Yu Dawei 201406 PHD Thesis-1
Yu Dawei 201406 PHD Thesis-1
Yu Dawei 201406 PHD Thesis-1
by
Dawei Yu
Dawei Yu
Doctor of Philosophy
2014
Abstract
Selective oxidation and sulfation roasting of nickel concentrate followed by leaching was
investigated as a novel route for nickel production. In the oxidation roasting stage, the iron
species in the nickel concentrate was preferentially oxidized to form iron oxides, leaving non-
ferrous metals (Ni, Cu, Co) as sulfides. The roasted product was then sulfation roasted to
convert the sulfides of the latter metals into water-soluble sulfates. The sulfates were then
leached into solution for further recovery and separation from iron oxides.
The oxidation of nickel concentrate was firstly studied by means of thermogravimetric and
differential thermal analysis over a wide temperature range. A reaction scheme was deduced, in
which preferential oxidation of iron sulfide species occurred over a wide temperature range up
to about 700 ºC, forming a Ni1-xS core with iron oxide shell. A batch fluidized bed roaster was
then constructed to study the oxidation and sulfation roasting of nickel sulfide concentrate.
Oxidation roasting tests were carried out at temperatures between 650 °C and 775 °C. It was
found that low temperatures (e.g. 650 °C) are favorable for the preferential oxidation of iron
sulfide species while minimizing the formation of nickeliferous oxides, i.e. trevorite and NiO.
Several parameters were varied in the sulfation roasting experiments, including the sulfation gas
ii
flowrate, sulfation roasting temperature, the addition of Na2SO4, sulfation roasting time, and the
oxidation roasting temperature. Under optimized conditions of sulfation gas composition (95%
air, 5% SO2), temperature (700 °C), Na2SO4 addition (10 wt%) and time (150 min), the
conversions to sulfates were 79% Ni, 91% Cu, and 91% Co. Only 5% Fe forms water-soluble
sulfate. The residue from the leaching of calcine in water contained 49% Fe and 10% Ni, which
is a suitable feedstock for the production of ferronickel alloys. Therefore, further studies were
also conducted to evaluate the reduction behavior of the residue with CO, H2 and graphite.
iii
Acknowledgments
I would like to express my deepest gratitude to the late Professor Torstein A. Utigard for
introducing me to this exciting topic and his continuous supervision even after he was diagnosed
with cancer. I am always grateful that he offered me the opportunity to pursue the Ph.D. degree
after I studied on this project for a year as a MASc candidate.
I would like to thank the rest of my thesis committee: Professor Doug D. Perovic, Professor
Charles Q. Jia, Professor Ramamritham Sridhar and Professor Christopher A. Pickles for their
insightful and constructive comments and advice.
I would like to thank George Kretschmann and Yanan Liu of the Geology Department of the
University of Toronto for their assistance in the characterization of some of the samples. The
assistance from Mingqian Zhu during the summer of 2012 is gratefully acknowledged. I would
also like to thank my colleague and friend Mark Li for the stimulating discussions and support.
I wish to thank Xstrata Process Support (Sudbury, Ontario, Canada) for providing the nickel
concentrate and both Xstrata and Vale for sponsoring the project. The funding for this research
was provided by the Natural Sciences and Engineering Research Council (NSERC) of Canada
and the Centre for Chemical Process Metallurgy.
Last but not least, I like to thank my mother Aihua Liu and my father Haijiang Yu for their love,
encouragement and support.
iv
Table of Contents
Acknowledgments .......................................................................................................................... iv
1 Introduction ................................................................................................................................. 1
1.4.4 Heat Transfer between a Fluidized Bed and an Immersed Surface ....................... 14
1.7 References.......................................................................................................................... 24
2.1 Introduction........................................................................................................................ 27
v
2.2.1 Sample ................................................................................................................... 31
2.5 References.......................................................................................................................... 61
3.1 Introduction........................................................................................................................ 64
3.5 References.......................................................................................................................... 86
4.1 Introduction........................................................................................................................ 89
Appendix 01: C Code for the Calculation of Enthalpy Change during the Roasting of the
Raglan Concentrate as a Function of Temperature.......................................................... 220
Appendix 02: C Code for the Calculation of the Heat Transfer Rate through the Quartz
Tube from the Air in the Electric Furnace to the Fluidized Bed ..................................... 227
Appendix 03: Photos of the Fluidized Bed Experimental Setup ............................................ 232
viii
List of Tables
Table 1.1. SO2 production and disposition for nickel flash smelters............................................. 6
Table 1.2. SO2 production and disposition for nickel electric furnace smelters............................ 6
Table 1.3. Summary of 2004 energy consumption and emissions for the Xstrata nickel smelter in
Sudbury and Vale smelter in Thompson. ...................................................................................... 7
Table 1.4. Constants for evaluating drag coefficient for three flow regimes. ............................. 13
Table 1.5. Volume changes for oxidation and sulfate formation reactions. ................................ 19
Table 2.1 Various reactions occurring during the oxidation of pentlandite at 10 ºC/min in an
air/oxygen flow of 0.2 L/min. ...................................................................................................... 28
Table 2.3. Reaction scheme for the oxidation of iron sulfide (Fe1-xS). ....................................... 30
Table 2.4. Chemical composition and estimated mineral contents of the Raglan concentrate. .. 32
Table 2.5. EPMA compositional analysis of individual grains of the Raglan concentrate and
their calculated stoichiometry, which shows the composition difference between particles of the
same mineral. ............................................................................................................................... 33
Table 2.7. Summary of the reaction sequence for the oxidation of the Raglan concentrate at 15
ºC/min in air. ................................................................................................................................ 57
Table 3.1. Activation energies for the acid dissolution of some minerals................................... 66
Table 5.1. Sulfate formation (%) and uncertainty limits for sulfation roasting tests under
optimized conditions.................................................................................................................. 146
ix
Table 6.1. Chemical and mineralogical compositions of the leach residue............................... 161
Table 6.2. Overall reaction rate expressions for different temperature ranges. ......................... 180
Table 7.1. Mass and heat balance for the oxidation roasting of 100 kg Raglan concentrate. ... 210
Table 7.2. Mass and heat balance for the sulfation roasting stage. ........................................... 211
Table 7.3. Mass and heat balance for the reduction of the leach residue with CO.................... 213
x
List of Figures
Figure 1.1. Schematic of two routes for producing a Ni-rich matte from nickel sulfide
concentrate. .................................................................................................................................... 2
Figure 1.6. Gibbs free energy of formation vs. temperature for oxides and sulfides calculated
using HSC Chemistry. ................................................................................................................. 16
Figure 1.7. Superimposed Fe-S-O predominance diagrams for 500 ºC, 700 ºC and 900 ºC
calculated using HSC Chemistry. ................................................................................................ 17
Figure 1.8. Superimposed Ni-S-O predominance diagrams for 500 oC, 700 oC and 900 oC
calculated using HSC Chemistry. ................................................................................................ 17
Figure 1.9. Shrinking core model for the oxidation roasting of a sulfide particle. ...................... 18
Figure 1.11. Predominance area diagram by superimposing Fe-S-O, Ni-S-O, Cu-S-O and Co-S-
O predominance area diagrams at 680 °C (calculated using data from HSC Chemistry). .......... 21
Figure 2.2. BSE image of the Raglan concentrate (pn: pentlandite; cpy: chalcopyrite; po:
pyrrhotite; py: pyrite; flux: silicate flux). .................................................................................... 32
Figure 2.6. Repeated tests for the determination of the melting temperature of Ag by DTA. .... 37
Figure 2.7 Sample mass change, rate of mass change, SO2 concentration and O2 consumption in
the offgas for the TGA run in which 100 mg concentrate was heated to 950 ºC at 15 ºC/min in
air. ................................................................................................................................................ 39
Figure 2.8. DTA curve of Raglan concentrate heated at 15 ºC/min in air. .................................. 40
Figure 2.9. DTA curve of Raglan concentrate heated at 15 ºC/min in O2. .................................. 40
Figure 2.10. XRD patterns for calcines quenched from intermediate temperatures after heating at
15 ºC/min in air in TGA runs. ..................................................................................................... 41
Figure 2.11. Chemical analysis results for the contents of the water soluble sulfates in the
calcines quenched from intermediate temperatures after heating at 15 ºC/min in air in TGA runs.
..................................................................................................................................................... 42
Figure 2.12. Fe-Ni-S diagram showing the change of chemical compositions during non-
isothermal heating in TGA runs analyzed by EPMA. The bottom part is the magnified area of
the trapezoid in the top ternary diagram. ..................................................................................... 42
Figure 2.13. SE image of an oxidized pyrrhotite particle quenched from 605 ºC in the TGA run.
..................................................................................................................................................... 45
Figure 2.14. SE image of an oxidized pyrrhotite particle quenched from 733 ºC in the TGA run
showing its characteristic columnar structure.............................................................................. 45
Figure 2.15. SE image of an oxidized pyrrhotite particle quenched from 880 ºC in the TGA run.
..................................................................................................................................................... 46
Figure 2.16. BSE image of a chalcopyrite particle after heating the Raglan concentrate in air at
450 ºC for 1 hour (Cpy: chalcopyrite, CuFeS2; Bor: bornite, Cu5FeS4). ..................................... 47
xii
Figure 2.17. BSE image of a partly oxidized pentlandite particle air-quenched from 733 ºC in
the TGA run. ................................................................................................................................ 50
Figure 2.18. XRD patterns for calcines quenched from intermediate temperatures after heating at
15 ºC/min in air in DTA runs. ..................................................................................................... 51
Figure 2.19. BSE image of an oxidized pentlandite particle air-quenched from 880 ºC in the
TGA run. Textures of the sulfide core are exhibited with enhanced contrast. ............................ 53
Figure 2.20. BSE image of an oxidized pentlandite particle air-quenched from 880 ºC in the
TGA run. ...................................................................................................................................... 53
Figure 2.21. BSE image of an oxidized pentlandite particle air-quenched from 950 ºC in the
TGA run. ...................................................................................................................................... 54
Figure 2.22. BSE image of a particle composed of a mixture of NiSO4 and MgSO4 quenched
from 880 ºC.................................................................................................................................. 56
Figure 2.23. Mass change in wt% vs. time with the variation of sample size. ............................ 59
Figure 2.24. Initial rate of mass change vs. sample size and bed thickness. ............................... 59
Figure 3.1. Experimental setup for the roasting of nickel concentrate. ....................................... 69
Figure 3.2. Temperature and SO2 concentration in the offgas during roasting of sample
Calcine650. ................................................................................................................................. 71
Figure 3.3. BSE image and elemental maps of Fe, Ni, and S of sample Calcine650. ................ 72
Figure 3.4. Temperature and SO2 concentration during roasting of sample Calcine650S. ........ 73
Figure 3.5. Temperature and SO2 concentration during roasting of sample Calcine750............ 73
Figure 3.7. Mass change, temperature and the SO2 concentration in the TGA test for sample
Calcine650. ................................................................................................................................. 75
xiii
Figure 3.8. BSE and optical images of the leach residue from the DAL of Calcine650. ........... 77
Figure 3.10. Concentrated HCl acid leaching behavior of sample Calcine650. ......................... 79
Figure 3.11. BSE image and elemental maps of Fe, Ni, and S of the leach residue from the CAL
of Calcine650. ............................................................................................................................. 81
Figure 3.12. XRD pattern for the residue from the CAL of Calcine650. ................................... 82
Figure 3.13. Schematic representation of the oxidative dissolution of Ni3S2 in HCl solution with
the presence of Fe3+. .................................................................................................................... 82
Figure 3.14. Optical and elemental mapping images showing the morphological features of the
nickel sulfide core covered with elemental sulfur rim................................................................. 83
Figure 3.15. Dilute HCl acid leaching results of the sample Calcine650S based on the ICP
analysis. ....................................................................................................................................... 84
Figure 3.16. Dilute HCl acid leaching results of sample Calcine750. ........................................ 85
Figure 4.1. Schematic of the batch-wise fluidized bed experimental setup. ............................... 90
Figure 4.3. Calculated minimum fluidization velocity as a function of particle size and
temperature for the roasting of 20 g sand + 5 g Raglan concentrate using air. ........................... 96
Figure 4.4. Pressure drop method for the determination of Vmf. ................................................. 97
Figure 4.5. Pressure drop across the bed vs. gas velocity............................................................ 98
Figure 4.6. Terminal velocity vs. particle size and temperature for the roasting of Raglan
concentrate with air.................................................................................................................... 100
Figure 4.7. Apparent terminal velocity (25 ºC) vs. particle size and temperature for the roasting
of Raglan concentrate using air. ................................................................................................ 101
xiv
Figure 4.8. Temperature profile above the fluidized bed. ......................................................... 101
Figure 4.9. Fluidized bed roasting test without cooling. ........................................................... 103
Figure 4.10. Images showing various stages of fluidized bed oxidation roasting test. ............. 104
Figure 4.11. Yellowish elemental sulfur particles formed on the upper wall of the quartz tube
from the condensation of the sulfur vapor. ................................................................................ 104
Figure 4.12. SO2 concentration right above the fluidized bed calculated from the measured SO2
concentration.............................................................................................................................. 107
Figure 4.13. Longitudinal and cross sections of the fluidized bed system showing the
temperature gradient. ................................................................................................................. 108
Figure 4.14. Flow chart for the calculation of the heat transfer rate through the wall of quartz
tube (please refer to Figure 4.13 and Appendix 02 for the meanings of the terms). ................. 109
Figure 4.15. Heat transfer coefficients (h) of the fluidized bed and the air around the quartz tube,
and the thermal conductivity (k) of the quartz tube as a function of the fluidized bed
temperature. ............................................................................................................................... 111
Figure 4.16. Temperature differences (in percentage) between the main body of the fluidized
bed and the inner wall of the quartz tube (labeled as Fluidized bed), between the inner wall and
outer wall of the quartz tube (labeled as Quartz tube), and between the outer wall and the
furnace wall (labeled as Furnace air), against fluidized bed main temperature, shown as the
stacked areas. Conductive heat transfer rate through the quartz tube is plotted as the blue curve.
................................................................................................................................................... 112
Figure 4.17. Temperature profile and the heat transfer rates by reactions, radiation, feeding air
and conduction through the quartz wall..................................................................................... 113
Figure 4.18. Temperatures measured at three locations in the fluidized bed. ........................... 115
Figure 4.19. Temperature, offgas SO2 and O2 concentrations, and pressure drop for roasting test
at 650 °C. ................................................................................................................................... 117
xv
Figure 4.20. Offgas SO2 profiles for the fluidized bed roasting tests by air with roasting
temperature variation. ................................................................................................................ 117
Figure 4.21. XRD patterns of the calcines for various temperatures (tv-trevorite, NiFe2O4; mss-
monosulfide solid solution, (Ni,Fe)1-xS; h-hematite, Fe2O3; NiO-nickel oxide; hz-heazlewoodite,
Ni3S2). ........................................................................................................................................ 119
Figure 4.22. Roasted pentlandite particles at 650 °C with 3 L/min air (mss: monosulfide solid
solution). .................................................................................................................................... 120
Figure 4.23. Concentration profiles in a cross section of a roasted pentlandite particle (650 °C)
along the arrow in Figure 4.22 measured by EPMA. ................................................................ 120
Figure 4.24. Degree of sulfur elimination as a function of roasting temperature. ..................... 121
Figure 4.25. Change of the pentlandite sulfide core compositions as a function of fluidized bed
roasting temperatures. ................................................................................................................ 122
Figure 4.26. Average Co contents of the roasted pentlandite sulfide cores as a function of the
fluidized bed roasting temperatures by EPMA. ......................................................................... 123
Figure 4.27. Water soluble species (%) in the roasted calcines vs. roasting temperature. ........ 124
Figure 4.28. XRD patterns for the fluidized bed roasting tests at 750 °C with roasting time
variation (tv-trevorite, mss-monosulfide solid solution, h-hematite, NiO-nickel oxide, hz-
heazlewoodite). .......................................................................................................................... 125
Figure 4.29. Degree of sulfur elimination as a function of roasting time at 750 °C. ................ 126
Figure 4.30. Fe-(Ni+Co)-S ternary diagram shows the composition change of the pentlandite
sulfide cores as a function of fluidized bed roasting time measured by EPMA. ....................... 127
Figure 4.31. Average Co concentrations of the roasted pentlandite sulfide cores as a function of
the fluidized bed roasting time measured by EPMA. ................................................................ 127
Figure 4.32. Water soluble species (wt%) in the roasted calcines vs. roasting time at 750 °C. 128
xvi
Figure 5.1. Effect of the sulfation roasting gas flowrate on the formation of sulfates. ............. 134
Figure 5.2. Sulfation roasted pentlandite particles with 1 L/min sulfation gas, exhibiting the
formation of thin NiSO4 layers on the nickel sulfide cores, and the formation of nickel iron
oxide. ......................................................................................................................................... 135
Figure 5.3. Leach residue of the sulfation roasted calcine with sulfation gas flowrate of 1 L/min.
................................................................................................................................................... 135
Figure 5.4. Effect of the sulfation roasting temperature on the formation of sulfates. .............. 137
Figure 5.5. Effect of the addition of Na2SO4 on the formation of sulfates. ............................... 138
Figure 5.6. SEM images of the oxidation roasted calcine blended with 10 wt% Na2SO4 (top-
left), and sulfation roasted calcines with the addition of Na2SO4 of 2% (top-right), 5% (bottom-
left), and 10% (bottom-right) with regard to the weight of the calcine (mss: monosulfide solid
solution). .................................................................................................................................... 139
Figure 5.7. Incomplete (left) and complete (right) conversion of nickel sulfide cores to NiSO4
with the addition of Na2SO4. ..................................................................................................... 140
Figure 5.8. Cluster of calcine particles agglomerated by sulfate mixtures with 5% Na2SO4
addition. ..................................................................................................................................... 140
Figure 5.9. Effect of the sulfation roasting time on the formation of sulfates........................... 141
Figure 5.10. Calculated relative amount of Ni existing as sulfide and oxide (%) in the leach
residue against sulfation roasting time. ..................................................................................... 142
Figure 5.11. Partly oxidation roasted pentlandite particles under 500 °C, 550 °C and 600 °C. 143
Figure 5.12. Effect of the temperature of the oxidation roasting stage on the formation of
sulfates in the sulfation roasting stage at 700 °C. ...................................................................... 144
Figure 5.13. Sulfation roasted pentlandite particles: single stage sulfation roasting at 700 °C
(top-left); two-stage roasting with the oxidation roasting temperature of 500 °C (top-right), 550
°C (bottom-left) and 600 °C (bottom-right). ............................................................................. 145
xvii
Figure 5.14. Repeated sulfation roasting tests conducted under optimized conditions. ............ 146
Figure 5.15. Sulfation roasted pentlandite particle with 20% addition of Na2SO4 at 700 °C for
30min. The sulfide core was isolated and shown in the top-right corner with enhanced contrast
illustrating the presence of two sulfide phases. ......................................................................... 148
Figure 5.17. Nickel sulfide samples uncoated (top-left) and coated with Na2SO4 (the other
three), sulfation roasted at 700 °C for 30 min. .......................................................................... 151
Figure 5.18. ICP-MS measurements of some of the precious metals in the Raglan concentrate by
line scan. .................................................................................................................................... 154
Figure 5.19. Four (4) repeated measurements of the concentrations of some precious metal
isotopes by ICP-MS in three samples: Raglan concentrate (black square); sulfation roasted
calcine (red cross); and leach residue of the sulfation roasted calcine (blue star). Averages of the
repeated measurements are shown in the diagonal band for each element and each sample. ... 155
Figure 6.1. Stability diagram for Fe-O system under CO-CO2 atmospheres (solid lines) and CO
partial pressure established by the Boudouard reaction (dashed line) (Calculated using
thermodynamic data from HSC Chemistry). ............................................................................. 160
Figure 6.3. Modification of the TGA setup by mounting graphite tubes on the ceramic tube. . 164
Figure 6.4. CO and CO2 evolution during the isothermal tests at 800 ºC with empty crucible
(#1), 20 mg graphite powder in the crucible (#2), and a mixture of 20 mg graphite powder and
50 mg residue in the crucible (#3), after the modification of the TGA setup............................ 165
Figure 6.5. TGA/DTA results of the continuous heating tests at 15 ºC/min for the reduction with
15% H2. ...................................................................................................................................... 167
Figure 6.6. Isothermal H2 (15%) reduction of the leach residue at 350 ºC, 400 ºC, 450 ºC and
500 ºC. ....................................................................................................................................... 169
xviii
Figure 6.7. Mass change (wt%) during the isothermal reduction of the leach residue by 15% H2
at various temperatures. ............................................................................................................. 169
Figure 6.8. BSE image of the residue partly reduced (22%) at 400 ºC with 15% H2. .............. 170
Figure 6.9. Time required to reach 90% reduction of the leach residue with 15% H2 as a function
of temperature. ........................................................................................................................... 171
Figure 6.10. Linear relations between the rate of reaction and the differential form of the 2D
diffusion model which is f(α)=[-ln(1-α)]-1 for the isothermal tests from 600 ºC to 1200 ºC. ... 172
Figure 6.11. Arrhenius plot for the isothermal reduction tests from 350 ºC to 1500 ºC (Data for
1200 ºC was calculated based on the initial reduction period). ................................................. 173
Figure 6.12. Morphology of the fully reduced residue (900 ºC with 15% H2). ......................... 174
Figure 6.13. XRD pattern for the fully reduced residue (900 ºC with 15% H2). ....................... 175
Figure 6.14. Effect of gas flowrate on the isothermal reduction of the leach residue at 800 ºC
(15% H2). ................................................................................................................................... 175
Figure 6.15. Effect of sample mass on the isothermal reduction of the leach residue at 800 ºC.
................................................................................................................................................... 176
Figure 6.16. Morphology of the residue fully reduced at 1100 ºC with 15% H2. ..................... 177
Figure 6.17. BSE image of the fully reduced residue from the isothermal test at 1200 ºC with
15% H2 (white area is the ferronickel alloy, grey areas are silicates with varying Fe contents).
