic addition rate increase are shown in Figure 2.49. Upon the

addition of caustic, the potassium levels decreased. This supports the theory proposed

above regarding the decrease in sodium levels due to surface chemistry.

In the previous study, the potassium ranged from 10 to 15 ppm until the concentrate

thickener, where the potassium concentration increased to 20 ppm (Haselhuhn et al.,


139
2012a). This same trend was found with both of the April 2nd sample sets; however, the

overall potassium levels were slightly lower.

Figure 2.49: Potassium concentration of each sample before and after caustic addition

on April 2nd, 2012

2.3.5.4.6 Calcium, Magnesium and Water Hardness

Common alkaline earth metals (Magnesium and Calcium) are the primary contributors to

water hardness. Control of Mg2+ and Ca2+ ions is essential to maintaining high iron

recoveries in deslime and flotation. Previous studies and ongoing research have shown a

strong link between the concentrations of calcium, magnesium, water hardness, zeta

140
potential and iron recovery. The calcium, magnesium, and total hardness concentrations

of the each location in the process are shown in Figure 2.50, 2.51, and 2.52.

On April 2nd, it was requested that the Glass-H rate remain constant throughout the day;

however, it was increased from 0.04 to 0.07 lbs/ton between 7 AM and 7 PM. The

increased caustic rate and Glass-H rate consistently lowered the calcium, magnesium and

total water hardness of the sample, but it cannot be inferred whether this loss in calcium

was due to the additional caustic or Glass-H.

The overall calcium concentration on April 2nd, 2012 was very similar throughout the

process compared to the November 30th, sample set; however, the magnesium

concentration on April 2nd was nearly twice that of November 30th. The total hardness did

not reflect the increased magnesium concentration suggesting that total hardness is not a

good indicator of magnesium concentration.

141
Figure 2.50: Calcium concentration of each sample before and after caustic addition on

April 2nd, 2012

142
Figure 2.51: Magnesium concentration of each sample before and after caustic addition

on April 2nd, 2012

143
Figure 2.52: Total Hardness of each sample before and after caustic addition on April

2nd, 2012

2.3.5.4.7 Soluble Iron

Soluble iron was measured to compare with trends found in a previous study performed

by Haselhuhn et al. (2012a). Soluble iron, in alkaline conditions, is mostly in the form of

Fe(OH)4- as shown in Figure 2.23 and is a contributor to water hardness when measured

by the EDTA titration method. The soluble iron concentrations of samples before and

after the increased caustic addition rate are shown in Figure 2.53. The increase in caustic

rate caused an increase in soluble iron concentration throughout the plant.

144
A previous study in October, 2010 performed by Haselhuhn et al. (2012a) showed a

progressive increase in soluble iron concentration ending at the flotation feed sample.

This trend was observed again in the April 2nd, 2012 samples except the end of the trend

occurred after the flotation feed sample. This difference was due to a change in sampling

location between studies. The October, 2010 flotation feed sample was taken from the

rougher feed box where etheramine was being added, whereas the April 2nd, 2012

flotation feed sample was taken from the output of the flotation feed distributor where

amine had not been added. This observation proved the hypothesis proposed during the

October 2010 study stating that the etheramine caused the reduction in soluble iron

concentration.

145
Figure 2.53: Soluble iron concentration of each sample before and after caustic addition

on April 2nd, 2012

2.3.5.4.8 Chloride and Nitrate (monovalent anions)

The chloride and nitrate concentrations before and after the increase in caustic addition

rate are shown in Figure 2.54 and Figure 2.55. The chloride concentration remained

between 220 and 320 ppm throughout the process and the change in caustic addition rate

seemed to have little effect on their respective concentrations. The deslime underflow

sample showed a significant decrease in chloride concentration for an unknown reason.

The nitrate concentration remained very low until the addition of recycled flux

concentrate and CO2. The nitrate concentration increased substantially when any flux was
146
added to the process making it apparent that there were small amounts nitrate salts within

the flux.

Figure 2.54: Chloride concentration of each sample before and after caustic addition on

April 2nd, 2012

147
Figure 2.55: Nitrate concentration of each sample before and after caustic addition on

April 2nd, 2012

2.3.5.4.9 Carbonates

The total carbonate concentration before and after the increase in caustic addition rate is

shown in Figure 2.56. The carbonate concentrations measured included both carbonate

(CO32-) and bicarbonate (HCO3-) forms. Carbonate concentration is an extremely

important factor because carbonate ions in solution tend to precipitate with metal cations

such as calcium and magnesium. As shown, the carbonate concentration remained fairly

stable until the addition of CO2. CO2 addition, as expected, increased the carbonate

148
concentration drastically. The change in caustic addition rate seemed to have very little

effect on the carbonate concentration.

Figure 2.56: Total carbonate concentration of each sample before and after caustic

addition on April 2nd, 2012

2.3.5.4.10 Sulfate

The sulfate concentration of each sample before and after the increase in caustic addition

rate is shown in Figure 2.57. The change in caustic addition rate seemed to have very

little effect on the sulfate concentration; however, the sulfate concentration seemed to

follow the same trend as soluble iron throughout the process; see Figure 2.58. Iron

sulfates are slightly soluble in water and could be could be a constituent of the incoming
149
ore in the form of natrojarosite mineral (NaFe3(OH)6(SO4)2) which has been found at the

Chicagon mine in Menominee, MI (Ralph et al., 2014). Slow dissolution of natrojarosite

would fully explain the phenomenon seen in both the soluble iron and the sulfate

concentrations.

Figure 2.57: Sulfate concentration of each sample before and after caustic addition on

April 2nd, 2012

150
Figure 2.58: Correlation between soluble iron and sulfate of the April 2nd, 2012 sample

sets

2.3.5.4.11 Total Dissolved Solids

The total dissolved solids (TDS) of each sample before and after the increase in caustic

addition rate is shown Figure 2.59. TDS is a measure of the total quantity (by mass) of

ions in a solution. As expected, the addition of caustic resulted in an overall increase in

TDS throughout the plant.

151
Figure 2.59: Total dissolved solids (TDS) of each sample before and after caustic

addition on April 2nd, 2012

2.3.5.4.12 Zeta Potential

The zeta potential at process conditions before and after the increase in caustic addition

rate is shown in Figure 2.60. The first noticeable trend was the increase in zeta potential

with the addition of Glass-H (sodium henicosapolyphosphate dispersant). This trend was

seen in all sample sets except for the November 30th, 2011 sample set. The average

increase in zeta potential was 6 mV.

Suppression of the surface charge was not seen at the cyclone overflow at the lower

caustic addition rate; however, it was seen at the flotation feed. The increase in caustic
152
addition rate caused an increase in surface charge suppression by starch at the cyclone

overflow but did not affect charge suppression at the flotation feed. Zeta potential curves

of these samples are included in the Appendix A.

All sample sets have shown an increase by 5 – 14 mV in zeta potential upon the addition

of CO2. The addition of flux decreased the zeta potential consistently to ~20 mV. This

could cause negative effects on filtration performance.

Figure 2.60: Zeta potential of each sample before and after caustic addition on April 2nd,

2012

2.3.5.4.13 Elemental Analysis

153
The concentration of calcium, magnesium, iron (Fe2+ and Fe3+), and acid insoluble

material was measured by acid leaching followed by atomic absorption spectroscopy and

gravimetry; see Figure 2.61, 2.62, 2.63 and 2.64. The magnesium content of the solids

was consistently higher than the calcium content. This was consistent with the increased

dissolved magnesium concentration in the water. Upon flux addition, the concentration of

calcium rose more than magnesium suggesting the use of a calcite-rich flux. As with

previous sample sets, calcite/dolomite rejection occurred during deslime but not during

flotation.

Figure 2.61: Calcium concentration in solids of each sample before and after caustic

addition on April 2nd, 2012

154
Figure 2.62: Magnesium concentration in solids of each sample before and after caustic

addition on April 2nd, 2012

The iron content and acid insoluble material content of the slurry solids are shown in

Figure 2.63 and 2.64. As expected, there was a large acid insolubles rejection during

deslime and flotation where silica and other gangue minerals were removed. The iron was

concentrated by both the deslime and flotation. The inconsistency between the deslime

underflow and the flotation feed was caused by the merging of section10-12 with section

7-9 at that point.

155
Figure 2.63: Iron concentration in solids of each sample before and after caustic

addition on April 2nd, 2012

156
Figure 2.64: Acid insoluble concentration in solids of each sample before and after

caustic addition on April 2nd, 2012

2.3.5.4.14 Conclusions

These sample sets provided a substantial amount of additional correlations between water

chemistry, zeta potential and process performance. Overall, the increase in caustic rate

led to less iron loss in deslime, more iron loss in flotation, an increase in soluble iron

concentration, and a lower overall recovery. The overall loss in recovery was primarily

due to the increased load on the flotation circuit caused by a lower total weight rejection

in deslime.

157
A previous study hypothesized that etheramine caused a reduction in soluble iron

concentration when added during flotation. This theory was solidified with this dataset

and also showed a correlation between soluble iron and sulfate concentration. The

association between iron and sulfate suggests the presence of natrojarosite minerals

(NaFe3(OH)6(SO4)2) in the ore. This finding was not verified by XRD.

This dataset also solidifies theories made from all previous datasets. The following are

the conclusions drawn from all sample sets:

1. Starch does noticeably and reproducibly suppress the surface charge of hematite in

both deslime and flotation

2. The effect of the starch is consistently more prominent in flotation

3. Starch does not significantly change the water hardness

4. The addition of CO2 has a significant effect on the pH, conductivity, water hardness,

and zeta potential. These additions result in a

4.1. Decrease in pH

4.2. Increase in conductivity

4.3. Increase in water hardness

4.4. Increase in zeta potential

5. Conductivity is directly proportional to the sodium concentration of the process water

6. Flux addition causes the zeta potential to equilibrate at ~-20 mV

7. Calcium and magnesium are rejected by the deslime thickener but not during flotation

158
2.3.5.5 Plant-wide Overall Trends Observed

2.3.5.5.1 Glass-H Addition

Glass-H addition caused a significant decrease in the magnitude of zeta potential (5-14

mV) in all sample sets except for the November 30th, 2011 sample set as shown in Figure

2.65. Glass-H ((NaPO4)21*Na2O, sodium henicosapolyphosphate) is a surface modifying

reagent that is typically used for its dispersive qualities. Glass-H requires positively

charged surface sites (un-hydrated iron or covalently bonded calcium/magnesium) for a

chemical bond to form causing dispersion.

Figure 2.65: Effects of Glass-H on Zeta Potential

159
2.3.5.5.2 Cyclone Overflow Starch Addition

Starch caused a significant decrease in the magnitude of zeta potential in all sample sets

except for the April 2nd, 2012 set at low caustic rate as shown in Figure 2.66. Starch has

been shown in literature to be a selective flocculent for hematite (Balajee et al., 1969).

Starch adsorption occurs at the surface of hematite due structural compatibility The

average distance between iron atoms at the surface of un-hydrated hematite particles is

2.852 A° (angstroms (A°) = 10-10 m), whereas the spacing between oxygen atoms at the

end of a starch chain is 2.85 A°. According to literature, a bond forms between these

oxygen atoms in starch and un-hydrated Fe surface sites (Pradip, 1994).

Figure 2.66: Cyclone overflow starch effects on zeta potential

The starch coats the hematite particles causing the magnitude of the surface charge to

decrease, thereby lowering the magnitude of the zeta potential. The amount that the
160
magnitude of zeta potential decreases is relative to the amount of starch adsorbed to the

hematite particle surfaces. As shown in Figure 2.67, the decrease in the magnitude of zeta

potential due to starch adsorption seemed to be dependent on the water hardness in the

sample. This suggested a certain degree of water hardness is required for starch

adsorption, and calcium/magnesium ions acted as an activator for starch adsorption on

hematite surfaces. However, it is common knowledge that too much calcium can cause a

loss in selectivity of the starch.

Figure 2.67: Starch effectiveness as a function of calcium hardness at cyclone overflow

2.3.5.5.3 Deslime Thickener

An important finding of this project was the correlation of zeta potential to the iron

concentration in the deslime overflow sample. Silica tended to have a much higher

magnitude of zeta potential (~-100 mV) in alkaline conditions than hematite (~15 mV),
161
therefore, the zeta potential of a hematite/silica mixture should be dependent upon the

amount of hematite in the mixture. As shown in Figure 2.68, this was indeed the case; as

the hematite concentration increased, the zeta potential decreased. Variations in this

behavior can be attributed to changes in water chemistry.