................................................................................................................................................... 178
Figure 6.18. Comparison of the continuous heating reduction with 15% H2 and 40% H2 at 15
ºC/min. ....................................................................................................................................... 179
Figure 6.19. Comparison of the isothermal reduction with 15% H2 and 40% H2 at 800ºC. ..... 180
Figure 6.20. Continuous heating (15 ºC/min) of the leach residue in CO. ................................ 181
xix
Figure 6.21. XRD patterns for the leach residue (a), and samples collected after continuous
heating in CO to 520 ºC (b) and 800 ºC (c). .............................................................................. 183
Figure 6.22. TGA isothermal reduction of the leach residue with CO. ..................................... 185
Figure 6.23. Variation of apparent activation energies as a function of extent of reduction (α) for
the isothermal reduction tests. lnt vs.1/T ×104 (isoconversional method) is also plotted as an
inset. ........................................................................................................................................... 187
Figure 6.24. Effect of the variation of the flowrates of CO on the isothermal reduction of the
leach residue. ............................................................................................................................. 188
Figure 6.25. Effect of the variation of sample sizes on the isothermal reduction of the leach
residue. ....................................................................................................................................... 188
Figure 6.26. Arrhenius plot for the isothermal reduction tests between 500 and 1100 ºC. The
relationship between the reduction rate and the 2D diffusion model (f(α)=[-ln(1-α)]-1) is also
plotted as an inset....................................................................................................................... 190
Figure 6.27. Isothermal reduction of the porous hematite particles with CO at various
temperatures............................................................................................................................... 191
Figure 6.29. Laminar structure of the silicates formed upon cooling the reduced residue from
1300ºC with a gas bubble formed in the melt (left) and interior of the bubble (right). ............. 194
Figure 6.30. Continuous heating of the leach residue with graphite in TGA and DTA. ........... 196
Figure 6.31. TGA isothermal reduction of the leach residue with graphite. ............................. 197
Figure 6.32. EPMA analysis on the alloy particles formed from the isothermal reduction tests by
graphite. ..................................................................................................................................... 198
Figure 6.33. Partial conversion of the nickel sulfide to alloy at 900 ºC. ................................... 199
xx
Figure 6.34. Alloy surrounded by monosulfide solid solution (Mss, (Ni,Fe)S) reduced at 1000
ºC (Px: Pyroxene, silicate containing Fe, Al). ........................................................................... 199
Figure 6.35. The composition of silicate phases in isothermal reduction by graphite. ............. 200
Figure 6.36. Formation of SixFe alloy particles from the silicate melt at 1200 ºC (Px: Pyroxene,
silicate containing Al, Mg, Fe). ................................................................................................. 201
Figure 6.37. Reduction of Fe and Si from a silicate particle forming Ni-Si-Fe alloy at 1200 ºC
(Px: Pyroxene, silicate containing Al, Mg, Na). ........................................................................ 201
Figure 6.38. Temperatures at which ΔG=0 for the reduction of FeSiO4 by graphite as a function
of the molar ratio of Si/(Fe+Si) of the alloy product. ................................................................ 202
Figure 6.40. An NixSiyFe alloy particle formed from the reduction at 1400 ºC. ....................... 204
Figure 6.41. XRD analysis on the product from the reduction at 1000 ºC, 1200 ºC and 1400 ºC.
................................................................................................................................................... 204
Figure 8.1. Proposed flow sheet for the selective oxidation and sulfation roasting of nickel
concentrate. ................................................................................................................................ 218
Figure A 2. Sample feeding and collection systems of the fluidized bed setup.........................233
Figure A 4. Interior of the brass cap. The gas outlet is covered with a piece of stainless steel
mesh to filter the dust from the offgas. A thin steel blade is fixed on the rotatable ceramic tube
by cement. During roasting, the dust accumulated on the mesh is periodically scraped by the
rotating blade and falls into the fluidized bed............................................................................235
Figure A 5. Upper part of the clear quartz tube showing the SO3 fume formed during roasting
and the condensed SO3 on the inner wall...................................................................................236
xxi
Figure A 6. Interior of the muffle furnace..................................................................................237
xxii
List of Appendices
Appendix 01: C Code for the Calculation of Enthalpy Change during the Roasting of the Raglan
Concentrate as a Function of Temperature…………………………………….………………220
Appendix 02: C Code for the Calculation of the Heat Transfer Rate through the Quartz Tube
from the Air in the Electric Furnace to the Fluidized Bed……………………………………..227
There are two routes for the processing of the nickel sulfide concentrate, namely flash smelting
and electric furnace smelting, as schematically shown in Figure 1.1 [3]. These two processing
routes share the same principle: 1) partial sulfur removal by oxidizing gas (air, oxygen-rich air
or industrial oxygen) forming gaseous SO2; 2) oxidation of Fe forming iron oxides which report
1
to the slag phase along with the gangue materials by combining with the siliceous flux, and 3)
separation of the matte from the slag due to their immiscible nature and the density difference.
The product from the smelting process is a Ni-rich and Fe-lean matte, which requires further
refining.
Figure 1.1. Schematic of two routes for producing a Ni-rich matte from nickel sulfide
concentrate [3].
The electric furnace smelting route to treat nickel concentrate comprises the following steps:
roasting, smelting and converting. Partial oxidation of the concentrate is performed in a
fluidized bed roaster to form iron oxide and nickel sulfide, the oxidation reactions being
represented by Reactions (1.1) and (1.2). In the oxidative atmosphere, a certain amount of Ni is
also oxidized to nickel oxide (NiO) and nickel ferrite (NixFe3-xO4). Separation of the iron oxide
from the sulfide is achieved by melting the calcine and fluxing with siliceous materials in the
electric furnace under a reducing atmosphere. The main reactions occurring in the electric
furnace are represented by Reactions (1.3) to (1.5). The recovery of Ni can be as high as 98%
2
due to the reducing environment in the electric furnace [3]. Converting is usually performed in
the Pierce-Smith converter to further reduce the Fe content in the electric furnace matte to 1–4
wt% as well as the sulfur content by blowing air or oxygen-enriched air.
2C + O2 = 2CO (1.3)
Flash smelting combines roasting and smelting in one smelter. Compared with electric furnace
smelting, it greatly reduces the fuel or electricity consumption by utilizing the heat released
from the oxidation of iron and sulfur in the concentrate. However, the nickel loss to the slag is
higher due to the more oxidizing atmosphere employed in the flash furnace, necessitating further
slag cleaning to recover the entrapped and oxidized Ni. Converting is also an integral part of the
flash smelting route to reduce the iron and sulfur contents of the matte.
3
Other industrial Electric power
sources generation
35% 23% Marine
transportation
4%
Non-ferrous
mining and smelting Other transportation
34% sources
Other sources 2%
2%
As for the nickel industry, sulfur rejection in the pyrometallurgical routes of nickel extraction
always involves gaseous SO2 as a by-product. SO2 fixation is necessary in order to meet the
more stringent environmental regulations. This is performed by converting it to sulfuric acid
(H2SO4) in the acid plant. The sulfuric acid plant requires the optimum SO2 concentration of the
gas input to be between 10% and 12% [3]. Generally speaking, the flash furnace and the
fluidized bed roaster would produce a continuous offgas stream with suitable SO2 concentration
for efficient production of sulfuric acid. On the other hand, the SO2 content in the offgas from
an electric furnace is too weak for this purpose. Usually this offgas is emitted to the atmosphere
without SO2 capture. The Pierce-Smith converter offgas is discontinuous and relatively weak,
which makes it difficult to capture. SO2 can be emitted to the environment during charging and
skimming, as well as from the interface between the hood and the converter during blowing [3].
The dependence of the SO2 disposition method on the SO2 concentration of the offgas can be
revealed by the nickel smelters’ practices worldwide. Table 1.1 and Table 1.2 provide a
summary of the SO2 production and its disposal in nickel smelters of both the flash smelting and
the electric furnace smelting routes [5]. These 11 smelters investigated and listed in these two
tables account for approximately 55.5% of the annual world primary Ni production in the year
4
2004, and the balance was mainly produced by the laterite processing plants. Based on the data
presented here, the amount of SO2 emitted to the atmosphere was some 900,000 tonnes per year
from the flash smelters, and 300,000 tonnes per year from the electric furnace smelters,
assuming the smelters operate 90% of the time. About 53% of the SO2 produced from the
smelting stage in the flash smelters was captured to make acid, while the value was only 8%
captured from the converting stage. As for the electric furnace smelters, only 33% of the SO2
produced from the roasting stage was captured for acid making, which is mainly due to the weak
SO2 produced by the traveling grates employed in the two nickel smelters in Russia. Although
the Vale nickel smelter in Thompson, Manitoba, Canada produces offgas with strong SO2 from
the fluidized bed roaster which is suitable for acid making, it is vented to the atmosphere
through stacks instead due mainly to the high cost of transportation of sulfuric acid to the
market. This makes it difficult for the company to meet the environmental regulations. Because
of this situation, as well as a lack of the raw materials, Vale will shut down the smelter in
Thompson in the year 2015 [7]. As for the Pierce-Smith converters in the electric furnace
smelters, 47% of the SO2 produced from them was fixed as sulfuric acid. This number is
relatively high compared to that of the flash smelters, which is mainly attributed to the capture
of SO2 from the converters for acid making practiced by the Pechenganickel smelter in Russia,
although its concentration in the offgas is very low.
Apart from the SO2 issues, the electric furnace smelters are also large consumers of electricity,
which is revealed in Table 1.3. This is a major drawback of the electric furnace smelters
compared with the flash smelters in which heat released from the exothermic oxidation of the
concentrate is utilized to melt the feed. The electric furnace uses consumable carbon paste
electrodes to conduct current, and create a more reducing environment with the addition of coke.
Substantial amounts of CO2 are emitted as a result, which can be seen in Table 1.3. Also a large
amount of heavy metals are emitted to the atmosphere from the smelters.
5
Table 1.1. SO2 production and disposition for nickel flash smelters [5].
DON Flash Smelters
Nadezda Inco Oxygen
Kalgoorlie, Jinchuan, Matallurgical Harjavalta, Flash Smelter,
BCL Smelter, Fortaleza
BHP Billiton Jinchuan Group Plant, Boliden Copper Cliff,
Plant BCL Limited, de Minas,
Nickel West, Ltd. Norilsky Harjavalta CVRD Inco,
Selebi Phikwe, Votoranti
Kalgoorlie, Gansu Nickel, Oy (Smelter), Sudbury,
Botswana m Metais,
Australia Province, China Norilsk, Harjavalta, Ontario, Canada
Brazil
Russia Finland
Annual Ni Production (t) 27400 100000 65000 140000 38000 7000 133400
Off-gas
Volume 87152 ______ 60000 56000 16000 13100 24000-28000
(Nm3/h)
SO2 Dry
Smelting 7.2 ______ 8 30-35 30 26 55 Dry basis
Basis (vol.%)
To liquid SO2
Off-gas
Atmosphere Acid Plant Acid Plant Atmosphere Acid Plant Acid Plant plant and to
Disposition
acid plant
80000 one 41000 to acid
50000-60000 140000 N/A N/A 140000
Off-gas converter plant
Volume 160000 two
(Nm3/h) converters in 60000 to stack ______ ______
stack
Converting 3-5% during
5.2 4% to acid plant 2.5-3.5 2.5 ______ ______
SO2 Dry regular blows
Basis (vol.%)
2.7% to stack ______ ______
Off-gas To acid plant or
To stack Acid Plant Atmosphere ______ ______ To stack
Disposition stack
Atmosphere
Slag Off-gas See flash furnace Baghouse
Atmosphere Atmosphere after dust ______ ______
Cleaning Disposition off-gas disposition and stack
recovery
Table 1.2. SO2 production and disposition for nickel electric furnace smelters [5].
Converting
Off-gas from slag-cleaning vessel, slag-making
Off-gas Disposition Stack Stack To acid Plant
converter, and finish converter to stack
6
Table 1.3. Summary of 2004 energy consumption and emissions for the Xstrata nickel smelter in
Sudbury and Vale smelter in Thompson*.
* Data estimated based on published information [5, 8, 9]. Assume electric furnaces operate
90% of the time. The furnace CO2 emissions do not include indirect emissions due to electricity
generation.
This investigation aims at developing a process where the electric furnace and the converters are
eliminated. Instead, roasting will be modified to make water-soluble nickel, copper and cobalt
sulfates so that leaching followed by various hydrometallurgical steps are used. Such a process
could potentially eliminate all the emissions and decrease the use of electric energy
dramatically. The roasting process would include two roasters in series, one oxidizing followed
by one sulfating. Both of these processes will be continuous with all gases captured and passed
through an acid plant, for nearly 100% capture of SO2 and heavy metals. The fuel value of the
sulfides will be maximized to dramatically decrease the use of electric energy and with no need
for coke or any other fossil fuel.
In the process to be investigated, iron sulfides are firstly oxidized selectively to iron oxides, and
the remaining metal sulfides are then roasted to form water soluble sulfates, thus eliminating
7
electric furnace smelting and converting. The valuable metal sulfates formed will then be
leached, purified and recovered. The process steps would be:
The main advantages are: 1) Eliminating the use of electric energy by the electric furnaces; 2)
Avoiding the use of coke as a reductant thereby reducing the CO2 emissions; 3) Significant
reduction of SO2 emissions; 4) Reducing heavy metals atmospheric emissions; 5) Decreasing
fugitive emission of gases and dust, improving work place conditions; 6) Simplified flowsheet
with lower capital cost. Since all the gases from the roasters will pass through various gas
cleaning steps and then enter a sulfuric acid plant, heavy metal emissions will also be minimal.
1) Roast Ni/Cu concentrate selectively to oxidize most of the iron sulfides to magnetite and
hematite, maintaining nearly all nickel and copper as sulfides.
2) Take the partially oxidized product from step one and sulfatize it further to form nickel,
copper and cobalt sulfates from the sulfides. The sulfation atmosphere has to be determined.
3) Leach product from step 2). The nickel, copper and cobalt sulfates are soluble while the iron
oxides do not dissolve.
4) Precipitation of any dissolved iron from the separated leach liquor by air oxidation after the
adjustment of pH and temperature.
5) Recovery of copper from the iron-free liquor by pH adjustment and liquid-liquid extraction.
6) Recovery of nickel and cobalt from the purified liquor by pH adjustment and liquid-liquid
extraction.
This thesis will be mainly focused on steps 1) and 2) since they are the main technical
challenges. Steps 3) to 7) should be fairly standard hydrometallurgical processes, except
8
possibly iron oxide filtration and removal as well as effluent treatment to meet the water
discharge regulations.
9
Figure 1.3. Fluidized bed roaster [10].
Fluidized bed technology is widely used for drying, calcining, roasting, and catalyzing, etc. Its
wide application draws a lot of interest in research in this area, resulting in a large amount of
literature. Some basic yet important aspects of the fluidized bed technique are briefly discussed
below.
10
rapid even with a few bubbles. Group B has particles with mean sizes between 40 µm and 500
µm, and with density between 1400 and 4000 kg/m3. A typical example of this group is sand.
Mixing is poor in the absence of bubbles. Group C are fine and cohesive particles, which are
difficult to fluidize. This difficulty arises because the inter-particular forces are greater than the
forces exerted by the fluid. As a result, the powder is lifted as a plug in small diameter tubes, or
channels badly. Group D is comprised of coarse (dp>500 µm) and dense particles, which is
spoutable.
11
which the pressure drop becomes relatively constant, which is because the drag force exerted on
the bed equals the gravitational force less the buoyancy.
The minimum fluidization velocity can also be theoretically predicted. For example, Thonglimp
et al. [13] developed a correlation for Remf derived from the pressure drop principles, which is
shown as Equation (1.6).
1�
𝑅𝑒𝑚𝑓 = (31.62 + 0.0425 ∗ 𝐴𝑟) 2 − 31.6 (1.6)
where Remf is the Reynolds number with minimum fluidization velocity, and Ar is the
Archimedes number. They are shown as Equations (1.7) and (1.8), respectively.
12
where dp is the diameter of the particle, Vmf is the minimum fluidization velocity, ρg is the
density of the fluid, μg is the dynamic viscosity of the fluid, ρs is the density of the particle, g is
the gravitational acceleration.
4
𝐶𝐷 𝑅𝑒𝑡 2 = 3 𝐴𝑟 (1.9)
where CD is the drag coefficient, Ret is the Reynolds number calculated with terminal velocity,
and Ar is the Archimedes number.
The drag coefficient could be determined from Re by Equation (1.10), in which a and b are
constants which are given in Table 1.4 for three flow regimes.
𝐶𝐷 = 𝑎� 𝑏 (1.10)
𝑅𝑒
Table 1.4. Constants for evaluating drag coefficient for three flow regimes [12].
13
1.4.4 Heat Transfer between a Fluidized Bed and an Immersed Surface
Heat transfer between a fluidized bed and an immersed surface is relatively fast due to the fluid-
like nature of the fluidized bed. It mainly involves the convective heat transfer associated with
the gas and the heat transfer associated with the contact between the fluidized particles and the
immersed surface [14]. Borodulya et al. [15] investigated the heat transfer between a surface and
a fluidized bed using a two-zone model. Considering the heat transfer coefficient as the sum of
the conductive (hcond), convective (hconv), and radiative (hr) components, Equation (1.11) was
obtained for computing the overall heat transfer coefficient. Thermophysical properties are
evaluated at (𝑇∞ − 𝑇𝑠 )/2.
0.14 0.24 2�
𝜌𝑝 𝐶𝑝 2 (1 − 𝜔) 3
𝑁𝑢 = 0.85𝐴𝑟 0.1
� � � � (1 − 𝜔) �3 + 0.046𝑅𝑒𝑃𝑟
𝜌𝑔 𝐶𝑔 𝜔
�
𝐷
+ 𝑘 𝜎 ∗ (𝑇∞2 + 𝑇𝑠2 )(𝑇∞ + 𝑇𝑠 ) (1.11)
𝑔
where ρp and ρg are the densities of fluidized particles and gas, respectively; Cp and Cg are the
� is the
heat capacities of the particles and gas; ω is the porosity or voidage of the fluidized bed; 𝐷
mean diameter of the particles; kg is the thermal conductivity of the gas; 𝑇∞ is the temperature
of the fluidized bed core; Ts is the temperature of the immersed surface;
�
ℎ𝐷
𝑁𝑢 = 𝑘𝑔
; where h is the overall heat transfer coefficient;
� 3 𝜌𝑔 (𝜌𝑝 −𝜌𝑔 )
𝑔𝐷
𝐴𝑟 = ;
𝜂𝑔 2
� 𝑉0
𝐷
𝑅𝑒 = ; where V0 is the superficial velocity of the gas; νg is the kinematic diffusivity of the
𝜈𝑔
gas;
𝜈𝑔
𝑃𝑟 = 𝛼 ; where 𝛼𝑔 is the thermal diffusivity of the gas;
𝑔
𝜎
𝜎∗ = 1 1 ; where σ=5.670*10-8 W·m-2·K-4; 𝜀𝑠 is the emissivity of the immersed solid; 𝜀𝑒 is
� + −1�
𝜀𝑠 𝜀𝑒
MeS(s)+2/3O2(g)→MeO(s)+SO2(g) (1.12)
(Ni,Fe)9S8+O2→Fe2O3+Ni1-xS+SO2 (1.13)
15
-50
Fe3O4
-100 CoO
NiO
-150 Cu2O
Fe7S8
∆Gf (KJ/mol O2 or S2)
-200 Ni3S2
Cu2S
-250 Co9S8
-300
-350
-400
Melting point
-450
Decomposition
Transformation
-500
0 200 400 600 800 1000 1200 1400
o
Temperature ( C)
Figure 1.6. Gibbs free energy of formation vs. temperature for oxides and sulfides calculated
using HSC Chemistry [17].
Thermodynamically, the stabilities of sulfide and oxide species depend not only on the
temperature, but also on the gas environment in the roaster. Therefore the roasting parameters
must be carefully chosen in order to achieve a good quality of the roasting product.
Predominance diagrams (Figure 1.7 and Figure 1.8) could provide guidance for the appropriate
selection of roasting conditions, such as roasting temperature and the roasting gas composition.
The roasting temperature of a fluidized bed could be adjusted by the water content of the
feeding slurry and by water cooling. The roasting gas composition could be indirectly controlled
by adjusting the gas to feed ratio as well as the feeding gas composition. Figure 1.7 and Figure
1.8 indicate that thermodynamically, sulfides are not stable under an oxidative environment at
normal roasting temperatures. Sulfates could form at relatively lower temperatures (e.g. 500 oC)
and with high partial pressure of SO2. Most probably the dead roasting product of a nickel-iron
sulfide concentrate would be Fe2O3 and NiO. A lower extent of the oxidation of NiS species
could be achieved by limiting the O2 supply, based on the theory of preferential oxidation of
iron species mentioned above.
16
8
6 Fe7S8
4 FeSO4
2 o
900 C
log pSO2(g) 0
Fe2(SO4)3 o
-2 Fe3O4 700 C
Fe2O3
-4
-6 500oC
-8
-10
-20 -15 -10 -5 0 5
log pO2(g)
Figure 1.7. Superimposed Fe-S-O predominance diagrams for 500 ºC, 700 ºC and 900 ºC
calculated using HSC Chemistry [17].
4
NiS NiSO4
2
log pSO2(g)
0
Ni3S2 900oC
-2
o
-4 700 C
NiO
-6
-8 500oC
-10
-20 -15 -10 -5 0 5
log pO2(g)
Figure 1.8. Superimposed Ni-S-O predominance diagrams for 500 oC, 700 oC and 900 oC
calculated using HSC Chemistry [17].
17
1.5.2 Kinetics
Reaction (1.12) which represents the main reactions occurring in a sulfide roasting process
indicates that solid and gas phases are involved both as reactants and as products. Figure 1.9
shows the shrinking core model for the oxidation roasting of a sulfide particle, illustrating
schematically the process of roasting. There are several steps involved in the oxidation roasting
mechanism of a sulfide mineral: 1) Oxygen diffusion through a thin gas film to the surface of
the particle; 2) Oxygen diffusion to the unreacted core (MeS) surface through the layer of solid
product (MeO); 3) Oxygen adsorption onto the surface of the sulfide core; 4) Electron transfer
from a sulfide ion to a nearby oxygen, and incorporation of the ions into the mineral lattice; 5)
Binding of the neutralized sulfur with another oxygen molecule adsorbed beside it; 6)
Desorption of the SO2 molecule and diffusion away through the MeO layer and the gas thin film
into the main gas stream, leaving a vacant sulfide ion site on the mineral surface; 7) Diffusion of
another sulfur ion from the interior of the sulfide core to occupy this site and continue the
reaction [18], or this site is occupied by another oxygen ion, forming a bond of the solid product
MeO.
Figure 1.9. Shrinking core model for the oxidation roasting of a sulfide particle.