Figure 2.68: Dependence of iron concentration on zeta potential in deslime overflow

samples

The zeta potential at the deslime feed had a large effect on the iron recovery of the

deslime thickener as seen in Figure 2.69. Zeta potential is a direct indicator of dispersive

stability; a higher magnitude of zeta potential indicates the slurry is highly dispersed; a

low magnitude of zeta potential indicates that the slurry particles can non-selectively

flocculate. In the deslime thickener, a highly dispersed slurry is required to maximize the

separation of iron oxides from gangue minerals.


162
Figure 2.69: Correlation between zeta potential at the deslime feed and the iron recovery

within the deslime thickener.

Water hardness played a vital role in the effectiveness of the deslime thickener. There

were two ways in which the calcium/magnesium concentration played a major role in the

deslime thickener:

1. Calcium and magnesium activated iron oxide surfaces for starch adsorption

2. Calcium and magnesium caused flocculation of fine particles

Low water hardness at the deslime thickener feed caused higher dispersion and increased

selectivity of starch adsorption; see Figure 2.70. This scenario yielded the highest iron

recovery. Higher water hardness both decreased the selectivity of starch and lowered

dispersion. Overall, a higher water hardness yielded a lower deslime recovery.

163
Figure 2.70: Correlations between calcium hardness and process performance. Both

11/30/11 and 5/5/11 samples were not included because they were taken during a process

upset.

Conductivity is a significant factor in water chemistry because it relates to the total ionic

content of the slurry. Each type of ion (OH-, Na+, Cl-, Ca2+, etc.) contributes to the

conductivity based on its concentration and a constant specific to that ion called the

limiting ion conductivity. Although conductivity is not related directly to zeta potential, it

can be assumed that a higher conductivity in alkaline conditions will result in a decrease

in the magnitude of zeta potential.

Conductivity can be used as a predictor of process performance because of its

relationship to the ionic content of the process water. A higher conductivity is associated

with a large abundance of ions in the water. These ions are electrostatically attracted to
164
the particles and cause a phenomenon called ‘double layer compression’. This

phenomenon reduces the net charge of the particles resulting in non-selective

flocculation. Non-selective flocculation, as seen with increased water hardness, causes a

lower iron recovery in the deslime thickener; see Figure 2.71. In contrast, a lower

conductivity is associated with increased repulsive charges of particles, thereby

increasing the dispersive stability and iron recovery.

Figure 2.71: Correlation between deslime iron recovery and conductivity of deslime feed.

Both 11/30/11 and 5/5/11 samples were not included because they were taken during a

process upset.

2.3.5.5.4 Flotation Feed Starch Addition

The flotation feed showed a significant decrease in the magnitude of zeta potential

(increase in zeta potential) upon starch addition as shown in Figure 2.72. This signified
165
selectivity of starch towards iron oxides because of the increased iron oxide content at

this point as opposed to the cyclone overflow. The large increases in zeta potential seen

on 5/5/2011 and 7/13/2011 were both accompanied by higher water hardness values. As

with the cyclone overflow samples, water hardness had a significant effect on starch

adsorption, as shown in Figure 2.73. By comparing the slope of the linear trend-line at the

cyclone overflow in Figure 2.67 to the slope of the linear trend-line at the flotation feed

in Figure 2.73, starch adsorption at the flotation feed was three times as sensitive to water

hardness.

Figure 2.72: Flotation feed starch effects on zeta potential

166
Figure 2.73: Starch effectiveness as a function of calcium hardness at flotation feed

2.3.5.5.5 Flotation Circuit

The zeta potential throughout flotation is extremely important because it affects both

starch and etheramine adsorption. In the alkaline environment, both silica and iron oxides

carry a negative surface charge. The surface charge of the iron oxides must be suppressed

to allow for selective adsorption of the cationic collector (etheramine) onto silica

surfaces. The flotation feed sample typically had a lower magnitude of zeta potential due

to the suppression of the surface charge of iron oxides by starch, as shown in Figure 2.74.

The silica retained its high magnitude zeta potential, and was floated to the tailings. The

rougher concentrate, primarily hematite, retained a low magnitude zeta potential.

167
Figure 2.74: Zeta potential throughout flotation cells during each sampling campaign

The zeta potential of the flotation feed has a large effect on the concentration of iron in

the flotation tailings as shown in Figure 2.75. Lower magnitude zeta potentials at the

flotation feed were caused by more complete coverage by starch, allowing for higher

selectivity in etheramine adsorption and therefore increased iron recovery. Higher

magnitude zeta potentials indicated a high level of dispersion and incomplete coverage by

168
starch. This reduced etheramine selectivity and ultimately caused a decrease in iron

recovery during flotation. This trend is opposite to that seen in the deslime thickener.

Figure 2.75: Effect of zeta potential of flotation feed on iron recovery. Both 11/30/11

and 5/5/11 samples were not included because they were taken during a process upset.

As stated previously, starch adsorption was affected by water hardness. Water hardness

was therefore directly related to the iron recovery during flotation, as shown in Figure

2.76. Prior to flotation, the iron oxide particles have completely hydrated, yielding no

positive surface sites for starch adsorption that aren’t created by adsorbed

calcium/magnesium. When the water hardness is low, the fully hydrated hematite could

be coated in starch. This limited etheramine selectivity and ultimately lowered iron

recovery. When water hardness is high (>10 ppm), positive surface sites formed on the

169
iron oxide surface allowing for starch adsorption. The calcium and magnesium ions can

also form positive surface sites on gangue mineral particles and reduce starch selectivity.

This indicates that there was an ideal water hardness to yield the highest flotation

recovery. Due to lack of data at water hardness values above 8 ppm, the only conclusion

that can be made is that the optimum water hardness was above 8 ppm.

Figure 2.76: Effect of calcium hardness in flotation on the iron recovery. Both 11/30/11

and 5/5/11 samples were not included because they were taken during a process upset.

The conductivity during flotation, through its relationship to zeta potential and the

surface chemistry of the particles, was a good predictor of flotation performance. Higher

conductivity was associated with lower dispersion and a higher iron recovery during

flotation as shown in Figure 2.77.

170
Figure 2.77: Effect of conductivity in flotation on the iron recovery. Both 11/30/11 and

5/5/11 samples were not included because they were taken during a process upset.

The effectiveness of the deslime thickener had a large bearing on how much iron was lost

during flotation. As shown in Figure 2.78, if the total weight rejected by the deslime

thickener was minimal, flotation rejected more iron. Although calcium hardness and zeta

potential did show an effect on the iron recovery, both of those parameters also affected

deslime performance in the opposite manner, thereby cancelling each other out. This

suggests that flotation was primarily capacity limited and depended on the desliming

process to be effective. Focusing the study on deslime thickening provides the most

potential for overall increases in iron recovery.

171
Figure 2.78: The effect of deslime weight rejection on iron loss during flotation. 11/30/11

sample excluded due to overgrind

Ultimately, from these samples, it was shown that the separation processes required

different processing conditions to be optimized; see Figure 2.79. Water hardness was

critical to effective separation in both deslime and flotation. At low water hardness

values, iron recovery was high in deslime and low in flotation. At high water hardness

values, iron recovery was low in deslime and high in flotation. From the data gathered,

there seemed to be little variability in the overall recovery, but these data sets were

missing processing conditions in which the water hardness was in the range of 6 and 18

ppm at the deslime feed.

172
Figure 2.79: The effect of water hardness on deslime, flotation and overall recovery.

Both 11/30/11 and 5/5/11 samples were not included because they were taken during a

process upset.

2.3.5.5.6 Dewatering

During dewatering, there were many water chemistry changes that accompanied the

addition of flux (calcite/dolomite), CO2, and dewatering aids/polymers. The addition of

CO2 to decrease the pH caused changes in the equilibrium distribution of carbonate ions

as shown in Figure 2.80. Carbonate ions (CO32-) form bicarbonate ions (HCO3-) at the

lower pH. Some of the sodium carbonate (NaCO3-) ions can also hydrolyze to form

aqueous sodium bicarbonate (NaHCO3).

173
Figure 2.80: Carbonate equilibrium distribution changes during dewatering.

Note: only 1st 4/2/2012 sample shown.

The decrease in pH caused some of the flux and possibly some of the calcium and

magnesium minerals present in the ore to dissolve into calcium and magnesium ions, as

shown in Figure 2.81 and Figure 2.82. At the higher pH of the rougher concentrate, most

of the aqueous calcium and magnesium was present in a carbonate form (CaCO3,

MgCO3). After the pH was lowered with CO2, the presence of aqueous carbonated

calcium and magnesium was diminished. This was accompanied by a rise in free

calcium/magnesium ions (Ca2+, Mg2+) and calcium/magnesium bicarbonate ions

174
(CaHCO3-, MgHCO3-). There was little change in the concentration of aqueous metal

sulfate ions (CaSO4, MgSO4).

Figure 2.81: Calcium equilibrium distribution changes during dewatering.

Note: only 1st 4/2/2012 sample shown.

175
Figure 2.82: Magnesium equilibrium distribution changes during dewatering.

Note: only 1st 4/2/2012 sample shown.

These water chemistry changes had a significant impact on zeta potential as shown in

Figure 2.83. The introduction of CO2 between the rougher concentrate and concentrate

thickener underflow decreased the magnitude of zeta potential for two reasons: it caused

higher concentrations of aqueous divalent cations which formed inner-sphere complexed

at hematite surface thereby increasing surface charge, and it caused a decrease in pH

causing hematite to approach its isoelectric point. The addition of flux at the filter feed

established a shifted equilibrium between calcium, magnesium and carbonate. This

176
shifted equilibrium was also associated with an increase in pH. The increase in pH caused

the zeta potential to decrease slightly to ~-20mV.

Figure 2.83: Zeta potential of sample during dewatering steps

2.3.5.6 Conclusions

During this phase of this project, samples were collected 5 times at 12 locations

throughout section 7-9 of the hematite ore concentrator. Sample locations were

strategically chosen to determine zeta potential effects at reagent addition points and zeta

potential effects on separation process performance. Samples were tested for water

quality parameters including pH, conductivity, water hardness, calcium, magnesium,

iron, sodium, potassium, sulfate, carbonate, nitrate and oxidation-reduction potential. The
177
solid content of each sample was analyzed for zeta potential (at process conditions),

elemental composition, thermogravimetric degradation, x-ray diffraction, x-ray

fluorescence and particle size distribution.

The following conclusions have been drawn from this study thus far:

1. The zeta potential of the slurry was largely controlled by 6 main factors: pH, water

hardness, conductivity, mineralogy, Glass-H adsorption and starch adsorption.

1.1. Increases in pH tended to cause a more negative zeta potential.

1.2. Higher water hardness caused inner-sphere complexation at all oxide surfaces

which yielded a less negative zeta potential.

1.3. Higher conductivity, due to metal cations in solution, caused both inner-sphere

and outer-sphere complexation with metal cations which created a phenomenon

called double-layer compression. Double layer compression caused a less

negative zeta potential.

1.4. The mineralogy, silica in particular, had a drastic effect on zeta potential.

Different minerals have different zeta potentials at any given pH. Silica has a

very negative zeta potential. A higher concentration of silica caused the overall

zeta potential of the slurry to be more negative.

1.5. Starch selectively coated hematite due to structural compatibility (Pradip, 1994).

Modified corn starch created a non-ionic coating around hematite thereby

suppressing its surface charge. This caused the zeta potential to become more

positive.

178
2. Conductivity was a function of many water chemistry parameters including pH,

sodium, potassium, chloride, calcium, and magnesium.

2.1. On most days, sodium levels throughout the plant varied drastically compared to

other ions due to the addition of caustic (NaOH) and glass-H ((NaPO3)21*Na2O).

This caused the conductivity to be nearly directly proportional to the sodium

concentration in the process water.

3. Calcium and magnesium present as minerals in the ore were rejected by the deslime

thickener but not by the froth flotation cells.

4. pH had a significant effect on both zeta potential and deslime performance.

4.1. Higher pH caused the surface charge to become more negative. It also caused

significant changes in ion types and concentrations in the water.

For instance, a high pH (>11) caused much of the calcium to be present as

aqueous CaCO3 but caused magnesium to be present as MgOH+. The magnesium

hydroxide caused unselective flocculation. At a low pH (~10.6), calcium was

present as both Ca2+ and CaCO3 whereas magnesium was present primarily as

aqueous MgCO3. The divalent calcium caused unselective flocculation as well.

Depending on the concentrations of magnesium and calcium, there was an ideal

pH to maximize overall dispersion.

4.2. pH had a direct influence on the amount of iron in the deslime underflow. An

ideal pH that maximizes overall dispersion will maximize the iron in the

underflow while minimizing the iron in the overflow.

179
4.3. The optimum pH for most calcium and magnesium conditions was near 11. This

pH was shown to yield the highest iron recoveries.

4.4. pH had a significant effect on conductivity. As stated previously, pH affected the

types of ions present in the water.

5. Particle size control was important for influencing deslime weight rejection. Finer

particles tended to be more liberated allowing for easier separation; however, fine

particles without adequate flocculation were easily lost in the deslime overflow. This

was witnessed on 11/30/2011 when the crushers were shutdown causing

overgrinding.

6. Water hardness at the cyclone overflow had a large influence on deslime

performance. Higher water hardness caused non-selective flocculation of particles

which lead to:

6.1. Lower deslime recovery

6.2. Lower magnitude zeta potential

7. Starch did noticeably and reproducibly suppress the surface charge of hematite during

both deslime and flotation. This was seen as a decrease in the magnitude of zeta

potential upon starch addition. The degree to which starch suppressed zeta potential

was dependent upon the water hardness at the time of starch addition. The effect of

the starch was consistently more prominent in flotation than desliming due to a higher

iron concentration in particles.

8. The addition of CO2 during dewatering had a significant effect on the pH,

conductivity, water hardness, and zeta potential. CO2 addition causes the initial

180
formation of carbonic acid (H2CO3) which then dissociates to carbonate (CO32-) and

bicarbonate (HCO3-) ions, lowering the pH. The increased ionic content caused an

increase in conductivity. The decrease in pH caused calcite and dolomite in the ore to

dissolve which yielded an increased water hardness. Both the decrease in pH and the

increase in water hardness caused the zeta potential to become less negative.

9. Flux addition caused the zeta potential to equilibrate at ~-20 mV.

10. Maximized deslime iron grade was associated with

10.1. Lower water hardness

10.2. Lower conductivity

10.3. Higher magnitude of deslime feed zeta potential

11. Maximized flotation iron grade was associated with

11.1. Higher water hardness (>8 ppm)

11.2. Higher conductivity

11.3. Lower magnitude of flotation feed zeta potential

12. Flotation iron recovery was largely dependent on deslime iron recovery; therefore,

flotation seemed to be capacity limited. An ore of lower grade entering flotation

caused higher iron losses; therefore, separation at the deslime thickener is critical to

preventing excessive iron loss during flotation.

Based on these results, it has been found that monitoring the water chemistry and zeta

potential at the deslime thickener offers the most potential for increasing iron recovery. It

was theorized that pH and water hardness at the deslime thickener can be optimized to

yield a higher overall recovery. Another way that could potentially increase overall iron

181
recovery is increasing the water hardness in the flotation feed distributors by adding

slaked lime.

2.3.6 Extended Deslime Thickener Analysis

2.3.6.1 Introduction

The major separation processes in the beneficiation plant (desliming, froth flotation,

thickening, and filtration) are all dependent on the interactions of particles, which are

largely controlled by the Zeta potential of the particles. Measurement and control of the

Zeta potential is therefore key to controlling the plant operations and preventing process

upsets. Zeta potential can be controlled by adjusting the water quality, adding flocculants

or dispersants, or by adjusting the incoming feed ore.

The objectives of this work were to measure process water quality and zeta potential

onsite at the mine to determine what process parameters were important for optimizing

the iron grade and recovery during deslime thickening. Samples were taken at the feed,

overflow, and underflow of the section 7-9 deslime thickeners over the course of one

week (5 days). It was important to analyze these samples as soon as possible because it

was hypothesized in earlier in the project that the surfaces change over time.

Samples were collected 3 times per day at 3 locations surrounding the deslime thickeners

of section 7-9. Samples were tested for water quality parameters including pH,

conductivity, calcium, magnesium, sodium, potassium, and oxidation-reduction potential.

The solid content of each sample was analyzed for zeta potential (at process conditions),

x-ray diffraction, and x-ray fluorescence.

182
2.3.6.2 Materials and Methods

During the study, process samples were taken and analyzed both onsite at the mine and

offsite at Michigan Technological University and an external laboratory. The analyses

were used to determine what process parameters were important for optimizing the

deslime thickener operation.

2.3.6.2.1 Plant Sampling

Slurry samples were taken on sections 7-9 at 7 AM, 1 PM, and 7 PM between March 17th

and March 21st, 2014. The samples were taken at 3 locations: deslime feed (Figure 2.26

location 5), deslime overflow (Figure 2.26 location 6), and deslime underflow (Figure

2.26 location 7). Each sample was obtained in a 500 mL plastic bottle and transported to

the analysis laboratory at the mine. The samples were analyzed for zeta potential

immediately after collection using a Malvern Zetasizer Nano ZS. The pH, conductivity,

calcium hardness ISE value, sodium ISE value, chloride ISE value, and ORP were also

measured onsite.

After onsite measurement, the samples were centrifuged and filtered to separate the liquid

and solid fractions. The liquid was analyzed for calcium, magnesium, potassium and

sodium concentration using inductively coupled plasma optical emission spectroscopy

(ICP-OES). The solids were dried and sent an external laboratory for X-Ray diffraction

and X-ray fluorescence spectroscopy. Plant process data was collected by mine

employees for analysis.

183
2.3.6.3 Results and Discussion

Due to the enormous amount of data generated during this week-long study, it was

important to decide what the most important factors were. As with any mineral

processing plant, the product grade and the mineralogical recovery are typically the most

important pieces of information. Determining what process parameters have significant

influence on product grade and recovery is of utmost importance. This data analysis

begins with all of the iron grade and recovery data, see Figure 2.84, and seeks to

determine the cause of fluctuations in the grade versus recovery curve.

184
Figure 2.84: Grade versus recovery curve for the deslime thickener during the week of

March 17th through March 21st, 2014. Ellipses represent groups of data falling in

different operational ranges: (A) Low Fe grade and high Fe recovery, (B) Average Fe

grade and recovery, (C) High Fe grade and low Fe recovery.

Strictly targeting either a higher Fe grade or a higher Fe recovery in the deslime thickener

is not as important as determining what range of grades and recoveries are conducive to a

higher Fe recovery in the entire plant. For example, compare a point in the ellipse labeled

(A) in Figure 2.84 and a point in the ellipse labeled (C) in Figure 2.84. Running the

deslime thickener in ellipse A will recover more iron at the expense of grade and cause

higher iron losses in the flotation circuit, as shown in phase 1. On the other hand, running
185
the deslime thickener in ellipse C will lose more iron during deslime but less iron will be

lost in the flotation circuit. The question is: where is the point on the deslime grade

versus recovery curve that is good for the entire plant?

With a continuously changing feed, it may be a better metric to look at an enrichment

ratio versus recovery plot to eliminate the effects of changing feed conditions.

Enrichment ratio is calculated as follows:

Ψ‫݁ݐܽݎݐ݊݁ܿ݊݋ܥ ݊݅ ݁ܨ‬
݁݊‫ ݋݅ݐܽݎݐ݄݊݁݉ܿ݅ݎ‬ൌ Equation 2.32
Ψ‫݀݁݁ܨ ݊݅ ݁ܨ‬

Obviously, if you start with a material with a higher Fe content, you expect to see a

higher Fe content in the concentrate. The enrichment ratio versus recovery, as shown in

Figure 2.85, essentially shows, essentially, the separation versus the recovery. Just like

with the grade versus recovery, higher enrichment ratios are typically accompanied by

lower recoveries. However; for these data, it shows that the high grade versus low

recovery (ellipse C) was really a combination of high feed grade with low separation

efficiency. This can be seen by comparing the horizontal translations of the data points in

ellipse C in Figure 2.86. The data points tend to translate left showing a lower than

expected enrichment ratio for that grade. The low grade versus high recovery data (ellipse

A), when comparing it by enrichment ratio, shows the same trend. This implies that these

outliers (points in ellipse A and C) were due to low separation efficiency.

186
Figure 2.85: Enrichment ratio versus recovery curve for the deslime thickener during

the week of March 17th through March 21st, 2014. Ellipses represent groups of data

falling in different operational ranges: (A) Low Fe enrichment ratio and high Fe

recovery, (B) Average Fe enrichment ratio and recovery, (C) High Fe enrichment ratio

and low Fe recovery.

187
Figure 2.86: Enrichment ratio and grade versus recovery curve for the deslime thickener

during the week of March 17th through March 21st, 2014. Ellipses represent groups of

data falling in different operational ranges: (A) Low Fe enrichment ratio and grade with

high Fe recovery, (B) Average Fe enrichment ratio, grade, and recovery, (C) High Fe

enrichment ratio and grade with low Fe recovery. Horizontal arrows indicate instances

during which the grade was superficially high compared to the enrichment ratio due to

high Fe grade in the feed.

188
Going back to the proposed fundamental question; the data points in ellipse B have very

little translation when comparing concentrate grade and enrichment ratio. This means that

the operational parameters seen during these times are the most conducive to higher

overall plant recoveries. The next question is: what are these operational parameters and

what needs to be done to maintain them? To answer that question, it is necessary to look

at the differences between operations in ellipses A, B and C.

2.3.6.3.1 Ideal Process Parameters Targeting Recovery

As seen in phase one of this project, there are many factors that can influence Fe recovery

in the deslime thickener. The key parameters from a water and surface chemistry

standpoint are the zeta potential, pH, water hardness (Ca +Mg), and the conductivity.

These parameters are summarized in Table 2.7.

189
Table 2.7: Operational parameters found to be important during phase 2 of the project

relating to the iron recovery during deslime thickening. The operational parameter

values were averaged for each recovery range.

Deslime Feed Conditions

Zeta Underflow
Deslime Fe Ca Mg Na Conductivity
Potential pH % Solids
Recovery (ppm) (ppm) (ppm) (mS/cm)
(mV)

High

(>93%, ellipse A in -37.3 10.88 5.0 1.8 567.6 2.69 54.94

Figure 2.84)

Mid

(86-93%, ellipse B -36.2 10.95 3.2 1.6 559.1 2.68 54.70

in Figure 2.84)

Low

(<86%, ellipse C in -36.9 10.88 3.6 2.1 562.8 2.68 52.99

Figure 2.84)

It was found that lower values of zeta potential (magnitude) typically caused

destabilization of the slurry. On a theoretical basis, this should result in a higher recovery

with a less selective separation. The zeta potentials reported in Table 2.7 are the feed zeta

potentials after starch addition. From the conclusions in phase 1, low zeta potential

magnitudes occurred because calcium adsorbed to all surfaces causing non-selective

flocculation when starch was added. This can be counteracted by higher dispersant

190
addition. From the data shown in Table 2.7, the lowest magnitudes of zeta potential

resulted in the points seen in ellipse B in Figure 2.84. From a recovery standpoint, having

a low feed zeta potential is conducive to better recoveries without a loss in enrichment

ratio.

The pH is a very important parameter for three reasons: (1) high pH results in a higher

dispersive stability. This results in a higher level of dispersion between gangue minerals

and iron oxides causing a higher chance that iron oxides will flocculate selectively. (2) It

has been reported that starch is more effective in a particular pH range. Results from a

study performed by Weissenborn (1996) are shown in Figure 2.87 depicting this range.

This study reported that as pH was increased above 10, the recovery declined; however,

the grade increased. And (3), as depicted in Figure 2.24 a and b, the pH range between

10.8 and 11.1 was ideal to allow free calcium and magnesium ions to complex with

carbonate and sulfate ions which limits their surface interactions. To recap, high pH was

good for dispersive stability, a pH higher than 10 limited starch activity, and a pH

between 10.8 and 11.1 was ideal to limit calcium and magnesium ionic effects on

surfaces. According to the data presented in Table 2.7, higher pH values put the process

in the mid-range for recovery at an acceptable grade (ellipse B in Figure 2.84).

191
Figure 2.87: Iron grade and recovery during selective flocculation and dispersion of an

ultrafine (9-% passing at 13.8μm) iron ore. Feed hematite grade was 53.2%. Starch with

a 74% amylopectin and 26% amylose content was used at 50 mg/L as a selective

flocculant. Adapted from Weissenborn (1996).

Another important parameter is the water hardness. This is typically a function of the

calcium and magnesium concentration within the process water. Water hardness

contributes to non-selective flocculation. This is good for iron recovery but detrimental to

the separation. As seen in Table 2.7, the higher recoveries seen in ellipse A in Figure 2.84

were associated with high calcium concentration. The recoveries in the 91% (Figure 2.84,

ellipse B) range were associated with the lowest calcium and magnesium concentrations.

The recoveries in the 83% (Figure 2.84, ellipse C) range had moderately low calcium and

192
magnesium concentrations but other process parameters were found to be the cause of

these low recoveries.