Depending on the roasting conditions and the nature of the sulfide itself, any step during the
above process could control the overall rate of the oxidation reaction. As is shown in Figure 1.9,
18
the gaseous oxidant has to diffuse inward through the MeO layer before reacting with the sulfide
core, and the gaseous product has to diffuse out through the MeO layer. This diffusion of gases
through the solid product layer could be the rate determining step depending on the diffusivity
of the gas in MeO as well as the morphology of MeO. If the solid product MeO is porous in
nature, the gas can easily access the surface of the unreacted core through the pores. It would
less likely be the rate controlling step. The porosity of the solid product is largely dependent on
the reaction mechanism and the volume change of the solid for a certain reaction. Table 1.5
summarizes the volume changes of the solids for the roasting reactions that might be involved
which are calculated based on the densities of the solids. The general rule is that oxidation of
sulfides leads to a volume decrease, meaning the space originally occupied by the sulfides
cannot be fully occupied by the solid product. This gives the solid product a potential to be
porous, which will result in an accelerated roasting reaction. This is particularly true for the
oxidation of pyrrhotite. Figure 1.10 is an optical microscopy image of a roasted pyrrhotite
particle, featuring its characteristic columnar structure. This particular morphological feature
results from its particular oxidation mechanism, i.e. the preferential oxidation proceeding along
certain crystallographic planes [19], as well as the occupied volume decrease illustrated in Table
1.5.
Table 1.5. Volume changes for oxidation and sulfate formation reactions.
19
Note: Densities of sulfides and oxides are cited from Reference [20] and those of sulfates are
cited from Reference [21].
20
(NiFe2O4), a by-product possibly from the solid-solid reaction between iron oxides and nickel
oxide due to local overheating in the fluidized bed, is typically detrimental to the sulfation roast.
Once formed, it is difficult to sulfatize, which would lead to the nickel loss into the residue
during the following leaching step [22-24].
Figure 1.11. Predominance area diagram by superimposing Fe-S-O, Ni-S-O, Cu-S-O and Co-S-
O predominance area diagrams at 680 °C (calculated using data from HSC Chemistry [17]).
1.6.2 Kinetics
Sulfate formation can take place by two possible reaction routes, which are represented by
Reactions (1.14) to (1.16). The former is the direct sulfation of sulfide (e.g. oxidation of nickel
sulfide and iron (II) sulfide), and the latter is the sulfation of oxide with sulfur trioxide (e.g.
formation of CuSO4) [25]. The sulfation of nickel sulfide is an extremely slow process due to
the rather dense nature of the nickel sulfate layer formed which inhibits further sulfation [26].
The formation of NiSO4 from NiS requires 2.5 times the space occupied by the original NiS by
calculation, which can be seen from Table 1.5. This volume increase leads to the dense nature of
21
the NiSO4 [27]. As a result, sulfation of nickel sulfide is so slow that it is impossible to be
employed in commercial production. To address the slow-kinetics problem, previous studies
have investigated the effect of Na2SO4 addition that could accelerate the reactions by one or a
combination of the following mechanisms [23, 28-33]:
1). Na2SO4 forms solid solution (β) or binary sulfates (ε,δ,γ) with NiSO4, which could be seen
from the NiSO4-Na2SO4 binary phase diagram in Figure 1.12 [34]. The formation of any of
these phases increases the stability of NiSO4 by lowering its activity, thus preventing the sulfate
decomposition [23]. Na2SO4 could also destroy the impervious NiSO4 layer by forming a melt
possibly below the roasting temperature [35]. As can be seen from Figure 1.12, the lowest
melting temperature is the eutectic at approximately 670 °C, which could be below the normal
roasting temperature (650–750 °C).
22
2). Na2SO4 destroys the nickel ferrite formed via Reaction (1.17) [22, 36]. If the Na2SO4 melt
contains Fe3+ ions, Reaction (1.18) will take place, in which the role of Na2SO4 is to provide the
melt conditions for the reactive Fe3+ ions [26, 37, 38].
3). Na2SO4 acts as a reservoir of SO3 for the sulfation of NiO via Reactions (1.19) and (1.20)
[22, 36, 39].
Most of the work on sulfation roasting was conducted between 1950 and 1990. The overall
technical challenges seem to be the low Ni yield (maximum around 80%) after leaching due
largely to the formation of nickel ferrite during roasting, as well as the slow kinetics which
made it economically unattractive at the time. Today, however, environmental regulations are
more stringent, to an extent that meeting the imposed limits on emissions is forcing the closure
of smelters. Energy conservation is more important than ever and emissions of heavy metals are
also coming under increasing scrutiny. Therefore, sulfation roasting today could be much more
attractive from energy, environmental and economic points of view if technical and economic
solutions can be found for the above mentioned problems.
23
1.7 References
[1] B. Terry, A.J. Monhemius, A.R. Burkin, General introduction, in: A.R. Burkin (Ed.)
Extractive Metallurgy of Nickel, John Wiley & Sons, New York, 1987, pp. 1-6.
[2] A.E.M. Warner, C.M. Diaz, A.D. Dalvi, P.J. Mackey, A.V. Tarasov, JOM World Nonferrous
Smelter Survey, Part III: Nickel: Laterite, Journal of Metals, 58 (2006) 11-20.
[3] F.K. Crundwell, M.S. Moats, V. Ramachandran, T.G. Robinson, W.G. Davenport,
Extractive Metallurgy of Nickel, Cobalt and Platinum-Group Metals, Elsevier Ltd., Great
Britain, 2011.
[4] F.K. Crundwell, M.S. Moats, V. Ramachandran, T.G. Robinson, W.G. Davenport, Overview
of the Hydrometallurgical Processing of Laterite Ores, in: Extractive Metallurgy of Nickel,
Cobalt and Platinum-Group Metals, Elsevier Ltd., Great Britain, 2011, pp. 117-122.
[5] A.E.M. Warner, C.M. Diaz, A.D. Dalvi, P.J. Mackey, A.V. Tarasov, R.T. Jones, JOM World
Nonferrous Smelter Survey, Part IV: Nickel: Sulfide, Journal of Metals, 59 (2007) 58-72.
[6] 2006-2007 Progress Report on The Canada-Wide Acid Rain Strategy for Post-2000,
Canadian Council of Ministers of the Environment, 2008.
[9] Moving Towards Sustainability, 2005 Good Neighbours Report on Health, Safety,
Environment and Community, Inco Limited, Toronto, Canada, 2005, pp. 93.
[10] J. Joseph R. Boldt, P. Queneau, Roasting, in: The Winning of Nickel, Longmans, Toronto,
1967, pp. 231-237.
[12] C.K. Gupta, D. Sathiyamoorthy, Generalities and Basics of Fluidization, in: Fluid Bed
Technology in Materials Processing, CRC Press, 1999, pp. 1-6.
[14] D.R. Poirier, G.H. Geiger, Correlations and Data for Heat Transfer Coefficients, in:
Transport Phenomena in Materials Processing, The Minerals, Metals and Materials Society,
Pennsylvania, 1994, pp. 248-279.
24
[15] V.A. Borodulya, Y.S. Teplitsky, I.I. Markevich, A.F. Hassan, T.P. Yeryomenko, Heat
transfer between a surface and a fluidized bed: consideration of pressure and temperature
effects, International Journal of Heat and Mass Transfer, 34 (1991) 47-53.
[16] D. Yu, T.A. Utigard, TG/DTA study on the oxidation of nickel concentrate,
Thermochimica Acta, 533 (2012) 56-65.
[17] A. Roine, HSC Chemistry, Outokumpu Research Oy, Pori, Finland, 2007.
[18] H.Y. Sohn, R.P. Goel, Principles of Roasting, Minerals Sci.Engng, 11 (1979) 137-153.
[19] P.G. Thornhill, L.M. Pidgeon, Micrographic Study of Sulfide Roasting, Journal of Metals,
9 (1957) 989-995.
[20] J.W. Anthony, R.A. Bideaux, K.W. Bladh, M.C. Nichols, Handbook of Mineralogy,
Mineralogical Society of America, Chantilly, VA 20151-1110, USA.
[22] P.G. Thornhill, U.S. Patent 2,813,016, Method of Roasting Nickeliferous Sulfide
Concentrates in a Fluidized Bed, United States, 1957.
[23] A.W. Fletcher, M. Shelef, The Role of Alkali Sulphates in Promoting the Sulphation
Roasting of Nickel Sulphides, in: Unit Process in Hydrometallurgy. Group C: Plant Operating
Practice - Economics - General, New York Gordon and Breach Science Publishers, 1963, pp.
946-970.
[24] A.W. Fletcher, M. Shelef, A Study of the Sulfation of a Concentrate Containing Iron,
Nickel, and Copper Sulfides, Transactions of the Metallurgical Society of AIME, 230 (1964)
1721-1724.
[25] J.G. Dunn, The oxidation of sulphide minerals, Thermochimica Acta, 300 (1997) 127-139.
[26] P.J. Saikkonen, U.S. Patent 4,464,344, Process for Recovering Non-ferrous Metal Values
From Ores, Concentrates, Oxidic Roasting Products or Slags, United States, 1984.
[27] M.C.B. Hotz, T.R. Ingraham, The Sulphation of Tricobalt Tetroxide and Nickel Monoxide
with Fused Sodium Hydrogen Sulphate, Canadian Metallurgical Quarterly, 4 (1965) 295-302.
[28] W.K. Sproule, P.E. Queneau, G.C. Nowlan Jr, Canadian Patent CA 593622, Method for
producing high grade hematite from nickeliferous iron sulfide ore, Canadian Intellectual
Property Office (CIPO), Canada, 1960.
[29] P.G. Thornhill, U.S. Patent 2,930,604, Fluidized Bed Roasting of Metal Sulfide
Concentrates, Unites States, 1960.
[32] P.G. Thornhill, The Falconbridge Iron Ore Process, AIME Transactions, LXIV (1961) 337-
344.
[33] J.R. Boldt, Jr., Falconbridge Pyrrhotite Process, in: P. Queneau (Ed.) The Winning of
Nickel, Longmans Canada, Toronto, 1967, pp. 331-336.
[34] N. Birks, G.H. Meier, F.S. Pettit, Hot Corrosion, in: Introduction to The High-Temperature
Oxidation of Metals, Cambridge University Press, United States of America, 2006, pp. 205-252.
[35] J.M. Toguri, A review on the methods of treating nickel-bearing pyrrhotite; with special
reference to the Sudbury Area pyrrhotite, Canadian Metallurgical Quarterly, 14 (1975) 323-338.
[36] M.C.B. Hotz, R.C. Kerby, T.R. Ingraham, The sulphation of nickel and cobalt ferrites and
sulphides by molten sodium pyrosulphate and sodium bisulphate, Canadian Metallurgical
Quarterly, 7 (1968) 205-210.
[37] P.J. Saikkonen, J.K. Rastas, The Role of Sulfate Melts in Sulfating Roasting, in: 25th
Annual Conference of Metallurgists, Toronto, Ontario, Canada, 1986, pp. 278-290.
[38] K.T. Jacob, Kinetics of Sulphation of Nickel and Cobalt Ferrites with Na2SO4 Melt
Containing Fe3+ Ions, Transactions of the Indian Institute of Metals, 40 (1987) 383-388.
[39] M.C.B. Hotz, T.R. Ingraham, Fused Sodium Pyrosulphate as a Sulphation Catalyst for NiO
and Co3O4, Canadian Metallurgical Quarterly, 5 (1966) 237-244.
26
2 TG/DTA Study on the Oxidation of Nickel Concentrate
2.1 Introduction
In the proposed selective oxidation-sulfation roasting route for nickel processing, the objective
of the first step, i.e. oxidation roasting, is to preferentially oxidize the iron sulfide species from
the nickel concentrate to form oxides. Therefore, it is of critical importance to understand the
oxidation roasting mechanisms of the nickel concentrate. Experiments were therefore conducted
to achieve this end by means of thermogravimetric (TG) and differential thermal analysis
(DTA).
Oxidizing roasting mechanisms of metal sulfide minerals are quite complex because they vary
depending on the sulfide species in the minerals, the roasting conditions employed, and the
interactions among all species present. Lab scale thermogravimetric and differential thermal
analysis studies on the oxidation of sulfide minerals have much milder oxidative conditions than
the industrial scale fluidized bed roasting in terms of the effective gas-solid contact area and the
heating rate. However, TG/DTA studies yield valuable information on the chemical reaction
routes and the kinetics of gas-solid reactions during roasting due to its ability to precisely
control the experimental conditions and to monitor the variables. In this investigation, attempts
were made to reveal the reaction sequences during the oxidation of a nickel concentrate by
TG/DTA. Gas analyzer was employed to continuously measure the offgas SO2 and O2 contents
during the TGA runs. X-ray powder diffraction (XRD), scanning electron microscopy (SEM),
energy dispersive X-ray spectroscopy (EDS), electron probe microanalysis (EPMA) and
inductively coupled plasma optical emission spectroscopy (ICP-OES) were employed to
characterize the concentrate and the oxidation products, namely calcines. The rate controlling
step in the TGA tests was also investigated.
Many researchers have studied the oxidation of various metal sulfides on a laboratory scale.
Studies on the oxidation of nickel, iron, and copper containing sulfide minerals, as well as
studies on the thermal decomposition of metal sulfates that could be formed during the roasting
of the nickel concentrate, are briefly reviewed in this section.
27
The oxidation of a natural pentlandite (Fe,Ni)9S8 was studied by Dunn and Kelly [1] using TG
and DTA. In their study, an exothermal event was detected by the DTA at 520 ºC which was
claimed to be due to the oxidation of iron sulfide (FeS) forming iron oxide (Fe2O3). Another
exothermal event at 575 ºC was suggested to be caused by the oxidation of the Fe8±xNi8±yS16±z
phase which forms upon heating the natural pentlandite. Upon reaching 800 ºC, the temperature
is in the vicinity of the melting point of the unquenchable compound Ni3±xS2, which leads to its
accelerated oxidation. This thermal event was also suggested in [2, 3]. And a reaction scheme
was deduced, Table 2.1 [1]. The oxidation of synthetic millerite (Ni0.994S) was also investigated
by Dunn and Kelly [2] with the same apparatuses in a dynamic oxygen atmosphere (0.2 L/min)
up to 1000 ºC. Under these conditions, a reaction scheme was proposed and is shown in Table
2.2.
Table 2.1 Various reactions occurring during the oxidation of pentlandite at 10 ºC/min in an
air/oxygen flow of 0.2 L/min [1].
<460 ºC (Ni,Fe)9±xS8±y→Fe8±xNi8±yS16±z+NiS+FeS
460-640 FeS+2O2→FeSO4
460-715 NiS+2O2→NiSO4
520 2FeS+3.5O2→Fe2O3+2SO2
575 Fe8±xNi8±yS16±z+O2→Ni2FeO4+SO2+NiO+Fe2O3
640-760 2FeSO4→Fe2O3+(2SO2+0.5O2)
(Ni,Fe)9±xS8±y→Ni3±xS2+ Fe8±xNi8±yS16±z
700-740 Fe8±xNi8±yS16±z+O2→Ni2FeO4+SO2
NiS+1.5O2→NiO+SO2
O2
775-805 Ni3±xS2 → NiO+(SO2+0.5O2)
>800 NiSO4→NiO+(SO2+0.5O2)
28
Table 2.2. Various reactions occurring in the oxidation of Ni0.994S [2].
400-785 ºC NiS+2O2→NiSO4
500-785 NiO+SO2+0.5O2→NiSO4
540-552 and 682-692 NiS+1.5O2→NiO+SO2
682-692 and possibly 3NiS+2NiSO4→Ni3S2+2NiO+3SO2
682-692 10NiS+2O2→Ni7S6+Ni3S2+2SO2
682-780 and/or Ni7S6+9.5O2→7NiO+6SO2
Ni7S6+12.5O2→6NiSO4+NiO
793 Ni3S2(s) →Ni3S2(l)
793 Ni3S2(l)+3.5O2→3NiO+2SO2
>800 NiSO4→NiO+SO2+1/2O2
Kennedy and Sturman [4] investigated the oxidation of iron(II) sulfide Fe1-xS in air and oxygen
atmospheres using TG, DTA, XRD and chemical analysis. A reaction scheme reported in this
study is shown in Table 2.3. The rate of oxidation of ferrous sulfide in a tiny bucket was
examined by means of a spring balance and X-ray powder diffraction analysis [5]. The bulk of
sample in the bucket had a disk-type shape, with several millimeters thickness. By varying the
sample sizes, the conclusion was drawn that the rate was controlled by the diffusion of oxygen
and/or gaseous products through the bed of the sample.
The oxidation of chalcopyrite was studied by means of TG, DTA, XRD and colorimetry
techniques [6]. The results indicate that in the range between 350 ºC and 440 ºC, the sample
surface, which had free access to oxygen, was oxidized into iron sulfate, CuSO4 and Fe2O3. On
the other hand, below the surface where oxygen was relatively starved, bornite (Cu5FeS4) could
be formed as an inert atmosphere product along with the formation of FeS and elemental sulfur.
Between 440 ºC and around 500 ºC these inert atmosphere products were oxidized significantly.
29
Iron sulfate was fully decomposed at 700 ºC, after which CuSO4 would be decomposed to CuO
with the formation of an intermediate compound CuO·CuSO4. In the temperature range between
800 ºC and 850 ºC, CuO, CuFe2O4 and Fe3O4 were the only stable species.
Table 2.3. Reaction scheme for the oxidation of iron sulfide (Fe1-xS) [4].
Fe1-xS+(2-x)O2→(1-x)FeSO4+xSO2
3Fe1-xS+(5-2x)O2→(1-x)Fe3O4+3SO2
425–520 ºC
4Fe1-xS+(7-3x)O2→(2-2x)Fe2O3+4SO2
8Fe1-xS+(2-4x)SO2→(1-2x)Fe3O4+(5-2x)FeS2
520–572 ºC 4Fe1-xS+(7-3x)O2→(2-2x)Fe2O3+4SO2
FeS2+3O2→FeSO4+SO2
575–625 ºC 12FeSO4+3O2→2[Fe2(SO4)3]2∙Fe2O3
4Fe3O4+O2→6Fe2O3
625–725 ºC [Fe2(SO4)3]2∙Fe2O3→3Fe2O3+6SO3
Thermal decomposition of nickel (II), copper (II), iron (II and III), and cobalt (II) sulfates have
been investigated by many researchers [7-13]. The decomposition temperatures of Fe2(SO4)3,
CuSO4, NiSO4 and CoSO4 in air were reported to be 575 ºC, 625 ºC, 675 ºC and 720 ºC
respectively by TG/DTA [7]. The thermal decomposition process of copper(II) sulfate consisted
of two steps, which are shown as Reactions (2.1) and (2.2) [7, 10, 13].
2CuSO4=CuO·CuSO4+SO3 (2.1)
CuO·CuSO4=2CuO+SO3 (2.2)
30
2.2 Experimental
2.2.1 Sample
The nickel concentrate investigated was Raglan concentrate from the Xstrata Sudbury Smelter.
The XRD analysis show that it is mainly composed of three minerals, i.e. pentlandite (Fe,Ni)9S8,
chalcopyrite CuFeS2 and pyrrhotite Fe1-xS, Figure 2.1. The chemical composition and estimated
mineral contents are presented in Table 2.4. Electron Probe Microanalysis (EPMA) was
employed to evaluate the stoichiometry of pentlandite, chalcopyrite, pyrrhotite, and pyrite in
different individual grains. The results of analysis are exhibited in Table 2.5. Figure 2.2 shows
the BSE image of the Raglan concentrate. Most of the sulfide grains in this image with particle
size larger than around 3 µm were analyzed by EDS for their mineralogical composition to
estimate the relative particle population of these four sulfides. 68 pentlandite grains (56.2%), 20
chalcopyrite grains (16.5%), 32 pyrrhotite grains (26.4%) and 1 pyrite grain (0.8%) were
identified in the image. The particle size of Raglan concentrate was determined by means of
sieve analysis using a Ro-Tap RX-29 type mechanical sieve shaker. The results of particle size
analysis are shown in Table 2.6. Particle size distribution of the material was also obtained using
a Laser Particle Size Analyzer (Hydro 2000S, Malvern Instruments) with the results shown in
Figure 2.3.
Figure 2.2. BSE image of the Raglan concentrate (pn: pentlandite; cpy: chalcopyrite; po:
pyrrhotite; py: pyrite; flux: silicate flux).
32
Table 2.5. EPMA compositional analysis of individual grains of the Raglan concentrate and
their calculated stoichiometry, which shows the composition difference between particles of the
same mineral.
Raglan concentrate Composition (wt%) Stoichiometry
Grain
Minerals Fe Ni Cu Co S Total Fe Ni Cu Co S
#
1 27.8 36.8 0.058 0.609 34.2 99.5 3.73 4.70 0.0069 0.0775 8
2 32.1 33.7 0.050 0.657 33.0 99.5 4.47 4.47 0.0062 0.0867 8
3 31.3 35.1 0.063 0.346 33.6 100.4 4.29 4.57 0.0075 0.0449 8
4 31.2 34.5 0.051 0.575 33.0 99.3 4.34 4.57 0.0062 0.0758 8
Pentlandite 5 30.3 35.7 0.022 0.633 33.2 99.9 4.19 4.70 0.0026 0.0829 8
(Fe,Ni)9S8 6 31.3 34.7 0.040 0.366 32.9 99.3 4.38 4.61 0.0049 0.0485 8
7 30.5 35.2 0.121 0.455 33.1 99.4 4.24 4.65 0.0148 0.0598 8
8 31.8 34.3 0.010 0.575 33.0 99.7 4.43 4.54 0.0012 0.0759 8
9 31.2 35.0 0.033 0.460 32.9 99.7 4.36 4.65 0.0041 0.0608 8
10 31.0 35.2 0.033 0.624 33.2 100.0 4.29 4.63 0.0040 0.0818 8
11 30.8 0.247 34.0 0.000 34.8 99.8 1.02 0.0077 0.986 0.0000 2
Chalcopyrite 12 30.5 0.164 34.2 0.000 35.2 100.0 0.994 0.0051 0.979 0.0000 2
CuFeS2 13 30.5 0.123 34.1 0.000 35.0 99.7 1.00 0.0038 0.986 0.0000 2
14 30.6 0.216 34.0 0.000 34.9 99.6 1.01 0.0068 0.985 0.0000 2
15 59.6 0.342 0.035 0.000 39.7 99.6 0.862 0.0047 0.0004 0.0000 1
16 59.2 0.693 0.004 0.000 39.4 99.3 0.862 0.0096 0.0001 0.0000 1
Pyrrhotite 17 59.5 0.792 0.142 0.000 38.8 99.2 0.879 0.0111 0.0019 0.0000 1
Fe1-xS 18 59.9 0.824 0.080 0.000 38.7 99.5 0.888 0.0116 0.0010 0.0000 1
19 59.7 0.646 0.054 0.000 38.7 99.1 0.887 0.0091 0.0007 0.0000 1
20 59.4 0.352 0.087 0.000 39.4 99.2 0.865 0.0049 0.0011 0.0000 1
21 45.3 0.183 1.69 0.168 53.8 101.1 0.966 0.0037 0.0316 0.0034 2
22 45.0 0.284 0.096 0.896 54.3 100.5 0.951 0.0057 0.0018 0.0180 2
Pyrite 23 46.2 0.084 0.000 0.789 53.4 100.5 0.994 0.0017 0.0000 0.0161 2
FeS2 24 44.5 0.111 0.067 2.25 53.2 100.2 0.960 0.0023 0.0013 0.0459 2
25 45.7 0.139 0.041 1.31 53.1 100.2 0.987 0.0029 0.0008 0.0268 2
26 44.2 0.101 0.080 1.90 53.1 99.3 0.956 0.0021 0.0015 0.0390 2
33
Figure 2.3. Particle size distribution of the Raglan concentrate.