Finally, the last two parameters presented in Table 2.7 are sodium concentration and

conductivity. In phase 1, a strong correlation was found between sodium concentration

and conductivity. The sodium concentration in this plant varies drastically due to the use

of caustic for pH adjustment. Sodium causes double-layer compression at the surfaces of

the mineral particles. This can diminish zeta potential (as seen in Figure 2.21) and result

in less effective reagent adsorption. At a pilot scale (discussed later) it was found that

sodium concentration was a key parameter for increasing both the grade and recovery of

the process. Lower sodium concentrations correlated to better grades and recoveries. This

is also shown in Table 2.7 for the plant data. The values of sodium in ellipse B of Figure

2.84 were lower than those in ellipse A and C.

One of the key process parameters that was found during this study was the deslime

underflow density. The deslime underflow density averaged around 55% solids except for

the Figure 2.84 ellipse C data. These points of low recovery were associated with low

underflow densities at around 50% solids. This could have been due to low solids

concentration in the feed, too high of an underflow flow rate or over-dispersion of the

pulp. In any case, the underflow density was the biggest outlier in the data set at these

times of low recovery.

193
2.3.6.3.2 Ideal Process Parameters Targeting Grade

Targeting a particular grade is another way to evaluate process performance. Table 2.8

shows the same process parameters for high (>49% Fe), low (<47% Fe) and mid-level

deslime underflow grades. Just as with recovery, a higher grade is not necessarily a good

thing because it is typically accompanied by a lower recovery. The best plant

performances were found when the grade was in the mid-range because these times also

had mid-range recoveries. By comparing Table 2.8 to Table 2.7, it is apparent that the

best operational parameters to achieve these mid-range grade values are similar to those

when pursuing the mid-range recoveries. These include:

• Low magnitude of zeta potential

• High pH

• Low calcium

• Low magnesium

• Low sodium

• Low conductivity

194
Table 2.8: Operational parameters found to be important during phase 1 of the project

relating to the deslime underflow iron grade. The operational parameter values were

averaged for each grade range.

Deslime Feed Conditions

Deslime Zeta
Ca Mg Na Conductivity
Underflow Fe Potential pH
(ppm) (ppm) (ppm) (mS/cm)
Grade (mV)

High (>49% Fe) -36.9 10.88 3.6 2.1 562.8 2.68

Mid (47-49% Fe) -36.0 10.96 3.2 1.6 558.0 2.68

Low (<47% Fe) -38.0 10.86 5.1 1.9 570.8 2.70

2.3.6.3.3 Ideal Process Parameters Targeting Enrichment Ratio

Targeting a high enrichment ratio is a better way to evaluate process performance than

targeting grade, at least in the deslime thickener. Enrichment ratio allows the removal of

feed ore inconsistencies from the evaluation, and, since flotation is used to reach the final

grade, having the highest grade at the deslime underflow is not as critical as reducing iron

loss. Table 2.9 shows the same process parameters for high (>1.4), low (<1.3) and mid-

level deslime enrichment ratios. Just as with grade and recovery, a higher enrichment

ratio is not necessarily a good thing because it is typically accompanied by a lower

recovery. The best plant performances were found when the enrichment ratio was in the

mid-range because these times also had mid-range recoveries. By comparing Table 2.9 to

Table 2.7 and Table 2.8 it is apparent that the best operational parameters to achieve

these mid-range enrichment ratios are similar to those when pursuing the mid-range

recoveries. An exception to this is the zeta potential. When targeting enrichment ratio, the
195
zeta potential did not seem to have a large effect. The targets for the process parameters

outlined in Table 2.9 include:

• High pH

• Low calcium

• Low magnesium

• Low sodium

• Low conductivity

Table 2.9: Operational parameters found to be important during phase 1 of the project

relating to the iron enrichment ratio. The operational parameter values were averaged

for each enrichment ratio range.

Deslime Feed Conditions

Zeta
Deslime Ca Mg Na Conductivity
Potential pH
Enrichment Ratio (ppm) (ppm) (ppm) (mS/cm)
(mV)

High (>1.4) -36.1 10.91 3.7 2.1 574.7 2.73

Mid (1.3-1.4) -36.5 10.95 3.2 1.6 555.2 2.66

Low (<1.3) -37.3 10.88 5.0 1.8 567.6 2.69

2.3.6.4 Conclusions

The objectives of this work were to measure process water quality and zeta potential

onsite at the mine to determine what process parameters were important for optimizing

the iron grade and recovery during deslime thickening. It was determined that the best

196
conditions for optimizing both iron grade and recovery during the deslime thickening

process include:

1. Low magnitude of zeta potential

2. High pH

3. Low calcium concentration in process water

4. Low magnesium concentration in process water

5. Low sodium concentration in process water

6. Low conductivity in process water

These conditions, when present in the plant, were effective at increasing the grade,

recovery and enrichment ratio of iron in the process. To verify these findings, a pilot

scale study was performed that assessed the effects of calcium concentration, magnesium

concentration, sodium concentration and pH on the zeta potential, iron grade obtainable

and iron recovery during deslime thickening.

Finely ground hematite ore was suspended in a solution of simulated process water at

varying pH, sodium concentration, calcium concentration and magnesium concentration.

Cooked acid modified corn starch was used as a selective hematite flocculant and a

Glass-H was used as a dispersant. The following conclusions were drawn from this study:

1. A pH between 10 and 11 yields the highest iron grade and recovery. At lower pH

values, the recovery drops significantly.

197
2. Lower sodium concentrations yielded both a higher grade and a higher recovery

of iron. Decreasing sodium in the process water could substantially affect the

process performance.

3. Calcium acts as a nonselective flocculant. Higher calcium concentrations decrease

grade but improve recovery while lower calcium concentrations increase grade

but decrease recovery.

Another set of experiments was performed in the pilot scale thickener assessing six

different types of dispersants. The six dispersants tested included sodium silicate, Glass-

H (21-member sodium polyphosphate), a mixture of both sodium silicate and Glass-H,

2,000 MW polyacrylic acid, 4,000 MW polyacrylic acid, and 7,000 MW polyacrylic acid.

The dosages of the dispersants were selected based upon operational data from the mine

and from manufacturer recommendations.

Of the inorganic dispersants tested, Glass-H yielded the best grade/recovery curve for

iron and was able to reject the most phosphorus during desliming. Sodium silicate and a

mixture of the two dispersants did not yield grade/recovery curves comparable to that of

Glass-H alone. The sodium silicate seemed to lessen the dispersive ability of the Glass-H.

All three polyacrylic acids, varying in molecular weight from 2,000 to 7,000 Da,

performed nearly identically in the process for both iron recovery and phosphorus

rejection. It was found that Glass-H, at 0.05 lb/ton of ore, yielded the best grade/recovery

curve for iron; however, 0.25 lb/ton 4000 MW PAA was able to reject up to 40% more

198
phosphorus with concentrate phosphorus concentrations as low as 0.022%. The iron

grade/recovery curve for the Glass-H was only slightly better than the 4000 MW PAA.

The following recommendations can be made from this study:

1. Lowering sodium concentration within the plant should increase both the grade

and recovery of iron. This can be accomplished by switching from caustic to

calcium hydroxide as a pH modifier. This will require a higher dispersant dose to

account for the excess calcium in the process water.

2. Maintaining a high pH (10.9 – 11.0) during desliming seemed to produce the

highest iron grade and recovery.

3. Switching from Glass-H to a low molecular weight (2000-7000 DA) polyacrylic

acid at the proper dose may increase phosphorus rejection. A study should be

performed to determine if PAA can be used economically in the plant. If an

economical dose is achieved, nearly all liberated phosphorus can be rejected

during deslime thickening.

4. Monitoring conductivity will be more important if caustic is replaced with

calcium hydroxide as a pH modifier. Conductivity is a good measure of sodium

concentration and seen in previously in this project.

2.3.7 References

ASTM, 2006, D1971-02 Standard Practices for Digestion of Water Samples for

Determination of Metals by Flame Atomic Absorption, Graphite Furnace Atomic

199
Absroption, Plasma Emission Spectroscopy, or Plasma Mass Spectroscopy, Water and

Environmental Technology, ASTM International, Pennsylvania.

ASTM, 2010a, D511-09 Standard Test Methods for Calcium and Magnesium in Water,

Water and Environmental Technology, ASTM International, Pennsylvania.

ASTM, 2010b, D512-04 Standard Test Methods for Chloride Ion in Water, Water and

Environmental Technology, ASTM International, Pennsylvania.

ASTM, 2010c, D1068-05 Standard Test Methods for Iron in Water, Water and

Environmental Technology, ASTM International, Pennsylvania.

ASTM, 2010d, D4192-08 Standard Test Method for Potassium in Water by Atomic

Absorption Spectrophotometry, Water and Environmental Technology, ASTM

International, Pennsylvania.

Balajee, S. R., and Iwasaki, I., 1969, "Adsorption Mechanism of Starches in Flotation and

Flocculation of Iron Ores," AIME Transactions, Vol. 244, pp. 401-406.

O'Brien, Robert J., Ajoy Raje, Robert L. Spicer, Liguang Xu, Shiqi Bao, Scott Lambert,

and Burtron H. Davis. "Technology Development for Iron Fischer-Tropsch Catalysts."

Proceedings of the First Joint Power & Fuel Systems Contractors Conference,

Pittsburgh, PA. 1996.

Dzombak, D. A., and Morel, F., 1990, Surface complexation modeling : hydrous ferric

oxide, Wiley, 393 pp.

200
Eggleston, C. M., and Hochella, M. F., 1992, "The Structure of Hematite (001) Surfaces

by Scanning Tunneling Microscopy - Image Interpretation, Surface Relaxation, and Step

Structure," American Mineralogist, Vol. 77, No. 9-10, pp. 911-922.

Haselhuhn, H. J., Carlson, J. J., and Kawatra, S. K., 2012a, "Water chemistry analysis of

an industrial selective flocculation dispersion hematite ore concentrator plant,"

International Journal of Mineral Processing, Vol. 102–103, pp. 99-106.

Haselhuhn, H. J., Swanson, K. P., and Kawatra, S. K., 2012b, "The effect of CO2

sparging on the flocculation and filtration rate of concentrated hematite slurries,"

International Journal of Mineral Processing, Vol. 112-113, pp. 107-109.

Keranen, C. U., 1986, “Reagent Preparation, Distribution and Feeding Systems at the

Tilden Mine,” Design and Installation of Concentration and Dewatering Circuits, A. L.

Mular and M. A. Anderson eds., Society for Mining, Metallurgy and Exploration, pp.

308-319.

Kinniburgh, D. G., and Jackson, M. L., 1982, "Concentration and pH dependence of

calcium and zinc absorption by iron hydrous oxide gel," Soil Science Society of America

Journal, Vol. 46, No. 1, pp. 56-61.

Kinniburgh, D. G., Jackson, M. L., and Syers, J. K., 1976, "Adsorption of Alkaline Earth,

Transition, and Heavy Metal Cations by Hydrous Oxide Gels of Iron and Aluminum,"

Soil Science Society of America Journal, Vol. 40, No. 5, pp. 796-799.

Knovel, 2003, Knovel Critical Tables (2nd Edition), Knovel.

201
Laskowski, J. S., and Ralston, J., 1992, Colloid Chemistry in Mineral Processing,

Elsevier Science, 428 pp.

Pradip, 1994, “Reagents Design and Molecular Recognition at Mineral Surfaces,”

Reagents for Better Metallurgy, Society for Mining Metallurgy, pp. 245-252.

Pyykkö, P., and Atsumi, M., 2009, "Molecular Single-Bond Covalent Radii for Elements

1–118," Chemistry – A European Journal, Vol. 15, No. 1, pp. 186-197.

Ralph, J., and Chau, I., 2014, "Chicagon mine, Menominee iron range, Iron Co.,

Michigan, USA," Mindat.org, http://www.mindat.org/loc-3840.html, Accessed:

12/16/2014

Sparks, D. L., and Grundl, T. J., 1998, Mineral-Water Interfacial Reactions, American

Chemical Society, 438 pp.

Stumm, W., Sigg, L., and Sulzberger, B., 1992, Chemistry of the solid-water interface,

Wiley, 448 pp.

Weissenborn, P. K., 1996, "Behaviour of amylopectin and amylose components of starch

in the selective flocculation of ultrafine iron ore," International Journal of Mineral

Processing, Vol. 47, No. 3-4, pp. 197-211.

202
3 Effects of Reagent Selection on Dispersion and Selective

Flocculation of Hematite Ore

3.1 Effects of Dispersant Adsorption on Settling Behavior of Iron Ore 5

3.1.1 Abstract

The settling behavior of ground hematite-rich iron ore at a pH of 11 was assessed in the

presence of aqueous calcium; sodium silicate, sodium tripolyphosphate (STPP), sodium

hexametaphosphate (SHMP) and sodium henicosapolyphosphate (Glass-H) inorganic

dispersants; and Ethylenediaminetetraacetic acid (EDTA). Without calcium, the inorganic

dispersants had no effect on the dispersive stability of the ore. This suggests that the

inorganic dispersants do not adsorb to the oxide surfaces at a pH of 11 due to their

inability to displace the negatively charged hydroxyl groups at the oxide-water interface.