Counterbalance
Cooling Water
Jacket
Alumina Crucible
-35.21 756
-40 807
-10 3000
2000
-51.97
Offgas O2 (ppm)
-60 2500
-15 -67.61 -70.22 0
-70.81
-80 2000
-20 161
-2000 1500
-100
Mass change
-25 Offgas O2 (drifting) 1000
-120 Offgas SO2 -4000
36
100 Solidification
Exo
80 Ar, 100mL/min; 15oC/min
60 ∆T #1
#2
40 #3
Heat flow (µV) Endo #4
20
Heat up
0
Figure 2.6. Repeated tests for the determination of the melting temperature of Ag by DTA.
SEM/EDS – The sample powders were mounted in an epoxy resin. After the epoxy was cured,
the surface was ground with silicon carbide papers with abrasive sizes of 180, 400, 600, 800,
and 1200 sequentially. Final surface preparation was made on a polishing pad with water based
0.3 µm alumina suspension as the polishing media. For samples which contain water soluble
sulfate species, oil based 1µm diamond suspension was used instead. The surface was then
coated with carbon to render it electrically conductive. The SEM used is a JEOL, JSM-840,
complemented by a PGT/AAT EDS detector (thin window) and an IXRF 500 digital pulse
37
processor, allowing both X-ray microanalysis and digital imaging, via SE, BSE and X-ray
signals.
EPMA - Compositional analyses were acquired on a Cameca SX50 electron probe x-ray
microanalyzer equipped with 3 tunable wavelength dispersive spectrometers. Operating
conditions were 40 degrees takeoff angle, and a beam energy of 20 keV. The beam current was
20 nA, and the beam diameter was 1 µm. Elements were acquired using analyzing crystals LiF
for Fe Kα, Ni Kα, Cu Kα, Co Kα, PET for S Kα, and PC1 for O Kα. The standards were
hematite for O Kα, cobaltite for Co Kα, pentlandite for S Kα, Fe Kα, Ni Kα, and chalcopyrite
for Cu Kα. The counting time was 20 seconds for Fe Kα, Ni Kα, Cu Kα, S Kα, 40 seconds for O
Kα, and 60 seconds for Co Kα.
38
mineralogical compositions. Their XRD patterns are provided in Figure 2.10. Due to the
complex nature of the system, assignments of some of the peaks were only tentatively
suggested. Results of the wet chemical analysis of the quenched samples by hot water leaching
and ICP is illustrated in Figure 2.11 as the percentages of water soluble sulfates calculated on
the basis of the content of each element in the calcines. In order to better understand the
oxidation mechanism of pentlandite in the concentrate, EPMA was applied to investigate the
chemical composition change of the pentlandite sulfide cores during the dynamic heating
period. The results are plotted in the Fe-Ni-S ternary diagram as shown in Figure 2.12.
3 12 8000
0 0
Change in mass (wt%)
5000
-1 -4
Change in mass 4000
-2 -8 Rate of mass change
Offgas SO2
-3 -12 Offgas O2 consumed 3000
-4 -16 2000
-5 -20
1000
-6 -24
0
300 400 500 600 700 800 900
Temperature (oC)
Figure 2.7 Sample mass change, rate of mass change, SO2 concentration and O2 consumption in
the offgas for the TGA run in which 100 mg concentrate was heated to 950 ºC at 15 ºC/min in
air.
39
80
70
Exo
60
50
Heat flow ( µV)
40
Endo
30
20
10
0
100 200 300 400 500 600 700 800 900 1000
Temperature (oC)
350
300
Exo
250
Heat flow ( µV)
200
150
Endo
100
50
0
100 200 300 400 500 600 o
700 800 900 1000
Temperature ( C)
40
10,11,12 1.(Fe,Ni)9S8; 2.CuFeS2; 3.Fe1-xS; 4.Ni1-xS; 5.(Fe,Ni)1-xS;
6.FeNi2S4; 7.CuO•CuSO4; 8.Ni3S4; 9.Ni3S2; 10.Fe2O3;
11.Fe3O4; 12.NixFe3-xO4; 13.Fe2(SO4)3; 14.NiS2; 15.FeSO4;
16.CuSO4;17.NiO; 18.NiSO4; 19.MgSO4
17,12,11
10
950oC 11,12,17
11,12 17
10 10 10,11,12
10 19 10 10 10
11,12 19 19 12
19
10,11,12
10
o
880 C
10 19 17,12,11 10 11,12,17
11,12 18 10
18 19 17 18 10 10,11,12 10 10
11,12 18 9
10
10,11,12
o
785 C 4
13,16 4 10
10 4 10 4
1816 1618
16,17 18 10 17,12 10,12
10 10
17
13
Counts
10 10,11,12 4
o
733 C 13,16 4
4
10 4 10 10,11,12
16 10 1318
11,12 16 18 10 17,12 18 10,12 10
4
o
10,11,12
605 C 4 10,15 4
4
13,15
10,15 15 10 10,11,12
11,12 15,16 10 12 10,15 12 4 10
2,1 7,10,11 4
4,5
509oC 7 10 74 5 4,5,14
11 8,14 2
1 8 1 10 2 1 11 10
1,2
7
495oC 4,5 10 7,10,11
74 5 4 1 4,5
1 8 11
7 18 81 1 2
2 8 2
1,2
450oC 1
3,4
1
22
1 7 3 54 3,5,1
6 103,5,7 1 1 6 22
11
15 20 25 30 35 40 45 50 55 60 65
2θ ( ο )
Figure 2.10. XRD patterns for calcines quenched from intermediate temperatures after heating at
15 ºC/min in air in TGA runs.
41
100
90 Fe
Ni
Figure 2.11. Chemical analysis results for the contents of the water soluble sulfates in the
calcines quenched from intermediate temperatures after heating at 15 ºC/min in air in TGA runs.
0 100
10 Pentlandite
20
90
80
80
450oC
)
30
ic%
70
S(
40 60 605oC
ato
om
50 50
mic
733oC
(at
60 40
70
%)
Fe
70 30
80 20 785oC
90
100
10
0
880oC
0 10 20 30 40 50 60 70 80 90 100
Ni (atomic%)
60
50
40
Figure 2.12. Fe-Ni-S diagram showing the change of chemical compositions during non-
isothermal heating in TGA runs analyzed by EPMA. The bottom part is the magnified area of
the trapezoid in the top ternary diagram.
42
Mass changes are barely seen below 350 ºC from Figure 2.7. Afterwards, there is a gradual
increase in mass till 498 ºC, where the mass reaches a small peak and starts to decrease.
Correspondingly, O2 in the feeding gas was consumed intensively and SO2 was emitted, which
is denoted by the intensive SO2 and O2 peaks in Figure 2.7, indicating the occurrence of
vigorous oxidation reactions. This is further confirmed by two partly overlapped exotherms in
the vicinity of 500 ºC in Figure 2.8, indicating the presence of at least two exothermic reactions
leading to the mass change and SO2 emission. These two reactions were greatly promoted by
increasing the oxygen partial pressure shown as a much stronger single peak at 437 ºC in Figure
2.9, which confirms the oxygen involvement in these two exotherms. The mass change between
450 ºC and 509 ºC in Figure 2.7 is mainly due to two competing reactions, which are the
oxidation of chalcopyrite forming CuO·CuSO4 and the oxidation of pyrrhotite forming Fe2O3.
These two reactions are expressed as Reaction (2.3) and Reaction (2.4), the former leading to
the mass gain while the latter leading to the mass loss. Reaction (2.3) dominates below 498 ºC,
causing an increase in the mass while the mass loss above 498 ºC is primarily due to Reaction
(2.4).
2CuFeS2+7O2→CuO·CuSO4+Fe2O3+3SO2 (2.3)
2Fe1-xS+(3.5-1.5x)O2→(1-x)Fe2O3+2SO2 (2.4)
The evidence to Reaction (2.3) is the XRD patterns, showing the dramatic reaction of
chalcopyrite and the formation of Fe2O3 and CuO·CuSO4 between 450 ºC and 509 ºC. The
assignment of Reaction (2.4) is based on the fact that pyrrhotite disappeared between 450 ºC and
509 ºC and that a substantial amount of Fe2O3 was formed at 509 ºC, as suggested by the XRD
patterns in Figure 2.10. Dunn and Kelly [1] observed the same reaction at 520 ºC. This
disagreement of temperature under which the oxidation of Fe1-xS occurs is partly caused by the
difference between the experimental conditions employed in the two studies. Another factor that
may contribute to the disagreement is the difference in the samples being studied. Since their
sample is a natural pentlandite, there would be no Fe1-xS unless it is formed from the
decomposition of pentlandite. While the sample in this study is a concentrate containing not
only pentlandite, but also around 16 wt% pyrrhotite, 13 wt% chalcopyrite and some siliceous
gangue. Figure 2.13, Figure 2.14, and Figure 2.15 illustrate the morphological features of the
43
oxidized pyrrhotite air-quenched from 605 ºC, 733 ºC and 880 ºC, respectively. Reaction (2.4)
would result in the volume reduction of 24% (calculated based on the densities of Fe2O3 and
Fe3O4). From this, one would expect a porous micro-structure of the oxidized pyrrhotite
particles. However, the pyrrhotite particle seems to be rather dense in Figure 2.13. It became
porous after heating to higher temperatures as shown in Figure 2.14 and Figure 2.15. We resort
to the Fe-S-O predominance diagram to help understand the change of micro-structure as a
function of temperature. Figure 1.7 is the Fe-S-O predominance diagram at 500 ºC, 700 ºC and
900 ºC calculated using HSC Chemistry [14]. The pyrrhotite particles experienced an oxidative
atmosphere with the O2 partial pressure of approximately 0.21 atm at the beginning of oxidation.
The propagation of the oxidation reaction as Reaction (2.4) in the bulk of the sample bed
consumed O2 from the local gas environment and released SO2, which in turn lead to a drop in
the partial pressure of O2 and an increase in the partial pressure of SO2. Once the partial pressure
of SO2 exceeded 10-5 atm, the gas environment in the bulk of sample bed would favor the
formation of Fe2(SO4)3 at 500 ºC. Indeed around 2.5 wt% S was detected in the oxidized
pyrrhotite particle in Figure 2.13 by EDS. This is indicative of the presence of Fe2(SO4)3 which
is disseminated in Fe2O3. The formation of Fe2(SO4)3 alone from pyrrhotite would result in a
volume increase by 223%. It is thus easy to imagine that even a small fraction of Fe2(SO4)3
would offset the volume reduction caused by the formation of Fe2O3. This formation of
Fe2(SO4)3 along with Fe2O3 resulted in the rather dense nature of the oxidized pyrrhotite at
relatively low temperature as seen in Figure 2.13. The formation of Fe2(SO4)3 became more
difficult with the increase of temperature indicated by the shrinkage of the predominant area for
Fe2(SO4)3 in Figure 1.7 as temperature increases. The particles became porous at higher
temperature as can be seen in Figure 2.14 and Figure 2.15 since Fe2(SO4)3 is no longer stable.
This characteristic columnar reaction interface with distinct orientations in these two figures was
reported to be due to the inward preferential oxidation along certain crystallographic plains [15,
16].
44
Figure 2.13. SE image of an oxidized pyrrhotite particle quenched from 605 ºC in the TGA run.
Figure 2.14. SE image of an oxidized pyrrhotite particle quenched from 733 ºC in the TGA run
showing its characteristic columnar structure.
45
Figure 2.15. SE image of an oxidized pyrrhotite particle quenched from 880 ºC in the TGA run.
Below 495 ºC, very little sulfates were formed, illustrated by the profiles in Figure 2.11. But an
increase in the content of water-insoluble CuO·CuSO4 is shown in the XRD patterns till 495 ºC.
Decomposition of chalcopyrite would result in the formation of bornite (Fe5CuS4), as indicated
by previous studies [6, 16]. This phase was not detected by XRD in any of the quenched
samples in this study, probably because the content of this phase was too low to be identified by
XRD. However, bornite was identified by SEM/EDS. As illustrated in the backscattered
electron (BSE) image of Figure 2.16, bornite formed around the chalcopyrite core with a clear
boundary between them. The porous surface layer is Fe2O3 formed by the preferential oxidation
of iron species from chalcopyrite. The intensity of the peaks for CuO·CuSO4 decreased after
495 ºC, meaning its content was decreasing as the heating proceeded. While at the same time,
the content of CuSO4 started to increase. This suggests that CuO·CuSO4 is a precursor for the
formation of CuSO4, which has also been reported by other researchers [17].
46
Figure 2.16. BSE image of a chalcopyrite particle after heating the Raglan concentrate in air at
450 ºC for 1 hour (Cpy: chalcopyrite, CuFeS2; Bor: bornite, Cu5FeS4).
Another reaction possibly has occurred up to 605 ºC based on the XRD patterns. The
consumption of pentlandite and formation of monosulfide solid solution (Mss), Ni1-xS, and
Fe2O3 are obvious and illustrated by the XRD patterns at 450 ºC, 495 ºC and 509 ºC in Figure
2.10. As a result, this reaction is suggested to be the oxidation of pentlandite forming (Ni,Fe)1-
xS, Ni1-xS, and Fe2O3 with the assignment of Reaction (2.5). SO2 produced from Reaction (2.5)
and the O2 in the feeding gas provide suitable atmosphere for sulfate formation, which is
revealed by the massive metal sulfate formation between 509 ºC and 605 ºC in Figure 2.11.
This, as a result, gives rise to an accelerating mass increase, indicated by the peak in the rate of
mass change curve in Figure 2.7, corresponding to the right uplifting shoulders of the offgas
curves. (Ni,Fe)1-xS in Reaction (2.5) represents the non-stoichiometric iron nickel sulfide, Mss
with hexagonal pyrrhotite structure [18]. Apart from the XRD observation, some other evidence
also supports the occurrence of Reaction (2.5). Firstly, Raglan concentrate was heated in argon
at 450 ºC for 1 hour and was quickly cooled down to room temperature with (Ni,Fe)1-xS, Ni1-xS
and Fe1-xS found in the product. Dunn and Kelly [1] also found (Fe,Ni)1-xS, NiS and FeS as the
decomposition product of pentlandite in a dynamic oxygen atmosphere below 460 ºC. Secondly,
Raglan concentrate was oxidized isothermally at 450 ºC for 1 hour and its product was
examined under microscope, a thin layer of iron oxide was found to have formed on the surface
of the pentlandite particle, with its core being converted to Mss. This suggests iron cations in
47
pentlandite are oxidized preferentially via outward diffusion mechanism. Preferential oxidation
of iron species from pentlandite has been studied by many researchers and is well recognized
[15, 16, 19, 20]. Iron cations tend to migrate out from the pentlandite particles towards the
oxygen/oxide interface close to the surface of the particles where they combine with O2 forming
Fe2O3.
(Ni,Fe)9S8+O2→(Ni,Fe)1-xS+Ni1-xS+Fe2O3+SO2 (2.5)
Violarite solid solution (Ni,Fe)3S4 was also formed in the TGA samples quenched from 450 ºC,
495 ºC and 509 ºC, which is denoted by two stoichiometric forms FeNi2S4 and Ni3S4. Although
(Ni,Fe)3S4 was reported to be stable only below 461±3 ºC in the presence of an equilibrium
sulfur vapor [21], the dynamic heating conditions applied in the present study sustained its
presence up to 509 ºC. In nature, violarite is often intimately associated with pentlandite.
Violarite forms readily from pentlandite because the pentlandite structure can be easily
converted to that of violarite with a minimum reorganization, by removal of excess metal atoms
and redistribution of the remainder [22]. Based on this, violarite in the present study is
considered to be produced by the transformation of pentlandite. The removal of the excess Fe
atoms was accelerated by the oxidizing condition forming Fe2O3 as a by-product. The
transformation reaction is represented as Reaction (2.6). This reaction coincides with the mild
mass increase and no significant SO2 release up to 500 ºC. Thermal decomposition of violarite
occurred above 509 ºC as indicated by the XRD results showing the disappearance of its peaks
at 605 ºC. Dunn and Howes [23] studied the oxidation of violarite and reported that its
decomposition in the oxidizing atmosphere leads to the formation of Mss and SO2 in the
temperature range 405–475 ºC. As Mss was also formed in the present study and significant SO2
release was occurring above 509 ºC, Reaction (2.7) is tentatively suggested.
The following thermal event is indicated by the spiking shoulder of the DTA curve from 550 ºC
to 690 ºC in Figure 2.8, corresponding to the wide peaks in the rate of mass change curve and
offgas curves at 660 ºC. In Figure 2.9, this exotherm is shown as a wide peak from 468 ºC to
48
700 ºC peaking at 562 ºC. This reaction can be deduced based on the XRD patterns in Figure
2.10 along with the change in composition of the pentlandite sulfide cores in Figure 2.12.
Although no Mss at 605 ºC was detected by XRD (Figure 2.10), its presence was confirmed by
EPMA in Figure 2.12. At 650 ºC, the points correspond to the sulfur-excess compound of
(Fe,Ni)1-xS with various Fe:Ni ratios. There was a relationship between the Fe:Ni ratios and the
size of the particles analyzed, which was that a smaller particle tended to have a lower Fe/Ni
ratio. This is expected because in a smaller particle which has a larger surface to volume ratio,
the average path for the Fe2+ ion to diffuse out of the sulfide core and be preferentially oxidized
was shorter, which means a shorter time was needed for the depletion of Fe2+ ion from the
sulfide core. This phenomenon also suggests that smaller particles usually have higher oxidation
rates, provided other conditions remain unchanged. At 733 ºC, all the pentlandite sulfide cores
were transformed into Ni1-xS containing around 2.5 atm% Fe whose percentages reached a
minimum and did not go any lower with the increase of temperature. XRD patterns also show
the increasing contents of Ni1-xS and Fe2O3, as well as the full consumption of the intermediate
product Mss from 495 ºC to 733 ºC. All of these suggest that this thermal event is the oxidation
of Mss forming Ni1-xS and Fe2O3, represented by Reaction (2.8). The mechanism behind this
reaction is also the preferential oxidation of iron species from Mss via diffusion, forming an iron
oxide shell and a nickel sulfide core in each pentlandite grain. Figure 2.17 is a BSE image of a
pentlandite particle quenched from 733 ºC showing its particular structure of a nickel sulfide
core surrounded by a porous layer of iron oxide. EDS analysis indicated the iron sulfates were
disseminated in the iron oxide. There is a discrepancy in the temperature range in which
Reaction (2.8) occurs in the TGA and DTA modes, which is probably caused by the geometrical
difference of the crucibles used in these two modes, as well as a larger sample size used in the
TGA mode.
49
Figure 2.17. BSE image of a partly oxidized pentlandite particle air-quenched from 733 ºC in
the TGA run.
(Ni,Fe)1-xS+O2→Ni1-xS+Fe2O3+SO2 (2.8)
Previously formed Ni1-xS was further oxidized giving rise to another exotherm peaking at 731
ºC in Figure 2.8. Quenched samples from 695 ºC and 785 ºC in the DTA runs were analyzed by
XRD to evaluate this exotherm, with their XRD patterns shown in Figure 2.18. The
disappearance of Ni1-xS and formation of heazlewoodite (Ni3S2) can be clearly seen on the basis
of these two XRD patterns. The formation of Ni3S2 is indicative of the unquenchable compound
Ni3±xS2, which was suggested by Dunn and Kelly [1], and was further confirmed in this study
using SEM/EDS. As a result, this exotherm is suggested to be the oxidation of Ni1-xS forming
Ni3±xS2, represented by Reaction (2.9). The temperature discrepancy between TGA and DTA
runs can be observed by comparison between the XRD patterns for TGA runs in Figure 2.10 and
those for DTA runs in Figure 2.18. The XRD patterns for 695 ºC in the DTA mode fit well with
that for 733 ºC in the TGA mode. The formation of Ni3S2 is only observed in the TGA sample
quenched from 880 ºC, which resemble the DTA sample quenched from 785 ºC. Due to this
temperature discrepancy, Reaction (2.9) did not occur by 785 ºC in the TGA mode, which can
be seen from the TGA curves and XRD results in Figure 2.7 and Figure 2.10, respectively.
50
10,21 1.(Fe,Ni)9S8; 2.CuFeS2; 3.Fe1-xS; 4.Ni1-xS; 5.(Fe,Ni)1-xS;
6.FeNi2S4; 7.CuO•CuSO4; 8.Ni3S4; 9.Ni3S2; 10.Fe2O3;
10 11.Fe3O4; 12.NixFe3-xO4; 13.Fe2(SO4)3; 14.NiS2; 15.FeSO4;
16.CuSO4; 17.NiO; 18.NiSO4; 19.MgSO4; 20.CuO; 21.Fe2(SiO4)
10
785oC 10 19 10 10,12
17,21
10 10
Counts
10 18 10,12 18 10 21,10
9 17
19
18
10
10,21
695oC 13,19
4
4,13 4 10
10 10 10
13 13 10 10 4 10 10
13 21
13 20 21,10
15 20 25 30 35 40 45 50 55 60 65
2θ ( o )
Figure 2.18. XRD patterns for calcines quenched from intermediate temperatures after heating at
15 ºC/min in air in DTA runs.