In the absence of calcium, EDTA was found to increase the dispersion of the ore,

suggesting EDTA adsorption to hydrated iron oxide without the presence of positively

charged surface sites. Calcium was found to lower the dispersive stability of the ore by

flocculating fine particles.

5
Some of the material contained in this chapter was previously published in the journal
“Minerals and Metallurgical Processing.” It is included in this dissertation with
permission from the journal.
Citation:
Haselhuhn, H. J., 2013, "Dispersant Adsorption and Effects on Settling Behavior of Iron
Ore," Minerals & Metallurgical Processing, Vol. 30, No. 3, pp. 188-189.
203
In the presence of 20 ppm of aqueous calcium, each dispersant affected the dispersive

stability of the iron ore differently. Sodium silicate caused little to no dispersion in the

presence of 20 ppm calcium. All of the phosphate based dispersants effectively re-

dispersed the ore in the presence of 20 ppm calcium. EDTA caused further flocculation

of the ore in the presence of 20 ppm calcium. It has been concluded that inorganic

dispersants require positively charged surface sites, such as calcium inner-sphere

complexes, to adsorb and cause dispersion of an iron ore pulp at a pH of 11. EDTA only

causes dispersion without positively charged surface site. The order of dispersant

effectiveness from lowest to highest in the presence of 20 ppm calcium at a pH of 11 was

found to be Ethylenediamenetetraacetic acid (EDTA) < sodium silicate < sodium

tripolyphosphate (STPP) < sodium henicosapolyphosphate (Glass-H) < sodium

hexametaphosphate(SHMP).

3.1.2 Introduction

Many reagents are added to enhance the separation of iron oxide from its impurities in

iron ore concentration processes. While processing fine (< 50 μm liberation size) non-

magnetic ores, such as hematite, the process relies on selective flocculation and

dispersion followed by reverse flotation to produce a concentrated iron oxide product

(Green et al., 1984). Selective flocculation and dispersion relies on several reagents to

effectively separate iron ore particles based on mineralogy. These reagents adsorb to the

particle surfaces causing a change in surface chemistry that either flocculates or disperses

the ore particles. An overall dispersant, such as those shown in Table 3.1, is used to

disperse iron ore particles from the gangue minerals present in the ore (Green et al., 1984;

204
Potapova et al., 2011). A selective flocculant, such as corn starch or a polyacrylamide, is

used to selectively flocculate iron oxide particles (Green et al., 1984; Iwasaki, 1989). The

flocs of iron oxide particles sink while the dispersed gangue minerals are rejected in the

overflow. The effectiveness of reagent adsorption onto particle surfaces is critical during

desliming to promote effective selective flocculation and dispersion.

205
Table 3.1: Chemical names and structures of common dispersants used in iron ore

concentration processes.

Chemical Name Trade Name Chemical Formula Structure


O
-
+ O Si
Sodium Silicate Water Glass Na2SiO3 Na -
O
+
Na
+
Na +
- -
Na
O O
O
P P
Sodium Tripoly- O O
-
STPP Na5P3O10 +
O O Na
+
phosphate Na
O
-
P O
-
O
+
Na
+
Na -
O
O O
P - +
O O P O Na
Sodium - P
O
O
+ O -
Hexameta- SHMP Na6P6O18 Na O Na
+

O P
phosphate O
O P O O
P
- -
O O
+ O Na
+
Na
O O
O
Sodium Henicosa- P +
Glass-H Na23P21O64 O
- P P - Na
polyphosphate O
-
O
O
- O
O
- O
19 23
+
Na
-
O
+
Na O
-
O
Ethylene-diamine- N O
EDTA Na4C10H12N2O8
tetraacetic Acid O N
-
O
+
O Na
-
O
+
Na

The mechanisms of ion adsorption to iron oxide and silica surfaces have been widely

established. An ion will adsorb to a hydrated metal oxide surface via either an outer-

206
sphere complex or an inner-sphere complex as shown in Figure 3.1. An outer-sphere

complex is an ion drawn to an oppositely charged surface by electrostatic attraction. The

ion does not form a bond with the surface and is surrounded by one or more water

molecules. An inner-sphere complex is an ion that is chemically adsorbed to the surface.

Inner-sphere complexes can drastically alter the surface charge and zeta potential of the

metal oxide (Potapova et al., 2011).

O H H
H H O
Outer-Sphere

H
H
H Cl- Na+

O
O

O
Complexes

H
H

H
O-
O
H H O
O H H
Inner-Sphere
P Mg+
Complexes
H H+ H
Surface
O O F O O O O-
Hydroxyls

M M M M M M M Metal

O O O O O O O Oxygen
O

Figure 3.1: Surface complex formation at the metal-oxide/water interface. Adapted from

(Stumm et al., 1992).

Adsorption of both anions and cations to hydrated iron oxide surfaces are a function of

pH as shown in Figure 3.2. At any given pH, certain ions are adsorbed more readily than

others. For instance, at a pH of 6.5, the extent of adsorption of anions on hydrated iron

oxide decreases as follows: H3SiO4- > H2PO4- > MoO42- > SO42- > Cl- = NO3- (Ryden et

207
al., 1987; Stumm et al., 1992). In general, at a low pH, the surface is positively charged

and attracts anions that can either form inner-sphere or outer-sphere complexes. At a

higher pH, the surface carries a negative charge and attracts cations that can form inner-

sphere or outer-sphere complexes.

208
(a)

Pb2+
Cu2+
Zn2+
Ca2+
Mg2+

H3SiO4-
(b)
H2PO4-
HPO42-
SO42-

Figure 3.2: pH dependance of cation (a) and anion (b) adsorption (inner-sphere

complexation) on hydrated iron oxide surfaces (Dzombak et al., 1990; Kinniburgh et al.,

1976; Stumm et al., 1992)

209
Concentration processes that separate hematite, goethite and/or magnetite from siliceous

gangue minerals typically use reverse cationic flotation as their primary means of

separation. To maximize dispersion in these processes, sodium hydroxide is used to

adjust their pH to between 10.5 and 11 (Benner et al., 1994). In an alkaline environment

such as this, adsorption of dispersive anions (silicates and polyphosphates) to hydrated

oxide surfaces can be limited due to the highly negative charge of the hydrated oxide

surfaces. Even in these highly alkaline process conditions, dispersion is still

accomplished with the aforementioned dispersants. There are three possible reasons for

this phenomenon:

• The dispersive anions coat surfaces with inner-sphere complexes, displace

hydroxyl groups at the surface, and decrease the surface charge of all particles

• The anions form stable complexes with calcium and magnesium in solution,

withdrawing them from the particle surfaces, thereby decreasing the surface

charge of all particles (Ding et al., 2007)

• The anions form stable surface complexes with positively charged calcium and

magnesium inner-sphere complexes at the surface. This neutralizes calcium and

magnesium surface complexes and decreases the surface charge causing

dispersion of particles (Green et al., 1984)

Illustrations for these methods of dispersion in alkaline conditions using a polyphosphate

as an example are shown in Figure 3.3.

210
Theory 2 Theory 3

O O

O- O- O- O-
Theory 1
P P

Poly
O
Poly
O O
O- O-

P P
P

O O O O
Poly
O
Ca

Ca Ca
Inner-Sphere
P
Complexes
Surface
O O O- O- O- O O
Hydroxyls

M M M M M M M Metal

O O O O
O
O O O Oxygen

Figure 3.3: Illustrations of the 3 theories behind dispersant action at metal oxide

surfaces.

Settling behavior is a good indication of the dispersive stability of a slurry of charged

particles. Particles that are highly dispersed will settle very slowly, if at all. Particles that

are flocculated will settle quickly. Settling rate can be determined by monitoring the total

suspended solids concentration over time. As solids settle, the totals suspended solids

concentration will decrease. Total suspended solids concentration can be measured by


211
UV/VIS spectroscopy at 810 nm with an appropriate linear calibration (Hach, 2010;

Krawczyk et al., 1959).

3.1.3 Materials and Methods

The iron ore that was used was sampled at a local hematite concentrator at a location

between the grinding circuit and the deslime thickeners. X-ray diffraction patterns show

the presence of quartz (SiO2), magnetite (Fe3O4), hematite (Fe2O3), as well as trace

amounts of siderite (FeCO3) and dolomite (CaMg(CO3)2), as shown in Figure 3.4.

Elemental analysis on the dried solid particles was conducted by acid digestion and

atomic absorption spectroscopy (ASTM, 2006). The results of the elemental analysis are

shown in Table 3.2. Particle size analysis using a Microtrac™ laser scattering particle

size analyzer yielded an 80% passing size of 30.99 μm.

212
Figure 3.4: XRD patterns of iron ore used for dispersant study with major peaks

identified.

Table 3.2: Elemental analysis of iron ore used in this study

Constituent Assay
Iron 38.52%
Calcium 0.29%
Magnesium 0.30%
Acid Insolubles 40.79%

Samples of filtered, dried iron ore were split using a micro-rotary riffle splitter into

representative 1 gram splits. The 1 gram splits were then hydrated and sealed for 24 hours

213
with distilled water adjusted to a pH of 11 with NaOH. Upon hydration, a 1 gram sample

was added to a 250 mL plastic beaker with 200 mL of distilled water that was adjusted

previously to a pH of 11 with NaOH. A small laboratory stand mixer with a plastic

impeller and shaft was used to keep the particles in solution suspended. Plastic laboratory

equipment was used to prevent any surface adsorption to laboratory instruments. During

experiments requiring the use of calcium, a 4.5 g/L (as Ca2+) solution of calcium chloride

was added using an adjustable pipette. Dispersants were added in a similar manner using

1 g/L solutions of their solid anhydrous sodium salts. The dispersant dosage (0.1 lb/ton

and 0.2 lb/ton) was chosen to be a comparable value that has been used in the local iron

ore concentrator that provided the iron ore sample.

The effectiveness of dispersants was determined by measuring total suspended solids

concentration in a 3 mL sample of the mixed slurry using a 5 minute time lapse UV/VIS

measurement at 810 nm. A Hach DR5000™ UV/VIS spectrophotometer was used in a

time-lapse mode with 10mm light path cuvettes for this test. Each measurement was

made in triplicate to ensure accuracy and statistical significance of data. The raw data

was normalized linearly to a distilled water trial in order to make direct comparisons

between trials. The absorbance at 810 nm was calibrated by comparison to gravimetric

total suspended solids measurements using various concentrations of the iron ore in

distilled water at a pH of 11.

3.1.4 Results and Discussion

Each dispersant showed differing levels of dispersive ability for the ground iron ore at a

pH of 11. Settling behavior evaluations for varying doses of sodium silicate (Na2SiO3)
214
are shown in Figure 3.5. Without calcium in the slurry, the dispersive stability remains

constant at the same level as distilled water regardless of sodium silicate dose. This

shows that the adsorption theory in which the anion displaces the hydroxyl groups does

not apply (theory 1 described above) in these conditions. When calcium is added at 20

ppm, the dispersive stability is lowered drastically. This is due to the addition of

positively charged calcium inner-sphere complexes to the negatively charged hydroxyl

groups causing flocculation of once-suspended particles. When sodium silicate is added

as a dispersant to the calcium-rich slurry, the dispersive stability is unchanged. This

shows that sodium silicate is not an effective dispersant at these doses for a slurry

containing 20 ppm calcium.

(A)

215
(B)

Figure 3.5: Sodium silicate effectiveness as a dispersant. Total suspended solids over

time of hematite ore in a distilled water solution at a pH of 11. Sodium silicate dispersant

added without calcium present (a) and after calcium had been added (b).

The settling behavior for varying doses of sodium tripolyphosphate (Na5P3O10, STPP) is

shown in Figure 3.6. Without calcium in the slurry, the dispersive stability remains

constant at the same level as distilled water regardless of the amount of STPP added. This

is the same phenomenon seen when adding sodium silicate. When calcium is added at 20

ppm, the dispersive stability is again lowered drastically. When STPP is added as a

216
dispersant to the calcium-rich slurry, the dispersive stability increased slightly. When

STPP was added at 0.1 lb/ton and 0.2 lb/ton, the final total suspended solids increased by

14 mg/L and 16 mg/L, respectively. This shows that STPP will re-disperse the ore in the

presence of calcium with diminishing returns on dispersant dose. This also shows that

STPP does not conform to theory 2 described above in which the dispersant removes

calcium inn-sphere complexes from the surface of the particles.

(A)

217
(B)

Figure 3.6: Sodium tripolyphosphate effectiveness as a dispersant. Total suspended

solids over time of hematite ore in a distilled water solution at a pH of 11. Sodium

tripolyphosphate (STPP) dispersant added without calcium present (a) and after calcium

had been added (b).

The settling behavior for varying doses of sodium hexametaphosphate (Na6P6O18,

SHMP) is shown in Figure 3.7. Without calcium in the slurry, the dispersive stability

remains constant at the same level as distilled water regardless of the amount of SHMP

added. This is the same phenomenon seen when adding sodium silicate and STPP. When

calcium is added at 20 ppm, the dispersive stability is again lowered drastically. When

SHMP is added as a dispersant to the calcium-rich slurry, the dispersive stability


218
increased substantially. When SHMP was added at 0.1 lb/ton and 0.2 lb/ton, the final

total suspended solids increased by 18 mg/L and 30 mg/L, respectively. This shows that

SHMP will re-disperse the ore in the presence of calcium with little diminish in returns

upon increased dosages. SHMP may conform to theory 2, theory 3 or both because the

dispersive effect of the SHMP changed progressively with increased dispersant dose.

(A)

219
(B)

Figure 3.7: Sodium hexametaphosphate effectiveness as a dispersant. Total suspended

solids over time of hematite ore in a distilled water solution at a pH of 11. Sodium

hexametaphosphate (SHMP) dispersant added without calcium present (a) and after

calcium had been added (b).

The settling behavior for varying doses of sodium henicosapolyphosphate (Na23P21O64,

Glass-H) is shown in Figure 3.8. Again, without calcium in the slurry, the dispersive

stability remains constant at the same level as distilled water regardless of the amount of

Glass-H added. This shows that theory 1 as described above does not apply for silicate or

phosphate based dispersants. When calcium is added at 20 ppm, the dispersive stability is

again lowered drastically. When Glass-H is added as a dispersant to the calcium-rich


220
slurry, the dispersive stability increased; however, not as much as when SHMP was

added. When Glass-H was added at 0.1 lb/ton and 0.2 lb/ton, the final total suspended

solids increased by 8 mg/L and 25 mg/L, respectively. This shows that Glass-H will re-

disperse the ore in the presence of calcium with little to no diminish in returns upon

increased dosages. Glass-H may also conform to theory 2, theory 3 or both because the

dispersive effect of the Glass-H changed progressively with increased dispersant dose.

(A)

221
(B)

Figure 3.8: Sodium henicosapolyphosphate effectiveness as a dispersant. Total

suspended solids over time of hematite ore in a distilled water solution at a pH of 11.

Sodium henicosapolyphosphate (Glass-H) dispersant added without calcium present (a)

and after calcium had been added (b).

The settling behavior for varying doses of ethylenediaminetetraacetic acid (C10H16N2O8,

EDTA) is shown in Figure 3.9. Unlike the silicate and phosphate dispersants, EDTA

causes a slight increase in dispersive stability without the addition of calcium in the

slurry. This shows that theory 1 as described above could be applied in this situation. As

before, when calcium is added at 20 ppm the dispersive stability is lowered drastically;

however, when EDTA is added as a dispersant to the calcium-rich slurry, the dispersive
222
stability further decreases. When EDTA was added at 0.1 lb/ton, the final total suspended

solids decreased by 37 mg/L; however, when EDTA was added at 0.2 lb/ton there was no

change in final total suspended solids. This shows that in the presence of calcium, EDTA

will cause flocculation, and in the absence of calcium EDTA will cause slight dispersion.

Also, this shows that EDTA does not follow theory 2 or 3, as mentioned above.

(A)

223
(B)

Figure 3.9: Ethylenediaminetetraacetic acid effectiveness as a dispersant. Total

suspended solids over time of hematite ore in a distilled water solution at a pH of 11.

Ethylenediaminetetraacetic acid (EDTA) dispersant added without calcium present (a)

and after calcium had been added (b).

3.1.5 Conclusions

The settling behavior of ground hematite-rich iron ore at a pH of 11 was assessed in the

presence of calcium and several common dispersants used in iron ore concentration

processes. Three possible adsorption mechanisms for these dispersants have been shown

in literature:

224
1. The dispersive anions coat surfaces with inner-sphere complexes by displacing

hydroxyl groups at the oxide surfaces (Stumm et al., 1992)

2. The anions form stable complexes with calcium in solution causing these ions to

dissolve from the particle surfaces (Ding et al., 2007)

3. The anions form stable surface complexes with positively charged calcium inner-

sphere complexes at the surface thereby neutralizing calcium surface complexes

(Green et al., 1984)

The dispersive ability of sodium silicate, sodium tripolyphosphate (STPP), sodium

hexametaphosphate (SHMP), sodium henicosapolyphosphate (Glass-H), and

ethylenediaminetetraacetic acid (EDTA) was evaluated with and without aqueous

calcium present in the slurry. Without calcium, the dispersants had no effect on the

dispersive stability of the ore, with EDTA being the only exception. This suggests that

the inorganic dispersants do not adsorb to the oxide surfaces at a pH of 11 due to their

inability to displace the negatively charged hydroxyl groups at the oxide-water interface.

EDTA did cause slight dispersion of the ore, indicating it can displace the hydroxyl

groups forming a stable complex.

Calcium was found to lower the dispersive stability of the ore by flocculating fine

particles. Each dispersant affected the dispersive stability differently when calcium was

present. Sodium silicate caused little to no dispersion in the presence of 20 ppm calcium.

All of the phosphate based dispersants effectively re-dispersed the ore in the presence of

20 ppm calcium. STPP re-dispersed the ore slightly with diminishing returns upon

increased dosage. This shows that STPP likely adsorbs to positively charged calcium

225
inner-sphere complexes thereby reversing the charge of the surface site. SHMP re-

dispersed the ore with little diminishing returns at the dosages tested. This shows that

SHMP may complex and remove calcium from the surface, complex with calcium inner-

sphere complexes without removing the calcium from the surface, or a combination of

both adsorption mechanisms. Glass-H re-dispersed the ore with little to no diminishing

returns at the dosages tested. The adsorption mechanism of Glass-H follows that of

SHMP. EDTA causes further flocculation suggesting that an aqueous EDTA molecule

may bind with two calcium complexes; sometimes, one calcium being bound to a

neighboring particle. This could form a more stable EDTA complex at the particle

surface resulting in flocculation.

It has been concluded that anionic dispersants require positively charged surface sites,

such as calcium inner-sphere complexes, to adsorb and cause dispersion of an iron ore

pulp at a pH of 11. The order of dispersant effectiveness from lowest to highest in the

presence of 20 ppm calcium at a pH of 11 was found to be ethylenediaminetetraacetic

acid < sodium silicate < sodium tripolyphosphate < sodium henicosapolyphosphate <

sodium hexametaphosphate.

3.1.6 References

ASTM, 2006, D1971-02 Standard Practices for Digestion of Water Samples for

Determination of Metals by Flame Atomic Absorption, Graphite Furnace Atomic

Absroption, Plasma Emission Spectroscopy, or Plasma Mass Spectroscopy, Water and

Environmental Technology, ASTM International, Pennsylvania.

226
Benner, B. R., and Goetzman, H. E., 1994, “Reagent Use in the Iron Ore Industry,”

Reagents for Better Metallurgy, Society for Mining, Metallurgy and Exploration, pp.

143-153.

Ding, H., Lin, H., and Deng, Y., 2007, "Depressing effect of sodium hexametaphosphate

on apatite in flotation of rutile," Journal of University of Science and Technology Beijing,

Vol. 14, No. 3, pp. 200-203.

Dzombak, D. A., and Morel, F., 1990, Surface complexation modeling : hydrous ferric

oxide, Wiley, 393 pp.

Green, R. E., and Colombo, A. F., 1984, “Dispersion-Selective Flocculation-Desliming

Characteristics of Oxidized Taconites,” US Bureau of Mines, Report of Investigations, RI

8867, 24 pp.

Hach, 2010, “Method 8006: Suspended Solids - Photometric Method,” Hach Methods

Iwasaki, I., 1989, “Bridging Theory and Practice in Iron Ore Flotation,” Advances in

Coal and Mineral Processing Using Flotation, Society for Mining, Metallurgy and

Exploration, pp. 177-190.

Kinniburgh, D. G., Jackson, M. L., and Syers, J. K., 1976, "Adsorption of Alkaline Earth,

Transition, and Heavy Metal Cations by Hydrous Oxide Gels of Iron and Aluminum,"

Soil Science Society of America Journal, Vol. 40, No. 5, pp. 796-799.

227
Krawczyk, D., and Gonglewski, N., 1959, "Determining Suspended Solids Using a

Spectrophotometer," Sewage and Industrial Wastes, Vol. 31, No. 10, pp. 1159-1164.

Potapova, E., Yang, X., Grahn, M., Holmgren, A., Forsmo, S. P. E., Fredriksson, A., and

Hedlund, J., 2011, "The effect of calcium ions, sodium silicate and surfactant on charge

and wettability of magnetite," Colloids and Surfaces A: Physicochemical and

Engineering Aspects, Vol. 386, No. 1–3, pp. 79-86.

Ryden, J. C., Syers, J. K., and Tillman, R. W., 1987, "Inorganic anion sorption and

interactions with phosphate sorption by hydrous ferric oxide gel," Journal of Soil

Science, Vol. 38, No. 2, pp. 211-217.

Stumm, W., Sigg, L., and Sulzberger, B., 1992, Chemistry of the solid-water interface,

Wiley, 448 pp.

228
3.2 Comparison of Dispersants for Selective Flocculation of Hematite

Ore at a Pilot Scale6

3.2.1 Abstract

Selective flocculation and dispersion is a process used to separate finely liberated

(<25μm) hematite ore from gangue minerals. It requires a specific water chemistry, a

selective flocculant and a dispersant to yield an effective separation. This research

focused on finding the most effective dispersant for the hematite concentration process

using a laboratory scale continuous deslime thickener. The dispersants studied were

sodium silicate, sodium henicosapolyphosphate (21-member polyphosphate), 2000

molecular weight (MW) polyacrylic acid (PAA), 4000 MW PAA and 7000 MW PAA. It

was found that sodium henicosapolyphosphate at 0.05 pounds of dispersant per ton of ore

yielded the best grade/recovery curve for iron. The 4000 MW polyacrylic acid at 0.25

pounds of PAA per ton of ore had an iron grade/recovery curve nearly that of sodium

henicosapolyphosphate but yielded a lower phosphorus concentration in the product

(down to 0.022% P).

3.2.2 Introduction

The United States produced 53.2 million metric tons of concentrated iron ore in 2012

(Tuck, 2013). As iron ore reserves containing high grade iron ore have been depleted for

6
The material contained in this chapter is under review in the journal “Minerals and
Metallurgical Processing.” It is included in this dissertation with permission from the
journal.
Citation:
Haselhuhn, H. J., Kawatra, S. K., 2015, "Design of a Continuous Pilot Scale Deslime
Thickener," Minerals & Metallurgical Processing, Accepted.
229
over five decades, the U.S. has relied on lower grade ores for all domestic production.

Most of the ore has been obtained in the form of low-grade unoxidized magnetite; this

ore has been selected because Fe3O4 is easily separated from the gangue minerals using

low-intensity wet magnetic separation. Fully oxidized iron ores containing hematite and

goethite are very abundant throughout the world; however, beneficiation of these ores has

been a stumbling block, both technologically and economically (Colombo, 1980).

Recently, there has been much recent interest in the beneficiation and processing of these

ores (Bolen, 2014; Carlson et al., 2010, 2013; Carlson, 2008; Halt, 2014; Halt et al.,

2014a; Halt et al., 2014b; Haselhuhn, 2012, 2013; Haselhuhn et al., 2012a; Haselhuhn et

al., 2012b; Kawatra et al., 2011; Liu et al., 2014; Manouchehri, 2014; Sandvik et al.,

2014; Semberg et al., 2014)

One commercial facility has been capable of economically separating low-grade fully

oxidized ores since 1974 (Colombo, 1980). This facility accounted for about 15% of U.S.

production in 2012 and relied on a process known as selective flocculation and dispersion

(Cliffs Natural Resources, 2011; Tuck, 2013). Selective flocculation and dispersion is a

process used to concentrate fine, non-magnetic mineral particles such as hematite. The

process relies on both water chemistry and surface chemistry to separate mineral

particles. It requires two types of reagents: a selective flocculant and a dispersant. For

concentrating hematite ore, cooked acid-modified cornstarch has been used as a selective

flocculant. The dispersant used has varied over time as polymer technology has

developed.