(3±y)Ni1-xS+(1+2x±y)O2→(1-x)Ni3±yS2+(1+2x±y)SO2 (2.9)
The next thermal event appears as a sharp peak at 813 ºC in the DTA curve in Figure 2.8,
corresponding to a large O2 consumption peak at 822 ºC in Figure 2.7. This exotherm is
believed to be the oxidation of Ni3±xS2 which has an incongruent melting point in the vicinity of
800 ºC. Ion diffusion in the sulfide was greatly increased after Ni3±xS2 melted, resulting in the
accelerated oxidation reaction rate. This exotherm was also reported by others [1, 2]. Along with
NiO, a substantial amount of NiSO4 was also formed in this exotherm, which can be clearly seen
in the XRD pattern at 880 ºC in Figure 2.10 and its chemical analysis results in Figure 2.11. The
very weak emission of SO2 at 822 ºC in Figure 2.7 compared with its corresponding intensive
O2 consumption peak also indirectly suggests the occurrence of sulfation rather than SO2
formation. As a result, this exotherm is tentatively suggested as Reaction (2.10).
51
Figure 2.19 and Figure 2.20 are BSE images of two partly oxidized pentlandite particles air-
quenched from 880 ºC, showing their morphological features. It comprises of complex layers of
iron oxides and a nickel sulfide core. It is difficult to determine the stoichiometry of the iron
oxides layers by EDS or EPMA when the oxide layers are either porous or too thin. Their
stoichiometries were tentatively suggested based on other observations as follows. The outer
dark porous layer of iron oxide shown in Figure 2.19 and Figure 2.20 was not seen in any partly
oxidized pentlandite particles quenched from below 880 ºC. The same phenomenon occurred for
pyrrhotite. As illustrated in Figure 2.14 and Figure 2.15, a similar dark porous layer of iron
oxide was formed on the surface when the sample was heated to 880 ºC. By noting XRD
patterns, it is obvious that magnetite reoccurred at 880 ºC. Thus this porous layer is believed to
be magnetite formed from the decomposition of hematite represented by Reaction (2.11). There
were conditions causing the decomposition of hematite between 785 ºC and 880 ºC. As is
depicted in Figure 1.7, the formation of Fe3O4 is easier at higher temperature since the
predominant area for Fe3O4 expands to the right with the increase of temperature. It suggests
that the already formed Fe2O3 would thermally decompose to Fe3O4 if the O2 partial pressure
drops to a level at which Fe3O4 is thermodynamically stable. The melting of nickel sulfide core
above 800 ºC accelerated its oxidation and would have lowered the oxygen partial pressure
within the bulk of the sample bed to a level that gives rise to the decomposition of Fe2O3,
forming Fe3O4. The porous nature of the Fe3O4 is possibly due to the volume reduction of
around 2% for Reaction (2.11).
The top right corner of Figure 2.19 shows the isolated sulfide core with enhanced contrast. This
sulfide core is composed of alternating layers of two nickel sulfide phases, one of which is
Ni3S2, the other is possibly Ni7S6. These two phases were probably formed by the exsolution of
the molten sulfide core during quenching, which provides evidence for the presence of the
unquenchable compound Ni3±xS2 in Reactions (2.9) and (2.10). On the surface of the sulfide
core where O2 is accessible, a layer of NiSO4 was formed via Reaction (2.10). The extent of
Reaction (2.10) was eventually retarded by the NiSO4 layer formed on the surface of the molten
sulfide core due to the limited O2 diffusion rate through this dense sulfate layer to reach the
surface of the sulfide where the reaction proceeded.
52
Figure 2.19. BSE image of an oxidized pentlandite particle air-quenched from 880 ºC in the
TGA run. Textures of the sulfide core are exhibited with enhanced contrast.
Figure 2.20. BSE image of an oxidized pentlandite particle air-quenched from 880 ºC in the
TGA run.
Figure 2.21. BSE image of an oxidized pentlandite particle air-quenched from 950 ºC in the
TGA run.
54
NiSO4→NiO+SO2+1/2O2 (2.12)
Ni3±xS2(l)+(3.5±0.5x)O2→(3±x)NiO+2SO2 (2.13)
There is another endotherm at 850 ºC after Reaction (2.10). Samples were collected at 825 ºC
and 885 ºC from DTA runs for XRD analysis in order to evaluate this endotherm. XRD patterns
indicate the increase of NiO and decrease of NiSO4 contents after this endotherm. It becomes
then clear that this endotherm is also the thermal decomposition of NiSO4. This means a single
reaction, which is represented by Reaction (2.12), gave rise to two endotherms in this test. The
explanation for this is tentatively interpreted as follows. The melting of Ni3±xS2 and the
following intensive thermal oxidation represented by Reaction (2.10) raised the temperature of
the sulfide core dramatically well above the furnace temperature, even higher than 942 ºC.
Meanwhile, the formation of NiSO4 layer on the molten sulfide core from Reaction (2.10)
eventually terminates the oxidation reaction by inhibiting the O2 diffusion. The local
temperature of the sulfide core is already high enough for the quick decomposition of the NiSO4
layer around the core, leading to the endotherm at 850 ºC. With the proceeding of the DTA run,
while the furnace temperature was increasing, the local temperature of the sulfide core drops
due to the termination of the exothermic Reaction (2.10) as well as the occurrence of the
endothermic Reaction (2.12). Reaction (2.12) stopped as a consequence due to this temperature
drop, which is indicated by the termination of the endotherm at 850 ºC. The decomposition of
NiSO4 resumes when the temperature is increased to 942 ºC, which gives rise to the second
endotherm.
Sulfate formation and decomposition is the second focus of this study. As can be seen in Figure
2.11, sulfation and sulfate decomposition which played a key role in the sample mass change,
spanned over a wide temperature range. Massive sulfation started from around 500 ºC, leading
to the increase of mass till 733 ºC. Sulfate decomposition prevailed at relatively higher
temperatures, resulting in the mass loss after 733 ºC. It is surprising to see that the sulfation for
Cu is over 85% at 785 ºC. It is also worth mentioning that Mg containing species, which is
siliceous gangue materials in the concentrate, is also prone to sulfation. This is illustrated by the
increasing sulfation for Mg to over 50% at 950 ºC. NiSO4 is relatively hard to produce, which is
due to its dense nature tending to inhibit further sulfation. The quick increase in NiSO4 from 785
55
ºC to 880 ºC is because of the melting of nickel sulfide that leads to the acceleration of the
kinetics for sulfation, as discussed earlier. The percentage of iron sulfate formation is low,
which is probably caused by the low stability of the sulfates. Sulfate formation can take place by
two possible reaction routes, which can be shown by Reactions (2.14) and (2.15) in general [20].
The first reaction represents the direct oxidation of sulfide, whereas in the second reaction the
sulfate is formed by SO2 and O2. No attempts have been made to determine by which route these
various sulfates were formed in this study. Sulfate decomposition can be generally represented
by Reaction (2.16). Temperature ranges for sulfates formation and decomposition for various
species can be seen in Figure 2.11 and will be summarized later.
It is also found from this study that at relatively high temperatures, various metal sulfates tended
to form sulfate mixtures. Figure 2.22 illustrates a particle of NiSO4-MgSO4 mixture in the
calcine quenched from 880 ºC. Mixtures of CuSO4-MgSO4, NiSO4-CuSO4-MgSO4, and even
Fe2(SO4)3-NiSO4-CuSO4-MgSO4 were also found to have formed. This is possibly because
sulfates of individual particles which have physical contact tend to diffuse into each other and
form mixtures. The formation of sulfates mixture increases the thermal stability of each
individual sulfate by lowering its activity, resulting in an increase in its decomposition
temperature.
Figure 2.22. BSE image of a particle composed of a mixture of NiSO4 and MgSO4 quenched
from 880 ºC.
56
MeS+2O2→MeSO4 (2.14)
MeO+SO2+1/2O2→MeSO4 (2.15)
MeSO4→MeO+SO2+1/2O2 (2.16)
For the oxidation of the Raglan concentrate at 15 ºC/min in air, the reaction scheme is
summarized in Table 2.7 based on this study.
Table 2.7. Summary of the reaction sequence for the oxidation of the Raglan concentrate at 15
ºC/min in air.
57
a
Temperature determined from TGA results.
b
Temperature determined from DTA results.
c
Temperature determined from ICP results.
A set of TGA runs was also performed to evaluate the effect of sample size on its oxidation
kinetics. Samples were heated in argon to 760 ºC and oxidized by switching the feeding gas to 1
L/min air and maintaining this temperature. Mass changes in weight percent were plotted
against oxidation time, as shown in Figure 2.23. At the onset of the oxidation, mass changes
were normalized to 0. The initial rates of mass change at the onset of oxidation in both
mg/second and weight%/second were plotted as a function of both the sample size and the
sample bed thickness in the crucible in Figure 2.24. For the runs with sample sizes larger than 5
mg, the initial rate of mass change (mg/sec) is roughly the same and independent of the sample
size. This indicates that reactions did not take place uniformly within the bulk of the sample, but
instead occurred gradually downward from the top surface of the sample. This is because
diffusion of O2 through the bed plays an essential part in the progress of reactions. Due to the O2
consumption reactions taking place at a certain depth in the bed, the O2 partial pressure must
have dropped to near zero in the area below the reaction level. As a result, the reaction front
where most oxidation reactions were taking place propagated downwards, leaving the oxide
products above and the materials below unaffected. The thickness of the reaction front must be
thin compared with that of the whole sample bed.
58
0
o
760 C, 1L/min air
2MeS+3O2=2MeO+2SO2
-1
2MeO+2SO2+O2=2MeSO4 100.2mg
-2
Mass change (wt%)
50.2mg
-3
25.0mg
-4
10.1mg
-5 5.0mg
2.2mg
-6
0 2 4 6 8 10 12
Time (min)
Figure 2.23. Mass change in wt% vs. time with the variation of sample size.
-2.0
0.1 -2.2
-2.4
0.0 -2.6
-2.8
-0.1 -3.0
-3.2
-0.2 -3.4
-0.3 -3.6
-3.8
-0.4 -4.0
-4.2
-0.5 -4.4
0 500 1000 1500 2000 2500
Bed thickness (µm)
Figure 2.24. Initial rate of mass change vs. sample size and bed thickness.
59
The progress of reactions can be envisioned on the basis of the mass change curves in Figure
2.23. At the beginning of oxidation, the reactions occurring are mostly oxidation reactions
releasing considerable amounts of SO2, leading to the initial quick mass loss. O2 from above the
sample bed diffuses through the oxidized bed and reaches the reaction front, where most of it is
consumed, releasing SO2 as a gaseous product. The resulting SO2 needs to diffuse upwards
through the oxidized bed before it is swept away by the feeding gas. The resulting counter
diffusion action of SO2 and O2, as well as the presence of the oxides as catalysts for the
oxidation of SO2 forming SO3, provides an ideal atmospheric environment for the sulfation of
oxides. With the downward propagation of the reaction front, the thickness of the oxidized bed
increases, giving rise to the prevalence of the sulfation reactions as opposed to the oxidation
reactions, resulting in the mass increase. Most of the oxidation stops when the reaction front
reaches the bottom of the crucible. Without the presence of SO2, the sulfation reactions stop
immediately, indicated by the sudden transition of the mass change curves in Figure 2.23.
Regarding this scenario, it is reasonable that the duration of this oxidation process is
proportional to the sample size, as shown in Figure 2.23. A sample size of 5 mg marks a
transition in the kinetics of oxidation, because the slopes of the mass change curves of 5 mg and
2.2 mg are identical. For the runs with sample sizes no larger than 5 mg in Figure 2.24, the
initial rate of mass change (mg/sec) becomes proportional to the sample size, indicating that the
O2 diffusion through the bed of the sample did not control the reaction rate. Probably in this
case, the diffusion in the particle became the rate controlling step. The sample size of 5 mg, with
the bed depth of around 125 µm, should roughly be the size of the reaction front. Given the
particle size of the Raglan concentrate, it is fair to say that even around 5 layers of the
concentrate particles would reduce most of the O2 partial pressure during the oxidation of the
concentrate at 760 ºC.
2.4 Conclusions
The oxidation mechanism of a nickel concentrate was investigated by means of TG/DTA.
Reaction products at intermediate temperatures were analyzed by XRD, SEM/EDS, EPMA and
chemical analysis. A reaction scheme was deduced for its oxidation in air from ambient up to
60
1000 ºC at 15 ºC/min. Between 350 ºC and 550 ºC, the main reaction was the decomposition of
pentlandite forming monosulfide solid solution, as well as the oxidation of iron sulfide. Violarite
is also involved as a transitional by-product from the decomposition of pentlandite at around
500 ºC. Preferential oxidation of iron sulfide species from the resulting monosulfide solid
solution occurred over a wide temperature range up to around 700 ºC, forming Ni1-xS core with
iron oxide shell. The Ni1-xS core was then transformed into Ni3±xS2 at around 730 ºC. The
melting of the nickel sulfide core at 813 ºC accelerated further oxidation as well as the
formation of nickel sulfate, which lowered the O2 partial pressure in the bulk of the sample bed,
leading to the decomposition of Fe2O3 forming Fe3O4. Decomposition of NiSO4 at 942 ºC
exposed the remaining nickel sulfide core to the oxidative atmosphere, leading to the complete
oxidation of the sulfide. Sulfates of various metals started to form at around 500 ºC probably
due to the emission of SO2 which favors the sulfation by providing a suitable atmosphere. These
sulfates tended to decompose at higher and varying temperatures, depending on their individual
thermal stability. Mixtures of sulfates were formed at relatively high temperature, which
increased their individual thermal stability.
The reaction rate controlling step was the O2 diffusion through the bed of sample for TGA runs
with sample sizes larger than 5 mg. Most of the oxidation reactions took place in a reaction front
of around 125 µm in depth which propagated downwards in the sample bed due to the limited
access of O2.
2.5 References
[1] J.G. Dunn, C.E. Kelly, A TG/MS and DTA study of the oxidation of pentlandite, Journal of
Thermal Analysis, 18 (1980) 147-154.
[2] J.G. Dunn, C.E. Kelly, A TG/DTA/MS study of the oxidation of nickel sulphide, Journal of
Thermal Analysis, 12 (1977) 43-52.
[3] A.C. Chamberlain, The effect of stoichiometry on the thermal properties of violarite and
pentlandite, Ph.D. thesis, Curtin University, Perth, Australia, 1997.
[4] T. Kennedy, B.T. Sturman, The oxidation of iron (II) sulphide, Journal of Thermal Analysis
and Calorimetry, 8 (1975) 329-337.
61
[5] K. Niwa, T. Wada, Y. Shiraishi, Roasting Reaction of Ferrous Sulfide, JOM, 9 (1957) 269-
273.
[6] M. Aneesuddin, P.N. Char, M.R. Hussain, E.R. Saxena, Studies on thermal oxidation of
chalcopyrite from Chitradurga, Karnataka State, India, Journal of Thermal Analysis, 26 (1983)
205-215.
[7] G.A. Kolta, M.H. Askar, Thermal decomposition of some metal sulphates, Thermochimica
Acta, 11 (1975) 65-72.
[8] T.R. Ingraham, Thermodynamics of the Thermal Decomposition of Nickel(II) Sulfate: The
Ni-S-O System from 1000º to 1150º K, Transactions of the Metallurgical Society of AIME, 236
(1966) 1064-1067.
[9] T.R. Ingraham, P. Marier, Kinetics of the Formation and Decomposition of Nickelous
Sulfate, Transactions of the Metallurgical Society of AIME, 236 (1966) 1067-1071.
[10] T.R. Ingraham, P. Marier, Kinetics of the Thermal Decomposition of Cupric Sulfate and
Cupric Oxysulfate, Transactions of the Metallurgical Society of AIME, 233 (1965) 363-367.
[11] P. Masset, J.Y. Poinso, J.C. Poignet, TG/DTA/MS Study of the thermal decomposition of
FeSO4·6H2O, Journal of Thermal Analysis and Calorimetry, 83 (2006) 457-462.
[12] P.G. Coombs, Z.A. Munir, The decomposition of iron(III) sulfate in air, Journal of Thermal
Analysis, 35 (1989) 967-976.
[13] R.V. Siriwardane, J.A. Poston Jr, E.P. Fisher, M.-S. Shen, A.L. Miltz, Decomposition of
the sulfates of copper, iron (II), iron (III), nickel, and zinc: XPS, SEM, DRIFTS, XRD, and
TGA study, Applied Surface Science, 152 (1999) 219-236.
[14] A. Roine, HSC Chemistry, Outokumpu Research Oy, Pori, Finland, 2007.
[15] P.G. Thornhill, L.M. Pidgeon, Micrographic Study of Sulfide Roasting, Journal of Metals,
9 (1957) 989-995.
[16] M. Zamalloa, T.A. Utigard, The behaviour of Ni-Cu concentrate in an industrial fluid bed
roaster, Canadian Metallurgical Quarterly, 35 (1996) 435-449.
[18] G. Kullerud, Thermal stability of pentlandite, The Canadian Mineralogist, 7 (1963) 353-
366.
[20] J.G. Dunn, The oxidation of sulphide minerals, Thermochimica Acta, 300 (1997) 127-139.
62
[21] J.R. Craig, Violarite Stability Relations, The American Mineralogist, 56 (1971) 1303-1311.
[22] K.C. Misra, M.E. Fleet, Chemical Composition and Stability of Violarite, Economic
Geology, 69 (1974) 391-403.
[23] J.G. Dunn, V.L. Howes, The oxidation of violarite, Thermochimica Acta, 282–283 (1996)
305-316.
63
3 Leaching Behavior of the Roasted Nickel Calcine
3.1 Introduction
It is known from the previous chapter that the roasting mechanism could be rather complicated
due to the complex mineralogy of the nickel concentrate, the heterogeneous nature of the
reactions, as well as the varying local roasting conditions. Therefore, it is expected that after
sulfation roasting, some unwanted by-products (e.g. NiFe2O4, NiO) could co-exist with the non-
ferrous metal sulfates in the calcine. A leach procedure therefore needs to be developed to
maximize the non-ferrous metal recovery and minimize the dissolution of iron species from the
calcine.
The main minerals in the nickel sulfide concentrate are: pentlandite (Fe,Ni)9S8, pyrrhotite Fe1-
xS, chalcopyrite CuFeS2, and siliceous gangue materials. The main species that the roasted
calcine may contain are: hematite (Fe2O3), magnetite (Fe3O4), nickel oxide (NiO), nickel ferrite
(NiFe2O4), cupric oxide (CuO), pyrrhotite (hexagonal type Fe9S10, monoclinic type Fe7S8),
heazlewoodite (Ni3S2), iron nickel monosulfide solid solutions ((Fe,Ni)1-xS), NiS, bornite
(Cu5FeS4), and chalcocite (Cu2-xS), as well as the water soluble sulfates [1-3]. The leaching of
water soluble sulfates is rather straightforward without the involvement of chemical
transformations. However, there is a possible scenario where the sulfates formed are enclosed in
the oxides and cannot be accessed by the leach solution. This would lead to the loss of metal
sulfates into the leach residue after the leach process. The major focus of this study is, as a
result, to investigate the leaching behavior of the minerals in the calcine and the possible
interactions among the species present in the leaching system.
Extensive studies have been done to investigate the leachability of some of the species involved
in this work. The leaching behavior of sulfide and oxide minerals of interest can be mainly
affected by the following conditions.
1. pH
In non-oxidative acid leaching, protons provided by the acidic leachants are used to attack the
solid. An example is the Falconbridge matte leach process, in which Ni3S2 in the smelter matte
64
is dissolved in concentrated HCl as shown as Reaction (3.1) [4, 5]. The effect of the acidity on
the dissolution rate can be generally described by Equation (3.2). Majima et al. [6] found the
leaching rate of hematite (Fe2O3) is first order with respect to a(H+) in hydrolic acid or
perchloric acid solutions, and is of half order in sulfuric acid solutions. While the leaching rate
of NiO has an order of 0.66 with respect to the H2SO4 acid concentration [7].
Ni3S2+6H+→3Ni2++2H2S+H2 (3.1)
Leaching reactions involving the transfer of electrons or the change of the valence state of
components could respond to the change of redox potential of the leaching system. Copper
concentrate could be leached under acidic oxidizing conditions, which can be expressed as
Reaction (3.3) [8]. Another commercialized oxidative leaching example is the Sherritt Gordon
Ammonia leach process to treat nickel sulfide concentrate, shown as Reactions (3.4) and (3.5)
[9]. Redox potentials are observed to have large effects on the leaching of oxides of the
transition elements Fe, Co, Ni, and Cu, because they are capable of forming oxides of variable
stoichiometry [10]. For the leaching of iron oxides, dissolution rate could be increased by the
addition of a small amount of reducing agents (including sulfide ions) [10-14], and the presence
of oxidizing agents can significantly reduce its dissolution rate [15]. On the contrary, the
dissolution rate of NiO tends to decrease if the leaching solution contains reductive ions, and the
ion with stronger reducing ability has larger effects on decreasing the dissolution rate of NiO
[7]. A study conducted by Rodenas et al. [16] revealed that the dissolution of nickel ferrite
(NiFe2O4) could be accelerated by both reductants (Fe2+) and oxidants (O2) because of the
presence of appreciable amounts of both Ni(II) and Fe(III) in the dissolving interface of the
particles.
CuFeS2+4Ox→Cu2++Fe2++2So+4Ox- (3.3)
NiS+5H2O+8Ox→NiO+SO42-+10H++8Ox- (3.4)
The temperature dependence of the dissolution rate of certain mineral could be quantitatively
described by the Arrhenius correlation. Table 3.1 summarizes the apparent activation energy for
the acid leaching of some minerals of interest.
Table 3.1. Activation energies for the acid dissolution of some minerals.
Normality Activation Energy
Minerals Acid References
(N) (kJ/mol)
Fe2O3 HCl 1 77.7 [6]
Fe2O3 H2SO4 2 104 [6]
pyrrhotite HCl 2.87 29.3±2.9 [17]
CuO HCl 0.2 50.6 [18]
NiO prepared at 873K H2SO4 0.02 75.7 [7]
NiO prepared at 1273K H2SO4 2 64.8 [7]
Both the thermal history and the storage conditions may have a great impact on the leachability
of certain minerals. This especially applies to the leaching behavior of NiO. Studies have shown
that the dissolution behavior of NiO is sensitive to the conditions of preparation [7, 19, 20].
They found that NiO samples heated to higher temperature were less reactive towards both acid
and ammoniacal solutions. This phenomenon was tentatively explained by a decrease in the
surface area as well as a change in defect (e.g. dislocation) concentration near the surface,
because the dissolution of NiO begins from the defect site (Ni2+ vacancy) on the surface of the
oxide particle [20, 21]. Astonishingly, the dissolution rate of fresh NiO is found to be over 40
times faster than that of the stored NiO powder [19], which is due to the change on its surface
during storage by the adsorption of foreign atoms or molecules [20].
Elemental sulfur is, to some extent, resistant to oxidation in spite of high oxidizing potentials in
the direct leaching of sulfide minerals [8]. The formation of elemental sulfur could detrimentally
66
affect the dissolution of sulfide minerals. Pyrrhotite is a non-stoichiometric mineral with an Fe:S
ratio smaller than 1. Due to the excess of sulfur contained in pyrrhotite, elemental sulfur tends to
form on the surface of the particle as part of the acid leaching product, apart from H2S. And the
fraction of elemental sulfur in the product is dependent on the stoichiometry of the pyrrhotite
itself. This sulfur layer formed on the surface of particles inhibits the progress of leaching. If the
minimum leaching conditions, i.e. temperature and acidity of the leachant, are not met, leach
would pause as a result of the refractory effect of the sulfur layer, leaving a fraction of the
pyrrhotite undissolved [17]. In the leaching of NiS (millerite) using HCl solution, H2S gas
would be oxidized to elemental sulfur if oxidizing conditions were present, resulting in a
reduced dissolution rate [22].
In some cases of acid leaching of oxides, aside from the hydrogen ions, anions from the
leachants also play an important role. The mechanism of such effect may be the complexation of
metal ions, enhancement of proton co-adsorption, or labilization of vicinal Me-O bonds [16].
Majima et al. [6] found that due possibly to the adsorption of anions of the leachants onto the
mineral surface which determines the leaching rate, the dissolution rate of hematite differed
greatly when leached with different acids although a(H+) were adjusted to the same value. The
study conducted by K. Nut [20] shows that in the acid dissolution of NiO, the presence of Co2+
could promote the dissolution of NiO, while the cations with stronger reducing ability tend to
have a larger effect in decreasing the leaching rate of NiO. In the same study, it was also found
that the reducing ion (e.g. Fe2+) present in the solution would preferentially be adsorbed on the
sites where the dissolution of NiO initiates. As a result, the dissolution rate is largely affected by
only a small coverage of the surface by reducing ions due to this preferential adsorption
behavior. Majima and Awakura [23] found the acid leaching of hematite as well as cupric oxide
could be enhanced by the addition of NaCl to HCl solution or NaClO4 to HClO4 solution. Rather
than employing the adsorption theory, they explained this enhancement of leaching as a result of
an increase of a(H+) by the addition of the appropriate salts.
7. Pressure
67
An example is the Dynatec Process developed by Sherritt Gordon in Canada. The main reaction
is shown as Reaction (3.6), taking place at 115–150 ºC and about 2000 kPa oxygen pressure
[24]. Pyrrhotite is insoluble in water under normal conditions, but can be slowly dissolved in
water at 110 ºC and 200 kPa oxygen pressure via Reaction (3.7) [24].
2CuFeS2+4H++5/2O2→2Cu2++2FeOOH+4S+H2O (3.6)
FeS+2O2(aq)→FeSO4(aq) (3.7)
As an intrinsic property of the mineral, the stoichiometry has a great impact on its leachability.
Take iron oxides as an example, the reactivity of the oxides in HCl solution is of the order
FeO>Fe3O4>Fe2O3 [10]. By taking advantage of this effect of stoichiometry on the leachability,
Dyson and Scott [25] devised an activation procedure which renders the nickel concentrate more
acid reactive by eliminating excess sulfur from it in a reducing atmosphere (natural gas) under
high temperature. With no excess sulfur in the system, the possibility of forming a protective
sulfur coating on the particles is eliminated.
3.2 Experimental
Nickel concentrate as received is Raglan concentrate from Xstrata Nickel’s smelter in Sudbury
(see Section 2.2.1 for more information on this material). Roasting of the nickel concentrate was
conducted in the experimental setup illustrated in Figure 3.1. In each roasting test, Raglan
concentrate of 5 grams was put in a porcelain boat. In order to maximize the heat and mass
transfer rate between the sample and its local surroundings, the concentrate was spread evenly in
the boat to cover an area of 28 cm2 with a thickness of 1.5 mm. The boat was then placed in a
sealed tube inside a resistance heating furnace. The samples were heated under flowing argon
and were subsequently subjected to the roasting gas (dried air and/or SO2 mixed with N2). The
offgas was analyzed for its SO2 content before it was scrubbed to eliminate the SO2 and SO3 for
purging to the atmosphere. The SO2 concentration of the off-gas was measured using a gas
analyzer (ABB EL3020) and was recorded using a computer controlled data acquisition system
68
(FLUKE Hydra Series II). After roasting, the tube was withdrawn from the electric furnace to
ensure a fast cooling rate.
In order to investigate the effects of both the roasting and leaching conditions on the leaching
behaviors of calcines, three types of leaching tests were conducted: Dilute Acid Leaching
(DAL), Hot Water Leaching (HWL), and Concentrated Acid Leaching (CAL). Conditions of the
leaching experiments are summarized in Table 3.2. Leaching took place in a 250 mL flask
equipped with a water condenser on exhaust to minimize the vaporization losses of the solution.
For precise temperature control within ±1 ºC, the flask was placed in a hot water bath. 1 mL
aliquot samples of leach liquor were drawn from the flask after predetermined time intervals.
After leaching, the pulp was filtrated and the residue was dried. Roughly 0.079 g residue was
taken and digested in aqua regia. Both the leach liquor samples and the solutions from residue
69
digestion were analyzed by ICP-OES after appropriate dilution. Calcines and leach residues
were analyzed by optical microscopy, SEM/EDS, and XRD.
70
2Fe1-xS + (3.5-1.5x)O2 → (1-x)Fe2O3 + 2SO2 (3.8)
Figure 3.2. Temperature and SO2 concentration in the offgas during roasting of sample
Calcine650.
One partially roasted calcine (named Calcine650S) was prepared by roasting the Raglan
concentrate at 650 ºC under a gas mixture of 500 mL/min air and 500 mL/min 10% SO2
(balance N2) until no apparent SO2 was emitted from the sample. The last calcine (named
Calcine750) was produced by roasting the concentrate at 750 ºC with air, aiming to investigate
how the roasting temperature would affect the leaching behavior of the calcine. The degree of
sulfur elimination for Calcine750 is 91.8%. The temperature and the SO2 concentration in the
offgas are plotted against roasting time in Figure 3.4 and Figure 3.5 for Calcine650S and
Calcine750, respectively.
71
Figure 3.3. BSE image and elemental maps of Fe, Ni, and S of sample Calcine650.
72
Figure 3.4. Temperature and SO2 concentration during roasting of sample Calcine650S.
Figure 3.5. Temperature and SO2 concentration during roasting of sample Calcine750.
73
3.3.2 Leaching Tests
74
Figure 3.6. Progression of leaching for DAL of Calcine650.
Figure 3.7. Mass change, temperature and the SO2 concentration in the TGA test for sample
Calcine650.
75
MeSO4 → MeO + SO2 + 1/2O2 (3.10)
Based on the EDS analysis of the Raglan concentrate and the calcines, all the Mg and Al are in
the siliceous gangue materials. As is shown in Figure 3.6, almost all the Mg was leached in 4
hours. A large portion of Al was also leached although it only weighs 0.40% in the concentrate.
However, only around 50% Si was dissolved. This shows that the Mg and Al in the siliceous
gangue materials were preferentially attacked by the acid. The percentage of Cu that was
leached was also very high, and the dissolution took place primarily in the first 30 minutes,
corresponding to the sulfate leaching period. This points to the formation of a large amount of
copper sulfate during roasting. The major portion of the Ni was still in the sulfide cores in the
calcine based on the EDS analysis. After 48 hours leaching, over 90% of Ni was leached out,
indicating that the nickel sulfide cores were gradually attacked by HCl acid. Apart from nickel
sulfide, NiO and NiFe2O4 could also be leached, contributing to the high percentage of Ni
extraction. However, the leached percentage of sulfur remains constant although the nickel
sulfide core was gradually attacked. The possible reason is the formation of elemental sulfur
during the acid leaching of the nickel sulfide cores, which can be represented by Reactions
(3.11) and (3.12). These two reactions require the presence of oxidants in the leaching solution
(e.g. Fe3+, Cu2+). The leaching profile of Co is similar to that of Ni because of the similarity of
these two elements as well as the co-presence of Co as impurities with Ni in the pentlandite. Of
all the elements of interests, Fe has the least extent of leaching, which can be seen in Figure 3.6.
In sample Calcine650, most of the iron is in the form of Fe2O3 and NiFe2O4, accounting for its
low dissolution rate. The formation of iron sulfates is limited under the examined conditions due
to their relatively low decomposition temperature. Figure 3.8 shows the BSE and optical images
of the leach residue. The remains are mainly nickel ferrous ferrite (NixFe3-xO4) and siliceous
gangue materials. Most of the oxide rims in the partly oxidized pentlandite particles were
broken down into fragments as can be seen in the BSE image. In the optical image, the oxide
rim was seen to be relatively intact, in which the sulfide core is mostly leached with very little
remains (shown as the white yellow part).
76
Figure 3.8. BSE and optical images of the leach residue from the DAL of Calcine650.
XRD analysis was used to determine the mineralogical composition of the leach residue, with
the spectrum shown in Figure 3.12. The formation of elemental sulfur is confirmed by the
dominant peaks of S8 in the XRD pattern. It also shows the presence of another phase NixS6 as a
by-product. Based on its stoichiometry, this phase should be an intermediate product during the
oxidative transformation of nickel sulfide Ni3S2 to Ni2+ and elemental sulfur S8. The dissolution
of Ni3S2 forming Ni2+ and S8 requires an oxidant. It is clear that the main oxidant in the leaching
solution should be Fe3+ from the dissolution of iron oxides. In this scenario, Reactions (3.14)
and (3.15) are tentatively suggested as the mechanism of the dissolution of nickel sulfide in the
present study: nickel sulfide core (Ni3S2) is first oxidized by Fe3+ in the highly acidic solution
79
with the preferential dissolution of metallic ions into the solution, forming solid NixS6 on the
surface of the sulfide; the NixS6 is then further oxidized by the Fe3+ forming elemental sulfur.
Figure 3.13 schematically illustrates the dissolution of a Ni3S2 particle with the presence of the
oxidant Fe3+ in the HCl solution. NixS6 should be found as a layer beneath the elemental sulfur
rim. The non-stoichiometry of NixS6, as well as the rate controlling step which should be ionic
diffusion through the solid product, suggest a decreasing content of Ni in NixS6 from the sulfide
core to the surface, which is exhibited in Figure 3.13. The morphological feature of the
incompletely leached nickel sulfide with elemental sulfur rim is shown in Figure 3.14. The
NixS6 layer could not be clearly differentiated from elemental sulfur using EDS due to its low
content of Ni.
80
Figure 3.11. BSE image and elemental maps of Fe, Ni, and S of the leach residue from the CAL
of Calcine650.
81
Figure 3.12. XRD pattern for the residue from the CAL of Calcine650.
Figure 3.13. Schematic representation of the oxidative dissolution of Ni3S2 in HCl solution with
the presence of Fe3+.
82
Figure 3.14. Optical and elemental mapping images showing the morphological features of the
nickel sulfide core covered with elemental sulfur rim.
83
Figure 3.15. Dilute HCl acid leaching results of the sample Calcine650S based on the ICP
analysis.
In order to investigate the leaching behavior of calcine roasted at higher temperature, the
calcine, generated by roasting the Raglan concentrate at 750 ºC, was leached with dilute HCl
acid. The degree of sulfur elimination of this calcine is 91.8%, much higher than that of the
Calcine650. As is shown in Figure 3.16, 98% of sulfur in the calcine was leached as sulfates
within 0.5 hour, and the balance is sulfides, which indicates that the weight ratio of sulfates to
sulfides in the calcine increases with the increase of roasting temperature. Compared with the
results of the Leaching test 1, much lower percentages of elements of interest, i.e., Ni, Cu, and
Co, were dissolved at the beginning of the leaching process, because of the lower percentages of
these elements existing as sulfates in the Calcine750. The iron profile in Figure 3.16 is lower
than that in Figure 3.6, especially at the beginning. Apparently, this has no relation with the
formation of sulfates, since no iron sulfates were formed in both calcines. But at higher degree
84
of roast, a larger amount of the iron species in the concentrate would be oxidized to Fe2O3 than
to Fe3O4. A previous study by other researchers demonstrated the relative reactivity of hematite,
magnetite, and nickel ferrite in HCl solution is in the order Fe3O4>>Fe2O3>NiFe2O4 [19], in
other words, the leaching rates of these three compounds would be in the same order. The lower
fraction of magnetite in the Calcine750 than in the Calcine650 is partially responsible for the
lower leaching rate of iron in Figure 3.16 than that in Figure 3.6. Similarly, more nickel ferrite
forms when the concentrate is roasted at higher temperature, especially above 700 ºC. The
leaching rate of nickel ferrite is rather slow, which is another factor that contributes to the
slower leaching rate of nickel in this leaching test, apart from the fact that less fraction of nickel
sulfate exists in the Calcine750.
85
3.4 Conclusions
Water leaching and non-oxidative HCl leaching tests were performed on the calcines produced
by roasting the nickel concentrate under various conditions. Results show that all sulfates could
be leached by water within 30 min. Longer leaching time results in the slow precipitation of
Cu2+ possibly due to its reaction with the nickel sulfide in the calcine. In acid leaching of
calcine, the formation of elemental sulfur would inhibit the dissolution of the nickel sulfide in
the calcine. Substantial amount of iron species was also leached along with the dissolution of
non-ferrous metal species in the non-oxidative acid leaching of the calcine. Leaching the calcine
with stronger HCl solution is much faster but less selective. Calcine produced at higher
temperature was less susceptible to acid leaching due to the formation of more acid-resistant
compounds, such as NiFe2O4 and Fe2O3.
3.5 References
[1] V.M. Zamalloa, Mechanisms of roasting, reduction and smelting of Ni-Cu concentrates,
Ph.D. thesis, University of Toronto, Toronto, Ontario, Canada, 1995.
[2] M. Zamalloa, T.A. Utigard, The behaviour of Ni-Cu concentrate in an industrial fluid bed
roaster, Canadian Metallurgical Quarterly, 35 (1996) 435-449.
[3] D. Yu, T.A. Utigard, TG/DTA study on the oxidation of nickel concentrate, Thermochimica
Acta, 533 (2012) 56-65.
[4] P.G. Thornhill, E. Wigstol, G.V. Weert, The falconbridge matte leach process, JOM, 23
(1971) 13-18.
[5] G.N. Lewis, M. Randall (Eds.) Thermodynamics, McGraw Hill Book Company, Toronto,
1961, pp. 316-317.
[7] K. Bhuntumkomol, K.N. Han, F. Lawson, The leaching behaviour of nickel oxides in acid
and in ammoniacal solutions, Hydrometallurgy, 8 (1982) 147-160.
86
[9] F.A. Forward, Ammonia pressure leach process for recovering nickel, copper, and cobalt
from Sherritt Gordon nickel sulfide concentrates, Transactions C.I.M., 56 (1953) 373.
[10] M.J. Nicol, The non-oxidative leaching of oxides and sulphides: an electrochemical
approach, in: K.Osseo-Asare, J.D. Miller (Eds.) Hydrometallurgy: Research, Development and
Plant Practice, Proc 3rd Int Symp Hydrometall, 112th AIME Annu Meet, Atlanta, Georgia, 1983,
pp. 177-195.
[11] N. Valverde, Investigation on the rate of dissolution of metal oxide in aqueous solutions
with addition of redox couples and complexing agents, Berichte der Bunsen-Gesellschaft für
Physikalische Chemie, 80 (1976) 333-340.
[12] M.J. Pryor, The reductive dissolution of ferric oxide in acid. Part III. The mechanism of
reductive dissolution, Journal of the Chemical Society (Resumed), 0 (1950) 1274-1276.
[13] M.J. Pryor, U.R. Evans, The reductive dissolution of ferric oxide in acid. Part I. The
reductive dissolution of oxide films present on iron, Journal of the Chemical Society (Resumed),
0 (1950) 1259-1266.
[14] M.J. Pryor, U.R. Evans, The reductive dissolution of ferric oxide in acid. Part II. The
reductive dissolution of powdered ferric oxide, Journal of the Chemical Society (Resumed), 0
(1950) 1266-1274.
[15] K. Jibicki, Acid decomposition reactions on compounds and minerals in the Fe-Ni-Sulphide
system, Ph.D. thesis, University of British Columbia, Vancouver, Canada, 1974.
[16] L.A. García Rodenas, M.A. Blesa, P.J. Morando, Reactivity of metal oxides: Thermal and
photochemical dissolution of MO and MFe2O4 (M=Ni, Co, Zn), Journal of Solid State
Chemistry, 181 (2008) 2350-2358.
[17] T.R. Ingraham, H.W. Parsons, L.J. Cabri, Leaching of pyrrhotite with hydrochloric acid,
Canadian Metallurgical Quarterly, 11 (1972) 407-411.
[18] I.H. Warren, G.I.D. Roach, Physical aspects of the leaching of goethite and hematite,
Transactions of the Institution of Mining and Metallurgy, 80 (1971) C152-155.
[19] Z.-Y. Lu, D.M. Muir, Dissolution of metal ferrites and iron oxides by HCl under oxidising
and reducing conditions, Hydrometallurgy, 21 (1988) 9-21.
[20] K. Nut, On the dissolution behavior of NiO, Corrosion Science, 10 (1970) 571-583.
[21] J.M. Diggle, Dissolution of oxide phases, in: J.M. Diggle (Ed.) Oxides and oxide films,
Marcel Dekker, New York, 1973, pp. 285-386.
[22] M.C. Jha, J.R. Carlberg, G.A. Meyer, Hydrochloric acid leaching of nickel sulfide
precipitates, Hydrometallurgy, 9 (1983) 349-369.
87
[23] H. Majima, Y. Awakura, Leaching of oxides and sulphides in acidic chloride media, in:
Extraction Metallurgy '85 Symposium, Institution of Mining and Metallurgy, London, UK,
1985, pp. 607-627.
[24] F. Habashi, A textbook of hydrometallurgy 2nd ed., Laval University, Quebec, Canada,
1999.
[25] N.F. Dyson, T.R. Scott, Acid leaching of nickel sulphide concentrates, Hydrometallurgy, 1
(1976) 361-372.
88
4 Fluidized Bed Oxidation Roasting
4.1 Introduction
In an attempt to lower the environmental footprint of nickel processing, and as an alternative
process to treat nickel sulfide concentrate, a two-stage oxidation-sulfation roasting process
followed by leaching was proposed and investigated. Sulfation roasting to treat nickeliferous
sulfide ores or low grade concentrates was investigated in the period from 1960s to 1990s [1-
10]. In general, the sulfation roasting process suffers from the drawbacks of slow kinetics and
low recovery of non-ferrous metals, Ni in particular, due to the formation of nickel ferrite
(NiFe2O4) at high temperatures (>700 °C) which is resistant to sulfation. These studies in
general lack detailed investigation on the kinetics and mechanisms of the sulfation roasting of
nickel sulfide concentrate. In addition, with more stringent environmental regulations, a
renewed interest in further studying the prospects of the proposed technology is warranted. The
scope of the thesis includes two-stage oxidation-sulfation roasting, leaching of the calcines, and
further recovery of Ni from the leach residue through high temperature reduction. This chapter
and the following one focus on the optimization of the roasting steps using the fluidized bed
technique with the aim of maximizing recovery of valuable metals into the leach solution. The
recovery of Ni from the leach residue was also studied and will be discussed later. This chapter
presents an overview of the methods and findings from fluidized bed oxidation roasting of
nickel concentrates. The results from the investigation of the second stage, i.e. sulfation
roasting, are presented in the next chapter.
89
4.2.2 Experimental
Compared to oxidation roasting, sulfation roasting is characterized by slow kinetics and strong
dependence on temperature [11]. In order to achieve the highest possible reaction rates and
accurate temperature control during the roasting process, fluidized bed technique was employed.
A laboratory scale, batch-operated fluidized bed roaster was designed and constructed to allow
conducting both the oxidation and sulfation roasting experiments (Figure 4.1).
The diameter of the lab scale fluidized bed roaster is a very important parameter partly due to
the wall effect, part of the fluidized bed near the roaster wall being less mobile than the centre
90
due to the uneven distribution of the feeding gas [10]. If the diameter of the fluidized bed is too
small, the wall effect would be so prominent that it would result in poor fluidization and prevent
smooth operation [12]. On the other hand, the gas consumption would be higher with higher
fluidized bed diameter, given the same gas flow velocity in the bed. In the present work, a clear
quartz tube with inner diameter of 36mm with an expanded top (66 mm inner diameter) was
used as the roasting reaction vessel. All dimensions are shown in Figure 4.2. A porous frit was
fused in the quartz tube acting as the gas distributor. The purpose of this design is to lower the
velocity of the gas in the freeboard above the fluidized bed, thus alleviating the gas entrainment
of solid particles. The quartz tube was mounted vertically in an electric furnace as shown in
Figure 4.1. Its top opening was sealed with a brass cap to prevent the leakage of the offgas.
Roasting gas was fed from the bottom of the quartz tube. The offgas was cleaned in a gas
scrubber (Buchi Scrubber B-414) to eliminate SO2 before venting to the fume hood. A
pneumatic dispenser was used to feed the concentrate to the sealed chamber using nitrogen as
the carrier gas. During the roasting experiment, the temperature of the fluidized bed was
measured using a K-type thermocouple. The pressure drop across the bed and the porous frit
91
was measured by a pressure transducer (OMEGA PX302-015GV). The offgas was directed to a
multi-channel gas analyzer (ABB EL3020) to measure its SO2 and O2 contents. These above
data along with the feeding gas flowrate were recorded using a computer controlled data
acquisition system (FLUKE Hydra Series II). A platinum catalyst was mounted beneath the
porous frit only for the sulfation roasting experiments. Once the fluidized bed roasting
experiment was completed, the calcine was withdrawn from the quartz tube immediately. This
was accomplished using a vacuum pump which collects the calcine into a cyclone via a long
ceramic tube. Quick collection of calcine out of the quartz tube was necessary to prevent
decomposition of sulfates at high temperature.
Preliminary experiments were carried out to fluidize the Raglan concentrate but all failed.
Channels were formed within the bed through which the feeding gas escaped. The concentrate
was then fused together due to the immobility of the bed and the highly exothermic reactions
occurring in the bed. This failure of fluidization is due to the fact that the Raglan concentrate
falls into Group C in the Geldart classification of powders according to the particle size
distribution of the Raglan concentrate in Figure 2.3 and the Geldart classification in Figure 1.4.