230
To facilitate adequate dispersion, sodium hydroxide has been used to increase slurry pH

and create a negatively charged zeta potential at mineral surfaces. The negative zeta

potential of all mineral surfaces is important to prevent the natural flocculation of the

mineral particles. Both high pH and low concentration of polyvalent cations are

necessary to achieve adequate natural dispersion of the particles (Haselhuhn, 2012).

In addition to sodium hydroxide, a number of different dispersants have been used to

increase the dispersion of the ore. Many of these dispersants have been chosen to combat

the effects of polyvalent cations in the process water (Haselhuhn, 2013). Since the

inception of the process in 1974 at a plant scale, these dispersants have included sodium

silicate, sodium tripolyphosphate, several experimental phosphate-based dispersants and

a phosphate polymer averaging 21 phosphate groups known as Glass-H (Siirak et al.,

1988).

Sodium silicate was the most widely used dispersant in the world for iron ore

beneficiation (Ma, 2011a). Sodium silicate was used as the dispersant of choice in

selective flocculation and dispersion processes until it was recognized that phosphorus

rejection could be accomplished using phosphate-based dispersants. Siirak and Hancock

(1988) showed that phosphorus occurred in the Lake Superior district ore body in two

forms: approximately 50% as apatite (Ca5(PO4)3(F,Cl,OH)) grains and the remainder as

non-liberatable phosphorus highly disseminated throughout the iron oxide grains. Sodium

tripolyphosphate was the first phosphate dispersant tested with some success at rejecting

phosphorus during desliming (Siirak et al., 1988). The work of Siirak and Hancock

(1988) went on to show that longer polyphosphate chain lengths (P6, P13 and P21 chain
231
lengths) increased rejection of phosphorus during desliming. It was found that a 21

member sodium polyphosphate could be used economically to reject phosphorus during

desliming (Siirak et al., 1988). This dispersant has been used since that time at the only

selective flocculation and dispersion iron ore beneficiation plant in North America.

Recently, the use of polyacrylic acids were researched as a suitable replacement for

sodium polyphosphate dispersants in iron oxide – kaolinite systems (Ma, 2011a; Ma,

2011b). The studies performed by Ma (2011a,b) suggested that a low molecular weight

dispersant could perform nearly as well as polyphosphates to disperse kaolinite in iron

ore over a wide pH range. This study did not assess the dispersants ability to reject

phosphorus from the ore. More research was needed to determine if low molecular

weight polyacrylic acids could both meet the dispersion capabilities of polyphosphates

while still rejecting phosphorus during the desliming stage.

This study compared the effectiveness of sodium silicate, sodium henicosapolyphosphate

(21 member polyphosphate) and three different polyacrylic acid dispersants with varying

molecular weights (2,000, 4,000, 7,000 Da). Unlike previous studies, this research was

conducted in a laboratory scale continuous deslime thickener. Iron recovery and

phosphorus rejection was reported for each dispersant over varying flocculant dosages.

The novel design of this thickener and the experimental procedures used were capable of

replicating the grade and recovery seen at an industrial hematite beneficiation plant using

the same technology and reagents.

232
3.2.3 Materials and Methods

3.2.3.1 Thickener Design

A laboratory scale deslime thickener was constructed using a 3 inch diameter, 50 inch tall

acrylic column with an acrylic overflow collector and a stainless steel cone bottom. The

feed slurry basin consisted of a powder-coated conical tank and a mixer on a small variac

to adjust the mixer speed. The feed slurry was pumped into the feed well of the thickener

using a Masterflex Easy-Load 7518-12 peristaltic pump. The thickener overflow was

gravity drained from the thickener overflow collector back into the feed basin. The

thickener underflow was pumped back into the feed basin using a Manostat Veristaltic

Pump Jr. peristaltic pump. This experimental setup, including pertinent dimensions, can

be seen in Figure 3.10. This thickener design was used to simulate a continuous deslime

thickener without the need for excessive amounts of iron ore by recycling the underflow

and overflow of the column back into the feed basin.

233
8.25in.

3.09in.

6.88in.
8.38in.

18.00in.

49.81in.
Deslime Feed
Thickener Tank
13.00in.

21.13in.

2.50in.

Feed
U/F 1.00in. Pump
4.63in. Pump

Figure 3.10: Scaled diagram of the continuous laboratory scale deslime thickener

234
3.2.3.2 Ore Preparation

Fully oxidized (hematite) iron ore was collected at an operating hematite beneficiation

plant prior to any dispersant or flocculant addition in the circuit. The ore was collected

using a cross-flow sample cutter on a falling stream of slurry between primary

autogenous grinding and the dispersant addition point as can be seen in Figure 1.2. This

sample point was chosen because no reagents aside from caustic soda (NaOH) had been

added up to that point. Sampling at this location also allowed for a smaller sample to be

representative of the bulk ore body due to the reduced particle size. Once sampled, the

slurry was dewatered and dried in a convection oven at 107°ÛC for 24 hours. Once dry,

the material was processed in a roll mill to break up chunks of dried material. A particle

size analysis on this material showed an 80% passing size of 1.41 mm. The particle size

required to liberate the iron oxide grains from the gangue minerals was known to be 82%

passing at 25 μm (Siirak et al., 1988). To reach this particle size, the dried ore was

ground in a laboratory rod mill for 70 minutes. The resultant ore was analyzed for particle

size using a laser scattering particle size analyzer. This gave an 80% passing size of 23.1

μm.

Once the ore was ground, it was split using a rotary riffle splitter into 400 gram samples.

The samples were placed in plastic bags and immediately inerted under argon to prevent

oxidation of any cleaved iron grains. One sample was saved and analyzed for both iron

and phosphorus concentration. The ore contained 39.7% iron and 0.0268% phosphorus.

3.2.3.3 Water Chemistry and Reagent Selection

235
The water used for these experiments was made using distilled water and laboratory

grade salts to match the water chemistry found at an operating hematite beneficiation

plant (Haselhuhn et al., 2012a). The water chemistry was held constant for each trial and

can be found in Table 3.3. Laboratory grade reagents of sodium hydroxide (NaOH),

sodium chloride (NaCl), potassium chloride (KCl), magnesium chloride (MgCl2·6H2O),

calcium chloride (CaCl2·2H2O), sodium carbonate (Na2CO3, anhydrous), and sodium

sulfate (Na2SO4, anhydrous) were used to reach the desired levels of ionic constituents in

the simulated process water.

An ~400 gram sample of ground iron ore was mixed in the feed basin at 2% solids in

simulated process water. A dispersant was added at a fixed dose and cooked acid-

modified cornstarch was added at 0.05 lb/ton7 increments up to 0.20 lb/ton. The tested

dispersants included 0.3 lb/ton sodium silicate, 0.05 lb/ton Glass-H, 0.05 lb/ton Glass-H

+ 0.3 lb/ton sodium silicate, 0.25 lb/ton 2000 MW PAA, 0.25 lb/ton 4000 MW PAA, and

0.25 lb/ton 7000 MW PAA. The dosages of the dispersants were selected based upon

operational data from an operating beneficiation plant and from manufacturer

recommendations. The structures of these dispersants can be seen in Table 3.4. Overflow

and underflow samples were taken 30 minutes (~1.6 underflow residence times) after

each starch addition. These samples were filtered and dried. The filtrate volume was

measured in a 500 mL graduated cylinder and the dried solids were weighed on an

7
lb/ton refers to pounds (U.S.) of the reagent per ton of ore in the feed material of the

thickener. This applies for every instance of lb/ton within this document.

236
analytical balance. Iron and phosphorus analysis using acid digestion of the dried solids

and UV-VIS spectroscopy in conjunction with solids concentrations yielded

grade/recovery curves for each dispersant at each starch dose.

Table 3.3: Water chemistry used during laboratory thickening tests

Component Concentration

Sodium 495 ppm

Potassium 10 ppm

Magnesium 2 ppm

Calcium 3 ppm

Chloride 20.2 ppm

Carbonate 396.5 ppm

Sulfate 399.3 ppm

pH 11

237
Table 3.4: Chemical structures of dispersants used in this study

Dispersant Structure

Silicate

Henicosapolyphosphate

(21-member polyphosphate)

Polyacrylic Acid

3.2.3.4 Iron and Phosphorus Analyses

The iron content of samples was measured using a UV-VIS spectrometer. A small sample

of dried solids (<0.4 grams) was digested in 20 mL of hydrochloric acid for 20 minutes.

The iron-bearing liquid was transferred to a 100 mL class-A volumetric flask and diluted

using distilled water. This was allowed to cool to room temperature. To prepare the

indicator solution, 5 mL of 0.1 M ammonium acetate solution, 5 mL of 10%

hydroxylamine hydrochloride solution and 5 mL of 0.1% 1,10-phenanthroline solution

238
were added to a 50 mL class- A volumetric flask. Once cool, 100 μL of the digested

solution was added to the indicator solution. The indicator solution was diluted to 50 mL

with distilled water. The absorbance at 510 nm of the indicator solution was correlated to

the iron concentration within the sample using a calibration curve prepared with an

atomic absorption spectroscopy standard. Phosphorus concentration was measure using a

phosphor-molybdenum-blue indicator and UV-VIS spectroscopy according to ASTM

standard E1070-11 (2011).

3.2.4 Results and Discussion

3.2.4.1 Iron Beneficiation

The laboratory deslime thickener was run with six different dispersants to determine

which dispersant yielded the best grade/recovery curve for iron beneficiation. Curves

were constructed by incrementally increasing the starch flocculant dose from 0.0 to 0.2

lb/ton. The grade and recovery curves for these tests can be seen in Figure 3.11 and 3.12.

As expected, with no flocculant the recovery of iron for all dispersants was low (42%-

52%). As starch dose was increased, the iron recovery increased; however, the grade

decreased.

The type of inorganic dispersant had a significant impact on the grade/recovery curve for

iron. Sodium silicate had the least desirable grade/recovery curve. Sodium

henicosapolyphosphate (21-member polyphosphate) had a significantly higher grade and

recovery at each starch dose. The addition of sodium silicate to the sodium

henicosapolyphosphate did not result in a better grade/recovery curve than the sodium

239
henicosapolyphosphate alone. This suggests that sodium silicate negatively impacts the

dispersive stability achieved by other dispersants when mixed.

Figure 3.11: Iron grade/recovery using inorganic dispersants with varying flocculant

concentration

The molecular weight of the polyacrylic acids used as dispersants did not significantly

alter the effectiveness of the dispersant, as shown in Figure 3.12. All three dispersants

resulted in grade/recovery curves that were very close to each other. The polyacrylic

acids did not perform as well as sodium henicosapolyphosphate for the recovery of iron

during desliming. However, the polyacrylic acids did out-perform sodium silicate and the

240
mixture of sodium silicate and sodium henicosapolyphosphate. This conclusion may be

different at different dosages of polyacrylic acid and sodium henicosapolyphosphate.

Figure 3.12: Iron grade/recovery using polyacrylic acid dispersants with varying

flocculant concentration

3.2.4.2 Phosphorus Rejection

The phosphorus concentration of both the thickener overflow and underflow solids was

analyzed to perform a mass balance around phosphorus in the process. This allowed for a

comparison to be made between the six different dispersants used in the study. Since both

241
the phosphorus rejection and the iron recovery were the two most important factors in

this study, they were plotted against each other for each dispersant. These plots can be

found in Figure 3.13 and 3.14.

As expected and reported by Siirak and Hancock (1988), the use of sodium

henicosapolyphosphate yielded the best phosphorus rejection and iron recovery at each

starch dose. The lowest phosphorus concentration achieved in the thickener underflow

using sodium henicosapolyphosphate as the dispersant was 0.0245% at an iron recovery

of 51.8%. Sodium silicate yielded the poorest phosphorus rejection and iron recovery at

each starch dose. The mixture of the two dispersants fell between the two, again

suggesting that the addition of sodium silicate is detrimental to the effectiveness of the

sodium henicosapolyphosphate dispersant.

242
Figure 3.13: Phosphorus rejection versus iron recovery for inorganic dispersants with

varying flocculant dose

243
Figure 3.14: Phosphorus rejection versus iron recovery for polyacrylic acid dispersants

with varying flocculant dose

3.2.5 Conclusions

A laboratory scale deslime thickener was used to compare the effectiveness of six

different dispersants ability to both enhance iron recovery and phosphorus rejection in the

selective flocculation-dispersion desliming process. The six dispersants tested included

sodium silicate, sodium henicosapolyphosphate (21-member polyphosphate), a mixture

of both sodium silicate and sodium henicosapolyphosphate, 2,000 MW polyacrylic acid,

4,000 MW polyacrylic acid, and 7,000 MW polyacrylic acid. The dosages of the

244
dispersants were selected based upon operational data from an operating beneficiation

plant and from manufacturer recommendations.