The interparticle forces are greater than the force exerted by the up-flowing gas, resulting in the
formation of channels instead of the concentrate being fluidized. On the contrary, sand can be
easily fluidized and is classified in Group B. It is reasonable to believe that the fluidity of the
concentrate could be enhanced by mixing it with sand. The particle size range of the sand should
be different from that of the Raglan concentrate to render the effective separation of calcine
from the sand after roasting by sieving. Trial experiments were performed to maximize the
fluidity of the mixture of the Raglan concentrate and sand. The results showed that the particle
size range of the sand should be as close to that of the Raglan concentrate as possible, to avoid
the appreciable physical separation of sand and concentrate into separate layers during roasting.
As a result, the particle size range of the sand was 150–212 µm. In addition, the weight ratio of
sand to concentrate should be no less than around 4 to render a fluidized bed with enough
mobility to avoid large agglomeration due to local overheating. Based on these results, a batch
of 5 gram Raglan concentrate was mixed with 20 gram sand with the particle size range between
150 μm and 212 μm before roasting tests. Separation between the resulting calcine and the sand
can be conducted based on the particle size range difference between these two simply by
sieving.
92
Two series of oxidizing roasting tests were performed to study the effect of temperature and
roasting time on the extent of oxidation and reaction products. In the first series of tests, the
temperature was varied in the range 650–775 °C. The roasting was terminated by withdrawing
the calcine (and sand) when the SO2 concentration of the offgas dropped below 0.3 vol%, which
is an indication of the near-completion of the roasting reactions. While in the second set, the
temperature was maintained at 750 °C and roasting times from 1 to 13 minutes were examined.
The collected calcines were subjected to various analyses to evaluate the roasting performances.
(1) Raise the temperature of the water bath to 90 ºC. Set the stirring rate to be 250 rpm.
(2) Measure 15 mL deionized water and put it into a 150 mL flask. Put a magnetic stirrer in the
flask. Heat it in the water bath while stirring.
(3) Weigh around 100 mg calcine to be leached and grind it using the mortar and pestle for 10
minutes.
(4) Weigh 60 mg calcine that was grinded and put it into the deionized water for leaching for 30
minutes.
(5) After 30 minutes leaching, cool the leachate in tap water and filter it. Wash the flask and the
residue.
The leach residues were fully digested in water by firstly fusing them in sodium peroxide
(Na2O2). And the procedure is as follows:
(3) Blend the Na2O2 and the sample thoroughly using a platinum wire.
(4) Weigh 1 gram of Na2O2 and spread it on top of the mixture as a thin layer.
(6) Heat the crucible with a propane blow torch using the outer flame with a distance of around
0.5 cm between the bottom of the crucible and the tip of the inner flame, and time it.
(7) Swirl the sample after it melts. After around 3 minutes when no solid residue can be
observed, cool the crucible by taking it away from the flame and putting it on a clean Al plate.
(8) Put a Teflon beaker with a cover on a hotplate, and put the Zr crucible into the Teflon
beaker.
(9) Add water slowly using syringe needle drop by drop into the Zr crucible until no evident
sputtering can be seen when water is added.
(10) Slowly add 30 mL 18 wt% HCl into the Zr crucible drop by drop to avoid sputtering. When
water or acid is added into the Zr crucible, Teflon cover should always be placed on top of the
beaker to prevent the solution from splashing out.
(11) Wash the underside of the Teflon cover and the inner wall of the Teflon beaker, and also
the entire Zr crucible. (At this time, no residue should be seen.)
(12) Take the Teflon beaker away from the hotplate and allow it to cool. Transfer the solution
into a 100 mL volumetric flask and take to the volume.
(13) Take 0.67 mL solution with pipette and transfer it into a 10 mL volumetric flask and take to
volume for ICP analysis.
Solutions from both leaching and Na2O2 fusion were properly diluted and analyzed by
inductively coupled plasma optical emission spectrometry (ICP-OES) for chemical composition
determination and calculation of the percentages of the formation of water-soluble species based
94
on Eq. (4.1), in which W(Me, leach solution) represents the weight of species Me in the leach solution
and W(Me, leach residue) represents the weight of species Me in the residue calculated based on the
ICP results.
Samples were also analyzed by XRD using a Philips PW2273/20 diffractometer. Calcines were
further examined by scanning electron microscopy (SEM, JEOL JSM6610-Lv) equipped with
EDS detector (Oxford/SSD) for characterization of the morphology and the mineralogy of the
phases formed. Composition of the sulfide cores of the calcines was determined by a Cameca
SX50 electron probe microanalyzer (EPMA) equipped with 3 tunable wavelength dispersive
spectrometers. Operating conditions were 40º takeoff angle, and a beam energy of 20 keV. The
beam current was 20 nA, and the beam diameter was 1 µm. Elements were acquired using
analyzing crystals LiF for Fe Kα, Ni Kα, Cu Kα, Co Kα, PET for S Kα, and PC1 for O Kα. The
standards were hematite for O Kα, cobaltite for Co Kα, pentlandite for S Kα, Fe Kα, Ni Kα, and
chalcopyrite for Cu Kα. The counting time was 20 seconds for Fe Kα, Ni Kα, Cu Kα, S Kα, 40
seconds for O Kα, and 60 seconds for Co Kα.
95
the increase of temperature. The average particle size of the fluidized bed is in the vicinity of
170 μm. The Vmf is approximately 0.013 m/s given the roasting temperature of 700 ºC.
Figure 4.3. Calculated minimum fluidization velocity as a function of particle size and
temperature for the roasting of 20 g sand + 5 g Raglan concentrate using air.
The Vmf is further determined experimentally using the pressure drop method at 700 ºC. During
roasting, the pressure measured beneath the porous frit has two components: the pressure drop
across the bed and the pressure drop across the porous frit. In Figure 4.4, the pressure drop
across the porous frit is plotted as a function of the gas velocity when no feeding material is
added. 20 g sand and 5 g Raglan concentrate was then put in the quartz tube, and the pressure
beneath the frit was measured again and was also plotted in Figure 4.4. The pressure drop across
the bed was calculated by taking the difference of the two curves in Figure 4.4, which is shown
in Figure 4.5. As can be seen, the pressure drop across the fixed bed increases with the increase
96
of gas velocity. The transition from fixed bed to fluidized bed is denoted by the sudden drop in
the pressure drop. The experimentally determined Vmf is 0.06 m/s at 700 ºC read from Figure
4.5. This minimum fluidization velocity corresponds to a feeding gas flowrate of 1.1 L/min.
4500
Empty Bed
4000
20g Sand + 5g Raglan calcine
3500
3000
Pressure drop (Pa)
2500
2000
1500
1000
500
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Gas velocity (m/sec)
97
160
140
120
80
60
40
20
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
Gas velocity (m/sec)
Figure 4.5. Pressure drop across the bed vs. gas velocity.
Apparent terminal velocity is plotted against particle size and temperature in Figure 4.7.
Apparent terminal velocity is calculated by converting the terminal velocity at that particular
temperature to the velocity at room temperature (25 ºC) using PV=nRT. This graph provides
98
more information since at any given time, the apparent velocity (converted velocity at 25 ºC) in
the quartz tube is constant, which is determined by the flowrate of the feeding gas, regardless of
the temperature variation along the longitudinal axis of the quartz tube. From Figure 4.7, it is
obvious that the apparent terminal velocity is largely affected by the temperature. Figure 4.8
illustrates the temperature profile measured along the height of the quartz tube above the
fluidized bed maintained at 700 ºC. The temperature gradually decreases from the fluidized bed
temperature at 700 ºC to around 100 ºC with the increase of height above the bed, because, as
can be seen in Figure 4.1, the top part of the quartz tube is out of the electric furnace and is
cooled by a fan. The movement pattern of particles above the fluidized bed could be envisaged
based on Figure 4.7 and Figure 4.8. If the flowrate of the gas is fixed to result in an apparent
velocity (at 25 ºC) of 0.1 m/s for instance, the relationship between the temperature and the size
of the particles which just experience their terminal velocity at that specific temperature can be
described by the curve a–b in Figure 4.7. In the fluidized bed at 700 ºC, particles smaller than 72
μm (read from the a–b curve) experience a gas flow velocity which is larger than their terminal
velocity. These particles flow upwards as a result. When they reach a cooler region in the
freeboard, the larger particles among them start to drop because the gas velocities they
experience drop below their individual terminal velocities due to this temperature drop. Smaller
particles are still carried upwards until a much cooler region is reached. This phenomenon
continues until the particles reach the top of the quartz tube where only particles smaller than 35
µm exit the quartz tube. Larger particles fall back to a region where the temperature is hot
enough to result in the ascent of the particles again. This upward-downward circulation occurs
to the particles within certain particle ranges. Relatively smaller particles circulate in the upper
part of the freeboard, while larger ones do in the lower part. This idealized regular circulation
could be compromised by the natural convection brought about by the temperature gradient in
the freeboard as well as the forced convection resulting from the turbulence of the fluidized bed.
Overall, the temperature gradient of the freeboard helps alleviate the gas entrainment problem.
99
Figure 4.6. Terminal velocity vs. particle size and temperature for the roasting of Raglan
concentrate with air.
100
Figure 4.7. Apparent terminal velocity (25 ºC) vs. particle size and temperature for the roasting
of Raglan concentrate using air.
800
700
600
Temperature (oC)
500
400
300
200
100
0
0 10 20 30 40 50 60
Height above the fluidized bed (cm)
102
chamber via the ceramic tube by the vacuum pump and was collected into the cyclone, which is
shown in Figure 4.10 (g) and (h).
A heat transfer and energy balance calculation for the fluidized bed system were carried out
using the recorded data from this oxidation roasting test. Four heat transfer components were
involved in the fluidized bed system:
1. Heat transfer between the feeding gas and the fluidized bed;
2. Radiative heat transfer between the electric furnace and the fluidized bed;
900 1000 18
Temperature
Pressure 16
800
SO2 800
14
700
Temperature (oC)
12
600
Pressure (Pa)
SO2 (vol.%)
600 10
8
500 400
6
400 4
200
2
300
0 0
0 1 2 3 4 5 6
Time (min)
103
Figure 4.10. Images showing various stages of fluidized bed oxidation roasting test.
Figure 4.11. Yellowish elemental sulfur particles formed on the upper wall of the quartz tube
from the condensation of the sulfur vapor.
104
Feeding gas _ For the calculation of the heat transfer between the feeding gas and the fluidized
bed, the case is simplified by making the following assumptions: a) The feeding gas is pre-
heated to the target temperature at 700 ºC when travelling in the lower portion of the quartz tube
before entering the fluidized bed through the porous frit; b) The solid and gas phases are in
thermal equilibrium in the fluidized bed part of the tube.
Radiation _ The calculation of the radiative heat transfer is simplified by the geometry of the
system where the fluidized bed is totally surrounded by the walls of the electric furnace.
Therefore, Equations (4.2) to (4.4) [13] are employed for the calculation of the radiative heat
transfer rate (W) between the walls (including the heating elements) of the electric furnace and
the fluidized bed. It is necessary to make several reasonable assumptions to render the
calculation possible. Because the fluidized bed is totally surrounded by the walls of the electric
furnace, and the surface area of the fluidized bed is much smaller compared with that of the
walls, any radiation emitted from the surface of the fluidized bed will mostly be absorbed by the
furnace walls after several consecutive reflections among the furnace walls. With this particular
set of geometry, the furnace walls could be considered as a black body with an emissivity of 1 as
far as the fluidized bed is concerned. The emissivity of the fluidized bed would change during
the progression of the roasting due to the change of the surface chemistry and morphology of the
particles as well as the temperature change. Due to the selective oxidation behavior of the iron-
containing sulfide minerals (i.e. pyrrhotite and pentlandite), the surface would be iron oxide
most of the time during roasting. The emissivity of Fe2O3 varies from 0.82 to 0.87 in the
temperature range from around 400 K to 1400 K [14]. The emissivity of the fluidized bed (εbed)
is assumed to be constant at 0.9, which is higher than the above mentioned emissivity range of
Fe2O3 because of the powder nature of the fluidized bed. Also we assume that both the furnace
walls and the fluidized bed are grey bodies, meaning their absorptivities are independent of the
incoming spectral energy distribution. For the calculation of the surface area of the fluidized bed
which is exposed to the walls of the electric furnace, assumptions are made that the geometry of
the fluidized bed is a column and the average porosity of the fluidized bed is 0.7, which is a
reasonable number because the measured porosity of the fix bed of mixture of 20 g sand and 5 g
Raglan concentrate is 0.49. Based on these necessary assumptions, the heat transfer rate (W) at
any given time during roasting could be calculated.
105
1
𝑞𝑓𝑢𝑟𝑛𝑎𝑐𝑒−𝑏𝑒𝑑 = 𝐴 ∙ �𝑒𝑏,𝑓𝑢𝑟𝑛𝑎𝑐𝑒 − 𝑒𝑏,𝑏𝑒𝑑 � 1 (4.2)
�𝜀𝑓𝑢𝑟𝑛𝑎𝑐𝑒 +1�𝜀𝑏𝑒𝑑 −1
4
𝑒𝑏,𝑓𝑢𝑟𝑛𝑎𝑐𝑒 = 𝜎 ∙ 𝑇𝑓𝑢𝑟𝑛𝑎𝑐𝑒 (4.3)
4
𝑒𝑏,𝑏𝑒𝑑 = 𝜎 ∙ 𝑇𝑏𝑒𝑑 (4.4)
where qfurnace-bed is the heat transfer rate in W; A is the area of the fluidized bed which is exposed
to the electric furnace; εfurnace is the emissivity of the furnace wall; εbed is the emissivity of the
fluidized bed; σ is a constant which is 5.669×10-8 W·m-2·K-4; Tfurnace is the furnace temperature;
Tbed is the fluidized bed temperature.
Reactions _ The heat generated from the roasting reactions is calculated based on the SO2
profile in Figure 4.9 and the mineralogical composition of the Raglan concentrate in the
following way. Three main reactions are taken into consideration for the calculation, which are
Reactions (4.5) to (4.7). These three reactions all result in the formation of SO2, which is shown
quantitatively in Figure 4.9 as the SO2 concentration in the offgas. Assume at any temperature,
the contribution of each reaction to the concentration of SO2 in the offgas is determined by the
molar ratio of these three minerals in the Raglan concentrate. Then the amount of the oxidation
occurring in any given period of time is determined by the amount of SO2 formed from the
fluidized bed. However, the SO2 profile in Figure 4.9 does not represent the SO2 concentration
in the offgas immediately produced from the fluidized bed because of the relatively large
volume of freeboard in the upper part of the quartz tube where the offgas produced from a
certain period of time mix together. This back-mix phenomenon results in the delay of reading
as well as the deviation of SO2 concentration measured from that of the offgas just produced
from the fluidized bed. The assumption is made that the gas in the freeboard is well mixed due
to the convection resulting from the temperature variation. The corrected SO2 profile can be
calculated from the measured SO2 profile, both of which are plotted in Figure 4.12. Three partly
overlapped peaks appear in the corrected SO2 profile, representing three predominant reactions
within certain temperature ranges. In order to calculate the enthalpy for these reactions at a
certain temperature, the standard enthalpy of formation and heat capacity at that temperature for
all the species involved in Reactions (4.5)–(4.7) are required. The standard enthalpy of
formation for pentlandite is -(837.37±14.59) kJ/mol [15]. The heat capacity of pentlandite at
106
298.15 K is 442.7 J/mol·K [16], which is used for the calculation at high temperatures since no
heat capacity at higher temperature is available. Standard enthalpy of formation and heat
capacity of other species are cited from the HSC Chemistry [17]. Please refer to the Appendix
01 which is written in C language for the whole calculation process.
20
SO2 corrected
18
16
SO2 concentration (vol.%)
14
12
10
SO2 measured
8
0
0 1 2 3 4 5 6
Time (min)
Figure 4.12. SO2 concentration right above the fluidized bed calculated from the measured SO2
concentration.
Conduction _ The heat transfer between the wall of the quartz tube and the fluidized bed via
conduction involves three components in series: heat transfer from the fluidized bed to the inner
wall of the quartz tube; conductive heat transfer in the quartz tube; and heat transfer from the
outer wall of the quartz tube to the air in the furnace at 700 ºC. These three components are
illustrated schematically in Figure 4.13. The inner wall of the quartz tube can be treated as the
107
immersed surface in the fluidized bed. As a result, the calculation of the heat transfer between
the quartz tube and the fluidized bed could be performed using Equation (1.11). Thermal
conductivity and heat capacity of quartz are required for the calculation of heat transfer through
the quartz, which are given as Equations (4.8) [18] and (4.9) [19]. Natural convection takes
place in the air near the outer surface of the quartz tube. Equations (4.10) to (4.12) are used for
the calculation. These three conductive heat transfer components are interdependent in terms of
their individual contribution to the total heat transfer because the thermal properties of the air,
quartz and even the fluidized bed are dependent on their individual temperatures. Due to the
relative complexity, three iteration processes are required, which are shown in Figure 4.14 as the
flow chart for the calculation process. Please refer to the Appendix 02 for the whole calculation
process.
Figure 4.13. Longitudinal and cross sections of the fluidized bed system showing the
temperature gradient.
k (W/(m·K))=418.4*(3.83*10-6T+0.00163) (4.8)
108
Initialize Toq, Tiq
Ar, Re, Pr
Nu
hb
2πL (Toq − Tb )
q1 =
1 ln( r1 / r0 ) Tiq=Tiq2
+
r0 hb kq
q1 ln(r1 / r0 )
Tiq 2 = Toq −
2πLk q
N
(Tiq2-Tiq)/Tiq<=1E-9
Pr, Gr
Nu
ha
2πL(973.15 − Toq )
q2 =
1
r1ha
q2 ln(r1 / r0 )
Toq 2 = + Tiq Toq=Toq2
2πLk q
N
(Toq2-Toq)/Toq<=1E-9
q1≈q2 N
Figure 4.14. Flow chart for the calculation of the heat transfer rate through the wall of quartz
tube (please refer to Figure 4.13 and Appendix 02 for the meanings of the terms).
109
ℎ𝐿
𝑁𝑢𝐿 = 𝑘 (4.10)
𝑓
𝜈 𝐶𝑝 𝜂𝑓
𝑃𝑟 ≡ 𝛼𝑓 = (4.11)
𝑓 𝑘𝑓
Some calculation results are plotted in Figure 4.15 and Figure 4.16. The heat transfer
coefficients of the fluidized bed and the air around the quartz tube (taking into account its
natural convection), as well as the thermal conductivity of the quartz tube are plotted against the
fluidized bed temperature in Figure 4.15. The heat transfer coefficient of the fluidized bed is
much higher than that of the air around the quartz tube due to the turbulent nature of the
fluidized bed. The thermal conductivity of the quartz tube increases with the increase of the
fluidized bed temperature. Due to the natural convection of the air which contributes to its heat
transfer, the heat transfer coefficient of the air reaches a valley at 973 K which is the furnace
temperature. In order to quantitatively study the relative thermal resistivity of these three
components, the temperature drops within these three components, which are proportional to
their individual thermal resistivity, are plotted in Figure 4.16 as a function of the fluidized bed
temperature. Note that the total temperature drop at a specific fluidized bed temperature is
normalized to 100%. As can be seen, the air around the quartz tube accounts for more than 94%
of the total thermal resistivity when the fluidized bed temperature is below 973 K. The quartz
tube accounts for less than 1%, which is shown as the red stripe. Above 973 K, the temperature
drop within the fluidized bed is around 80% of the total temperature drop, meaning that the
thermal resistivity of the fluidized bed prevails. The heat transfer rate (in W) is also plotted,
revealing that the heat transfer via conduction mainly occurs when the fluidized bed temperature
is below 973 K.
110
9 3.0
340
8
320 Air 2.5
Fluidized bed 7
300
Furnace temperature
W/(m2.K)
280 6 2.0
k(quartz) W/(m.K)
h(air) W/(m2.K)
260 5
973K
1.5
240
4
h(fluid bed)
220
Quartz
3 1.0
200
2
180
0.5
160 1
140 0 0.0
400 600 800 1000 1200
Fluidized bed temperature (K)
Figure 4.15. Heat transfer coefficients (h) of the fluidized bed and the air around the quartz tube,
and the thermal conductivity (k) of the quartz tube as a function of the fluidized bed
temperature.
The heat transfer rates associated with reactions, radiation, feeding air and conduction are
plotted in Figure 4.17 against the roasting time for comparison. As can be seen, at the beginning
when the concentrate is just fed to the quartz tube, the main source of heat is the radiation due to
the large temperature difference between the furnace and the fluidized bed. Radiative heat
transfer gradually decreases due to the temperature increase of the fluidized bed. Exothermic
oxidation reactions prevail as the main source of heat afterwards due to the high temperature of
the fluidized bed as the main contributing factor. As a result of the reactions occurring in the
fluidized bed, the temperature increased above the furnace temperature. The radiation becomes
the main channel for the heat release out from the fluidized bed to the furnace until the
temperature gradually drops back to the furnace temperature. The feeding air and the conduction
contribute much less to the heat transfer compared with the other two means.
111
100 22
40 8
6
30
4
20
2
10 0
0 -2
400 600 800 1000 1200
Fluidized bed temperature (K)
Figure 4.16. Temperature differences (in percentage) between the main body of the fluidized
bed and the inner wall of the quartz tube (labeled as Fluidized bed), between the inner wall and
outer wall of the quartz tube (labeled as Quartz tube), and between the outer wall and the
furnace wall (labeled as Furnace air), against fluidized bed main temperature, shown as the
stacked areas. Conductive heat transfer rate through the quartz tube is plotted as the blue curve.
112
400 900
800
350
Temperature (oC)
700
300
600
400
200
Reactions 300
150 200
100
Heat transfer rate (W)
100
Radiation
0
50
-100
0 -200
-150 -600
0 1 2 3 4 5 6
Time (min)
Figure 4.17. Temperature profile and the heat transfer rates by reactions, radiation, feeding air
and conduction through the quartz wall.