Of the inorganic dispersants tested, sodium henicosapolyphosphate yielded the best

grade/recovery curve for iron and was able to reject the most phosphorus during

desliming. Sodium silicate and a mixture of the two dispersants did not yield

grade/recovery curves comparable to that of sodium henicosapolyphosphate alone. The

sodium silicate seemed to lessen the dispersive ability of the sodium

henicosapolyphosphate.

All three polyacrylic acids, varying in molecular weight from 2,000 to 7,000 Da,

performed nearly identically in the process for both iron recovery and phosphorus

rejection. It was found that sodium henicosapolyphosphate, at 0.05 lb/ton of ore, yielded

the best grade/recovery curve for iron; however, 0.25 lb/ton 4000 MW PAA was able to

reject up to 40% more phosphorus with concentrate phosphorus concentrations as low as

0.022%. The iron grade/recovery curve for the sodium henicosapolyphosphate was only

slightly better than the 4000 MW PAA.

3.2.6 References

ASTM, 2011, E1070-11 Standard Test Method for Determination of Phosphorus in Iron

Ores by Phospho-Molybdenum-Blue Spectrophotometry, ASTM Interantional,

Conshohocken, PA.

Bolen, J., 2014, "Modern Air Pollution Control for Iron Ore Induration," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 103-114.

245
Carlson, J. J. and Kawatra, S. K., 2008, "Effect of Particle Shape on the Filtration Rate in

an Industrial Iron Ore Processing Plant," Minerals & Metallurgical Processing, Vol. 25,

No. 3, pp. 165-168.

Carlson, J. J., and Kawatra, S. K., 2010, "Effects of CO2 on the Zeta Potential of

Hematite," International Journal of Mineral Processing, Vol. 98, No. 1-2, pp. 8-14.

Carlson, J. J., and Kawatra, S. K., 2013, "Factors Affecting Zeta Potential of Iron

Oxides," Mineral Processing and Extractive Metallurgy Review, Vol. 34, No. 5, pp. 269-

303.

Cliffs Natural Resources, 2011, Steel Starts Here, Operations Data Fact Sheet, 4 pp.

Colombo, A. F., 1980, “Selective Flocculation and Flotation of Iron-Bearing Materials,”

Fine Particles Processing, AIME, pp. 1034-1056.

Halt, J. A., 2014, "Increasing the preheat strength of cornstarch-bonded pellets," Minerals

& Metallurgical Processing, Vol. 31, No. 3, pp. 179.

Halt, J. A., and Kawatra, S. K., 2014a, "Review of organic binders for iron ore

concentrate agglomeration," Minerals & Metallurgical Processing, Vol. 31, No. 2, pp.

73-94.

Halt, J. A., Nitz, M. C., Kawatra, S. K., and Dube, M., 2014b, "Iron Ore Pellet Dustiness

Part I: Factors Affecting Dust Generation," Mineral Processing and Extractive

Metallurgy Review, Vol. 36, No. 4, pp. 258-266.

246
Haselhuhn, H. J., 2012, "Water Chemistry Effects on the Zeta Potential of Concentrated

Hematite Ore," Minerals & Metallurgical Processing, Vol. 29, No. 2, pp. 135-136.

Haselhuhn, H. J., 2013, "Dispersant Adsorption and Effects on Settling Behavior of Iron

Ore," Minerals & Metallurgical Processing, Vol. 30, No. 3, pp. 188-189.

Haselhuhn, H. J., Carlson, J. J., and Kawatra, S. K., 2012a, "Water chemistry analysis of

an industrial selective flocculation dispersion hematite ore concentrator plant,"

International Journal of Mineral Processing, Vol. 102–103, pp. 99-106.

Haselhuhn, H. J., Swanson, K. P., and Kawatra, S. K., 2012b, "The effect of CO2

sparging on the flocculation and filtration rate of concentrated hematite slurries,"

International Journal of Mineral Processing, Vol. 112-113, pp. 107-109.

Kawatra, S. K., and Halt, J. A., 2011, "Binding effects in hematite and magnetite

concentrates," International Journal of Mineral Processing, Vol. 99, No. 1-4, pp. 39-42.

Liu, S., Zhau, Y., Wang, W., and Wen, S., 2014, "Beneficiation of a Low-grade,

Hematite-Magnetite Ore in China," Minerals & Metallurgical Processing, Vol. 31, No. 2,

pp. 136-142.

Ma, X., 2011a, “The Dispersion of Kaolinite,” Proceeding of the Iron Ore Conference,

The Australian Institute of Mining and Metallurgy, Perth, Australia, pp. 471-474.

247
Ma, X., 2011b, "Effect of a low-molecular-weight polyacrylic acid on the coagulation of

kaolinite particles," International Journal of Mineral Processing, Vol. 99, No. 1–4, pp.

17-20.

Manouchehri, H. R., 2014, "Pyrrhotite Flotation and its Selectivity against Pentlandite in

the Beneficiation of Nickeliferous Ores: An Electrochemistry Perspective," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 115-125.

Sandvik, K. L., and Larsen, E., 2014, "Iron Ore Flotation with Environmentally Friendly

Reagents," Minerals & Metallurgical Processing, Vol. 31, No. 2, pp. 95-102.

Semberg, P., Andersson, C., and Bjorkman, B., 2014, "Interaction between Iron Oxides

and Olivine in Magnetite Pellets during Reduction at 500°-1,300°C," Minerals &

Metallurgical Processing, Vol. 31, No. 2, pp. 126-135.

Siirak, J., and Hancock, B. A., 1988, “Progress in Developing a Flotation Phosphorus

Reduction Process at the Tilden Iron Ore Mine,” Proceedings of the XVI International

Mineral Processing Congress, Stockholm, Sweden, pp. 1393-1404.

Tuck, C. A., 2013, “Mineral Commodity Studies - Iron Ore,” US Geological Survey.

248
Conclusions

This dissertation discussed the importance of water chemistry, surface chemistry, and

reagent selection on the hematite dispersion and selective flocculation process. The

majority of the dissertation focused on the dependence of the previously under-studied

desliming process on water chemistry and surface chemistry phenomena. Experiments

were performed in a laboratory setting, in a pilot scale deslime thickener, and at a full

scale beneficiation process making this work the first of its kind to show these effects at

all three process scales.

These studies proved many hypotheses proposed throughout literature and found new

phenomena that had been previously untested in other literature studies. The links

between water chemistry, surface chemistry and reagent selection that are critical to the

only hematite beneficiation process successfully implemented at a full scale have been

shown and proven. In particular, it was shown that decreasing the sodium ion

concentration in the process increases both iron grade and recovery. This is an important

finding because increasing both the grade and the recovery is a very rare phenomenon

and is extraordinarily beneficial to this process. It was also proven that calcium causes

non-selective flocculation resulting in increased iron recovery at the expense of iron

grade in the product, a typical scenario for process changes. The optimal pH for the

process was found to be dependent upon the water chemistry during the separation but is

typically optimal between 10.5 and 11. These findings have shown the importance of

249
certain water chemistry and surface chemistry parameters on the hematite beneficiation

process.

Various dispersants were also tested at a laboratory and pilot scale for their effectiveness

in the selective flocculation and dispersion deslime process. It was found that the

polyphosphate (glass-H, sodium henicosapolyphosphate) dispersant yielded both high

iron recovery and phosphorus rejection. Polyacrylic acid dispersants yielded the highest

phosphorus rejection of tested dispersants and have the potential to replace

polyphosphates in the full scale plant in the near future.

250
Appendix A – Zeta Potential Analyses from Plant Scale Studies

Figure A.1: Zeta potential as a function of pH for screen underflow solids at process

conditions for each sample set

251
Figure A.2: Zeta potential as a function of pH for screen underflow with Glass-H solids

at process conditions for each sample set

252
Figure A.3: Zeta potential as a function of pH for cyclone overflow solids at process

conditions for each sample set

253
Figure A.4: Zeta potential as a function of pH for cyclone overflow with starch solids at

process conditions for each sample set

254
Figure A.5: Zeta potential as a function of pH for deslime thickener overflow solids at

process conditions for each sample set

255
Figure A.6: Zeta potential as a function of pH for deslime thickener underflow solids at

process conditions for each sample set

256
Figure A.7: Zeta potential as a function of pH for flotation feed solids at process

conditions for each sample set

257
Figure A.8: Zeta potential as a function of pH for scavenger returns solids at process

conditions for each sample set

258
Figure A.9: Zeta potential as a function of pH for scavenger tailings solids at process

conditions for each sample set

259
Figure A.10: Zeta potential as a function of pH for rougher concentrate solids at process

conditions for each sample set

260
Figure A.11: Zeta potential as a function of pH for concentrate thickener underflow

solids at process conditions for each sample set

261
Figure A.12: Zeta potential as a function of pH for filter feed solids at process conditions

for each sample set. Note: filter feed sample was not taken on May 5th, 2011.

262
Appendix B - Thermogravimetric Analysis for Plant Scale

Studies

Figure B.1: Thermolysis curves for screen underflow solids. Thermal decomposition

prior to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

263
Figure B.2: Thermolysis curves for screen underflow with Glass-H solids. Thermal

decomposition prior to 200°C is primarily due to moisture and loosely bound water at

mineral surfaces. Decomposition near 300°C is due to the dehydration of goethite into

hematite. Decomposition from 400 to 700°C is due to the decarbonation of dolomite,

ankerite, calcite and siderite.

264
Figure B.3: Thermolysis curves for cyclone overflow solids. Thermal decomposition

prior to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

265
Figure B.4: Thermolysis curves for cyclone overflow with starch solids. Thermal

decomposition prior to 200°C is primarily due to moisture and loosely bound water at

mineral surfaces. Decomposition near 300°C is due to the dehydration of goethite into

hematite. Decomposition from 400 to 700°C is due to the decarbonation of dolomite,

ankerite, calcite and siderite.

266
Figure B.5: Thermolysis curves for deslime thickener overflow solids. Thermal

decomposition prior to 200°C is primarily due to moisture and loosely bound water at

mineral surfaces. Decomposition near 300°C is due to the dehydration of goethite into

hematite. Decomposition from 400 to 700°C is due to the decarbonation of dolomite,

ankerite, calcite and siderite.

267
Figure B.6: Thermolysis curves for deslime thickener underflow solids. Thermal

decomposition prior to 200°C is primarily due to moisture and loosely bound water at

mineral surfaces. Decomposition near 300°C is due to the dehydration of goethite into

hematite. Decomposition from 400 to 700°C is due to the decarbonation of dolomite,

ankerite, calcite and siderite.

268
Figure B.7: Thermolysis curves for flotation feed solids. Thermal decomposition prior

to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

269
Figure B.8: Thermolysis curves for scavenger returns solids. Thermal decomposition

prior to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

270
Figure B.9: Thermolysis curves for scavenger tailings solids. Thermal decomposition

prior to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

271
Figure B.10: Thermolysis curves for rougher concentrate solids. Thermal decomposition

prior to 200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite.

272
Figure B.11: Thermolysis curves for concentrate thickener underflow solids. Thermal

decomposition prior to 200°C is primarily due to moisture and loosely bound water at

mineral surfaces. Decomposition near 300°C is due to the dehydration of goethite into

hematite. Decomposition from 400 to 700°C is due to the decarbonation of dolomite,

ankerite, calcite and siderite.

273
Figure B.12: Thermolysis curves for filter feed solids. Thermal decomposition prior to

200°C is primarily due to moisture and loosely bound water at mineral surfaces.

Decomposition near 300°C is due to the dehydration of goethite into hematite.

Decomposition from 400 to 700°C is due to the decarbonation of dolomite, ankerite,

calcite and siderite. Note: filter feed sample was not taken on May 5th, 2011.

274
Appendix C – Particle Size Analyses for Plant Scale Studies

Figure C.1: Particle size distribution of screen underflow solids for each sample set.

275
Figure C.2: Particle size distribution of screen underflow with Glass-H solids for each

sample set.

276
Figure C.3: Particle size distribution of cyclone overflow solids for each sample set.

277
Figure C.4: Particle size distribution of cyclone overflow with starch solids for each

sample set.

278
Figure C.5: Particle size distribution of deslime overflow solids for each sample set.

279
Figure C.6: Particle size distribution of deslime underflow solids for each sample set.

280
Figure C.7: Particle size distribution of flotation feed solids for each sample set.

281
Figure C.8: Particle size distribution of scavenger returns solids for each sample set.

282
Figure C.9: Particle size distribution of scavenger tailings solids for each sample set.

283
Figure C.10: Particle size distribution of rougher concentrate solids for each sample set.

284
Figure C.11: Particle size distribution of concentrate thickener underflow solids for each

sample set.

285
Figure C.12: Particle size distribution of Filter Feed solids for each sample set.

286

You might also like