113
around the quartz tube. However, the main thermal resistivity comes from the fluidized bed
above 973 K, not from the air around the quartz tube, which can be seen in Figure 4.16. Another
cooling method is to add drops of water into the fluidized bed. This is effective because the
vaporization of water is fast, ensuring a quick heat absorption. However, the addition of water
causes a serious problem that the water tends to condense on the upper part of the quartz tube,
which results in the adhesion of the calcine powders on it. Another potential issue is the removal
of sulfates from the calcine due to their dissolution into water, which is detrimental to the
kinetic study on sulfate formation. The last cooling method is to add a stream of sand into the
fluidized bed, which is proven to be the most effective way. The sand is firstly added in the sand
feeder (Figure 4.1) before roasting. During roasting, when the temperature exceeds the target
temperature, a stream of sand with controlled flow is applied to absorb the extra heat.
The temperature gradient was quantitatively evaluated by measuring the temperatures at various
locations in the fluidized bed during oxidation roasting. A mixture of 5 g Raglan concentrate
and 20 g sand was roasted at 700 ºC with the introduction of sand for temperature control.
Temperatures at three locations were measured by three thermocouples: Temperature#1, center
of the fluidized bed; Temperature#2, between the fluidized bed center and the wall of the quartz
tube; Temperature#3, close to the wall of the quartz tube. Temperature profiles are illustrated in
Figure 4.18. In this figure, the quick temperature drop after 1min is a result of the feeding of the
sample into the quartz tube. Because thermocouple #3 is close to the wall of the tube,
Temperature#3 is higher than the other two due to the radiation received from the muffle
furnace in this period of time. Fluctuation occurred after 2.5 min resulting from the manual
114
discontinuous addition of sand to suppress the overheating. Between 2.5 and 10 min, the
magnitudes of temperatures are in the order: Temperature#1 < Temperature#2 < Temperature#3,
meaning that higher temperature is observed at locations which are closer to the wall of the tube.
The highest temperature difference between Temperature#1 and Temperature#3 is roughly 20
ºC, which is observed at 3 min when the rate of sand addition is at its maximum. With the
progress of roasting, the temperature gradient in the fluidized bed became smaller. Based on this
evaluation, the temperature gradient in the fluidized bed can be considered insignificant and a
relatively uniform temperature could be achieved during roasting experiment.
750
700
650 720
Temperature#1
600
Temperature#2
Temperature (oC)
500
450 700
115
SO2 and O2 contents, and pressure drop across the porous frit and the fluidized bed were
continuously measured, and have been plotted in Figure 4.19, for the roasting test at 650 °C as
an example. Similar plots were obtained for the roasting tests at other temperatures, exhibiting
similar trends. As seen in Figure 4.19, at around 0.5 min, a rapid temperature drop and a
corresponding ΔP increase were recorded, marking the start of feeding. The temperature
recovered after reaching a low of around 300 ºC. The temperature rise corresponds with a
decline in O2 and an increase in SO2 indicating the progress of roasting reactions. The low in the
O2 curve is around 2 vol%, corresponding to the SO2 peak of around 13 vol%. The volume of
the freeboard, which is the volume above the fluidized bed in the quartz tube, is relatively large
compared with that of the fluidized bed. Therefore, the offgas was held up and mixed to some
extent in the freeboard during roasting before reaching the gas analyzer. As a result, the offgas
O2 content recorded in Figure 4.19 lags the actual value in the fluidized bed, as does the
measured SO2. It is believed that the lowest O2 content in the fluidized bed was below 2 vol%.
This indicates that at the time when the fluidized bed atmosphere was relatively starved of O2,
the roasting reactions were so fast that the O2 in the roasting gas was consumed immediately
when passing through the fluidized bed. The roasting rate controlling step would be the supply
of O2 to the fluidized bed, rather than the roasting reactions or the gaseous diffusion through the
oxide layers. At later roasting stage when the partial pressure of O2 in the fluidized bed was
high, the gas diffusion through the oxide layers could dominate and control the overall roasting
rate. The offgas SO2 profiles for all roasting tests with temperature variation are plotted in
Figure 4.20, which exhibits the similarities in terms of the progress of roasting at different
temperatures.
116
Figure 4.19. Temperature, offgas SO2 and O2 concentrations, and pressure drop for roasting test
at 650 °C.
Figure 4.20. Offgas SO2 profiles for the fluidized bed roasting tests by air with roasting
temperature variation.
117
For oxidation roasting with air under normal roasting temperatures, the equilibrium phases are
always oxides of various metals (Fe, Ni, Cu, Co) from a thermodynamic point of view.
However, due to the limitations imposed by kinetics (e.g. gas/solid diffusion, chemical
reactions) and the varying local gas environment, sulfides and even sulfates could be present in
the calcine products. XRD analysis was employed to evaluate the mineralogical composition of
the calcine products. Their XRD patterns are presented in Figure 4.21. As can be seen, the main
compounds in these calcines are hematite (Fe2O3), trevorite (NixFe3-xO4), nickel oxide (NiO),
monosulfide solid solution (mss), and heazlewoodite (Ni3S2). There is no pentlandite,
chalcopyrite, or pyrrhotite in the calcines. The content of hematite in the calcines does not
change much with the increase of roasting temperature from 650 °C to 775 °C since there is
little difference with regards to its peak intensities. This is because iron sulfide species in the
form of pyrrhotite, pentlandite or chalcopyrite could be preferentially oxidized at lower
temperatures [20]. Reaction (4.13) represents the complete oxidation of pyrrhotite below 650
°C. Figure 4.22 illustrates the morphology of partly oxidized pentlandite particles roasted at 650
°C. Iron sulfide was preferentially oxidized from pentlandite particles forming a hematite shell
around the nickel sulfide core (mss). The partial oxidation of pentlandite could be represented
by Reaction (4.14). Concentration profiles of a roasted pentlandite particle along the arrow in
Figure 4.22 were measured by EPMA and are illustrated in Figure 4.23. Gradients of the Fe and
Ni concentrations can be observed in the sulfide core. The centre of the sulfide core has higher
Fe concentration (7.5 at%) and lower Ni concentration (41.5 at%) than the its edge
(approximately 5 at% Fe, 44 at% Ni). These gradients indicate the outward diffusion of Fe to the
surface of sulfide core where oxidation occurred. The hematite layer formed on the surface has
approximately 2 at% Ni, suggesting that a small portion of Ni was also oxidized along with the
preferential oxidation of Fe species. Co was oxidized at a faster rate than Ni, resulting in the co-
presence of Co in the hematite layer.
The formation of trevorite and NiO is obvious at temperatures higher than 700 °C. Monosulfide
solid solution disappears at temperatures higher than 700 °C, which could be represented by
Reaction (4.15), and heazlewoodite becomes the only sulfur-containing species. The amount of
heazlewoodite in the calcines becomes smaller with the increase of roasting temperature, which
can be seen by the decreasing intensities of the heazlewoodite peak at 31.2° with the increase of
temperature. This suggests the oxidation of heazlewoodite forming NiO according to Reaction
118
(4.16). Based on the XRD analysis, more species in the calcine appear in the form of oxides at
higher temperatures and in the form of sulfides at low temperatures.
Figure 4.21. XRD patterns of the calcines for various temperatures (tv-trevorite, NiFe2O4; mss-
monosulfide solid solution, (Ni,Fe)1-xS; h-hematite, Fe2O3; NiO-nickel oxide; hz-heazlewoodite,
Ni3S2).
119
Figure 4.22. Roasted pentlandite particles at 650 °C with 3 L/min air (mss: monosulfide solid
solution).
Figure 4.23. Concentration profiles in a cross section of a roasted pentlandite particle (650 °C)
along the arrow in Figure 4.22 measured by EPMA.
120
Figure 4.24 reveals the degree of sulfur elimination as a function of roasting temperature. A
previous study [21] shows that the roasting rate was limited by the mass transfer of O2 to the
reaction interface through the oxide rims which were formed upon oxidation. Considering this,
the increase of sulfur elimination with the increasing roasting temperature is possibly caused by
the acceleration of O2 diffusion through the porous oxide layers at higher temperatures. A
previous study [20] has shown that different sulfide species in the Ni concentrate were oxidized
at various temperature ranges. Some sulfides were oxidized with a multi-step mechanism, which
is also temperature dependent. This temperature dependency of oxidation roasting reactions
could be another important factor that causes the dependency of the degree of sulfur elimination
on temperature.
In order to further understand the roasting behavior of the concentrate under different roasting
temperatures, roasted pentlandite sulfide cores were analyzed by EPMA for their chemical
compositions, which are shown in Figure 4.25 and Figure 4.26. Figure 4.25 illustrates the
chemical composition change in a ternary diagram. As can be seen, Fe contents of most of the
pentlandite derived particles are around 2 at%, i.e. they do not change with the increase of
roasting temperatures. This value appears to represent the lowest iron concentration in the
121
pentlandite by preferential oxidation. The Fe concentration reported in Figure 4.23 is
approximately 6 at%, which is much higher. This is because of its relatively large particle size
(~70 µm) and the relatively low roasting temperature (650 °C) employed. Our previous study
[20] has shown that smaller pentlandite particles have higher preferential oxidation rate of Fe
species, which is in good agreement with the results from this study. Sulfur content decreases
from around 50 at% to 44 at% when temperature increases from 650 °C to 700 °C, with a
corresponding increase in Ni contents from 48 at% to 54 at%. This change in sulfur contents
coincides well with the disappearance of mss with the general formula (Fe,Ni)1-xS above 675
°C, which is illustrated in Figure 4.21 and represented by Reaction (4.15). Further temperature
increase above 700 °C does not contribute to an obvious change in composition.
Figure 4.25. Change of the pentlandite sulfide core compositions as a function of fluidized bed
roasting temperatures.
122
Figure 4.26. Average Co contents of the roasted pentlandite sulfide cores as a function of the
fluidized bed roasting temperatures by EPMA.
During the oxidation roasting of the nickel concentrate, Co species were also partially oxidized
and removed from the sulfide cores along with the oxidation of the Fe species. Figure 4.26
indicates the Co content of the roasted pentlandite sulfide cores as a function of roasting
temperature. The average Co content in the pentlandite particles prior to roasting is 0.41 at%.
After roasting the concentrate at 650 °C, the Co content in the pentlandite sulfide cores dropped
to 0.29 at%. A substantial decrease in Co content in the roasted pentlandite particles can also be
observed between 650 °C and 700 °C. Above 700 °C, the Co content dropped at a much lower
rate.
A roasting sequence of metal species in the pentlandite can thus be tentatively suggested as
follows. Preferential oxidation starts with the oxidation of iron sulfide from the pentlandite
(Fe4.5Ni4.5S8) forming FexNi1-yS with around 2 at% Fe. Further roasting leads to the oxidation of
nickel sulfide species towards FexNi3-yS2, which is a more sulfur deficient phase. During the
decrease in sulfur content, substantial percentages of Co in the pentlandite is oxidized and
removed from the sulfide cores. For complete/dead roasting, the final oxidation should be the
oxidation of the remaining nickel sulfide to form nickel oxide.
123
In the industrial oxidation roasting practice, the oxygen utilization efficiency is close to 100%.
However, excessive amount of air was used for roasting in this study, resulting in high partial
pressure of O2 in the fluidized bed and in the offgas. This provides more favorable conditions
for sulfate formation compared with the industrial roasting practice. In order to evaluate metal
sulfate formation during oxidation roasting of this study, calcines were leached with hot water
for 30 minutes to produce leachates for ICP analysis. Results are plotted in Figure 4.27 as
percent extraction of elements from the calcines by water as a function of the roasting
temperature calculated based on Equation (4.1). As can be seen, iron sulfate extraction is close
to nil under these temperatures. Copper sulfate formation drops to close to zero at 675 °C.
Nickel as sulfate stays at around 20 wt% when the temperature is below 700 °C, above which it
drops substantially. Cobalt sulfate content drops continuously from 650 °C to 775 °C, but a
substantial drop occurs when the roasting temperature is above 700 °C. The magnesium sulfate
content remains relatively constant, at around 27 wt%. Soluble silicon content is constant as
well, at around 5 wt% of silicon in the feed. Smaller amounts of sulfates were formed at higher
temperatures because oxides were preferentially formed rather than sulfates.
Figure 4.27. Water soluble species (%) in the roasted calcines vs. roasting temperature.
124
4.3.3 Effect of Roasting Time
Another series of roasting experiments was performed at 750 °C to evaluate the effect of the
roasting time. Calcines were analyzed with XRD for their qualitative mineral compositions, and
their XRD patterns are provided in Figure 4.28. As expected and can be seen from this figure,
considerable roasting took place within one minute, resulting in the disappearance of all the
original sulfide minerals and formation of various oxidized phases, i.e. trevorite (NiFe2O4),
hematite (Fe2O3), mms ((Fe,Ni)1-xS), NiO, and heazlewoodite (Ni3S2). The preferential
oxidation of iron sulfide species forming iron oxides occurred within this first minute. Roasting
reactions could be represented by Reactions (4.13) and (4.14). Further roasting led to the
disappearance of mss after 4 min (Reaction (4.15)), as well as the formation of more oxides, i.e.
hematite, trevorite, and NiO.
Figure 4.28. XRD patterns for the fluidized bed roasting tests at 750 °C with roasting time
variation (tv-trevorite, mss-monosulfide solid solution, h-hematite, NiO-nickel oxide, hz-
heazlewoodite).
125
Figure 4.29 illustrates the degree of sulfur elimination as a function of roasting time at 750 °C.
Within 1min the sulfur removal reached 63%, indicating a high roasting rate at the beginning.
The rate of sulfur elimination then decreased until it becomes practically zero after around 8
min, where the curve reaches a plateau at around 79%. The remaining 21% sulfur in the calcine
is in the form of heazlewoodite based on the XRD analysis shown in Figure 4.28. Figure 4.30
shows the chemical composition change of the roasted pentlandite sulfide cores as a function of
roasting time. As can be seen, the iron content in the pentlandite particles is reduced to the
minimum 2 at% after 1 min (Reaction (4.14)). Longer roasting time leads to the further
oxidation of nickel sulfide (NiS) to form Ni3S2 (Reaction (4.15)). In Figure 4.31, the average Co
content of roasted pentlandite sulfide cores is plotted as a function of roasting time. As
observed, the oxidation rate of cobalt from pentlandite is relatively high at the beginning of
roasting then slows down as the roasting proceeds. The average Co content dropped to around
0.05% after 13 min.
Figure 4.29. Degree of sulfur elimination as a function of roasting time at 750 °C.
126
Figure 4.30. Fe-(Ni+Co)-S ternary diagram shows the composition change of the pentlandite
sulfide cores as a function of fluidized bed roasting time measured by EPMA.
Figure 4.31. Average Co concentrations of the roasted pentlandite sulfide cores as a function of
the fluidized bed roasting time measured by EPMA.
127
Calcines from these tests were also subjected to water leaching and ICP analysis. The results are
presented in Figure 4.32. An interesting observation is that within the first two minutes, only
Mg and Si were dissolved among all the metal elements. The formation of NiSO4 started after 2
minutes. This is possibly because within the first two minutes iron from the pentlandite particles
was preferentially oxidized, and the nickel sulfide would not likely be sulfated before the iron
content in the pentlandite reaches its minimum, i.e. 2 at%. Figure 4.32 also illustrates that no
sulfates of iron or copper could be formed at the temperatures as high as 750 °C under the
current roasting conditions.
Figure 4.32. Water soluble species (wt%) in the roasted calcines vs. roasting time at 750 °C.
4.4 Conclusions
A laboratory scale, batch fluidized bed roaster was constructed to study the oxidation roasting of
nickel sulfide concentrate. Roasting tests were carried out at temperatures between 650 °C and
775 °C. At all temperatures, oxidation roasting reactions mainly occurred within the first minute
and usually lasted no longer than 10 minutes. Roasting reactions proceed as following: The
128
reactions start with the preferential oxidation of iron sulfide species forming iron oxides until
the content of Fe in the pentlandite sulfide cores drops to 2 at%. The remaining nickel sulfide
core in the pentlandite particles is either NiS or Ni3S2 depending on the roasting temperature, i.e.
the nickel sulfide formed tends to be more sulfur deficient at higher temperatures. Formation of
sulfates of Ni and Co occurs after the oxidation of iron sulfide. Low temperature (e.g. 650 °C) is
favorable for the preferential oxidation of iron sulfide species while minimizing the formation of
nickeliferous oxides, i.e. trevorite and NiO. The formation of these phases is inevitable for the
roasting at temperatures higher than 700 °C. In the range of 650−775 °C no iron sulfate was
formed. A higher degree of sulfur elimination could be achieved at higher temperatures. Sulfate
formation was inhibited at higher temperatures at the cost of forming more oxides. Unlike the
oxidation of iron sulfide, the oxidation of Co species through diffusion is a slow process. The
oxidation of the nickel sulfide core Ni3S2 is the last step if the roasting temperature is high
enough.
4.5 References
[1] J.R. Boldt, Jr., Falconbridge Pyrrhotite Process, in: P. Queneau (Ed.) The Winning of
Nickel, Longmans Canada, Toronto, 1967, pp. 331-336.
[2] A.W. Fletcher, K.D. Hester, A New Approach to Copper-Nickel Ore Processing, AIME
Transactions, (September 1964) 282-291.
[3] A.W. Fletcher, M. Shelef, A Study of the Sulfation of a Concentrate Containing Iron, Nickel,
and Copper Sulfides, Transactions of the Metallurgical Society of AIME, 230 (1964) 1721-
1724.
[4] A.W. Fletcher, M. Shelef, The Role of Alkali Sulphates in Promoting the Sulphation
Roasting of Nickel Sulphides, in: Unit Process in Hydrometallurgy. Group C: Plant Operating
Practice - Economics - General, New York Gordon and Breach Science Publishers, 1963, pp.
946-970.
[5] T.C. Frankiewicz, U.S. Patent 4,110,106, Selective Sulfation Process for Partitioning
Ferrous and Non-Ferrous Values in an Ore, United States, 1978.
[6] M.C.B. Hotz, R.C. Kerby, T.R. Ingraham, The sulphation of nickel and cobalt ferrites and
sulphides by molten sodium pyrosulphate and sodium bisulphate, Canadian Metallurgical
Quarterly, 7 (1968) 205-210.
129
[7] P.J. Saikkonen, U.S. Patent 4,464,344, Process for Recovering Non-ferrous Metal Values
From Ores, Concentrates, Oxidic Roasting Products or Slags, United States, 1984.
[8] P.J. Saikkonen, J.K. Rastas, The Role of Sulfate Melts in Sulfating Roasting, in: 25th Annual
Conference of Metallurgists, Toronto, Ontario, Canada, 1986, pp. 278-290.
[9] P.G. Thornhill, U.S. Patent 2,813,015, Method of Roasting Metal Sulfide Concentrates in a
Fluidized Bed, United States, 1957.
[10] P.G. Thornhill, U.S. Patent 2,813,016, Method of Roasting Nickeliferous Sulfide
Concentrates in a Fluidized Bed, United States, 1957.
[11] H.Y. Sohn, R.P. Goel, Principles of Roasting, Minerals Sci.Engng, 11 (1979) 137-153.
[13] D.R. Poirier, G.H. Geiger, Transport Phenomena in Materials Processing, The Minerals,
Metals and Materials Society, Pennsylvania, 1994.
[14] G.G. Gubareff, J.E. Janssen, R.H. Torborg, Thermal Radiation Properties Survey,
Honeywell Research Center, Minneapolis, MN, 1960.
[15] L. Cemič, O.J. Kleppa, High temperature calorimetry of sulfide systems, Phys Chem
Minerals, 14 (1987) 52-57.
[16] G.A. Berezovskii, V.A. Drebushchak, T.A. Kravchenko, Low-temperature heat capacity of
pentlandite, American Mineralogist, 86 (2001) 1312-1313.
[17] A. Roine, HSC Chemistry, Outokumpu Research Oy, Pori, Finland, 2007.
[18] H.E. Seemann, The Thermal and Electrical Conductivity of Fused Quartz as a Function of
Temperature, Physical Review, 31 (1928) 119-129.
[19] B.S. Hemingway, Quartz; heat capacities from 340 to 1000 K and revised values for the
thermodynamic properties, American Mineralogist, 72 (1987) 273-279.
[20] D. Yu, T.A. Utigard, TG/DTA study on the oxidation of nickel concentrate,
Thermochimica Acta, 533 (2012) 56-65.
[21] M. Zamalloa, T.A. Utigard, The behaviour of Ni-Cu concentrate in an industrial fluid bed
roaster, Canadian Metallurgical Quarterly, 35 (1996) 435-449.
130
5 Fluidized Bed Selective Sulfation Roasting
5.1 Introduction
A two-stage oxidation-sulfation roasting process for treatment of nickel sulfide concentrate is
proposed. The previous chapter covered the first step, the oxidation roasting with an aim to
maximize oxidation of iron while minimizing the formation of nickel ferrite. In this chapter, the
second stage, the sulfation roasting step, is discussed with the evaluation of several parameters
that may affect the sulfate formation, including the roasting gas flowrate, sulfation roasting
temperature, the addition of Na2SO4, the sulfation roasting time, and the oxidation roasting
temperature.
5.2.2 Experimental
The same fluidized bed roaster described in the previous chapter was employed. For sulfation
roasting experiments, a column of alumina pellets coated with platinum was mounted beneath
the porous frit to catalyze the oxidation of SO2 by O2, forming SO3.
Calcines were also mounted in the resin, then ground and polished with oil-based diamond
suspension for examination under a SEM (JEOL JSM6610-Lv) which was equipped with an
EDS detector (Oxford/SSD).
133
which has access to the sulfation gas through the gas channels in the oxide layer(s). The
formation of sulfate on the sulfide surface is inhibited if there is a firm contact of the relatively
dense oxide layer which isolates the sulfation gas. The oxide layer(s) (NixFe2-xO3) formed is
thermodynamically stable under the current sulfation roasting conditions, which would not be
converted to sulfate. Based on the above discussion, the formation of NiSO4 is dependent on the
access of the sulfide surface to the sulfation gas, which is further determined by the porosity of
the oxide layer(s) and the spatial relation between the sulfide surface and the oxide layer(s).
Figure 5.3 illustrates the water leach residue of the sulfation roasted calcine. The sulfate layers
were removed after leaching, leaving large gaps between the oxide layer and the nickel sulfide
core.
Figure 5.1. Effect of the sulfation roasting gas flowrate on the formation of sulfates.
134
Figure 5.2. Sulfation roasted pentlandite particles with 1 L/min sulfation gas, exhibiting the
formation of thin NiSO4 layers on the nickel sulfide cores, and the formation of nickel iron
oxide.
Figure 5.3. Leach residue of the sulfation roasted calcine with sulfation gas flowrate of 1 L/min.
135
5.3.2 Effect of the Sulfation Roasting Temperature
The dependence of the sulfate formation on the sulfation roasting temperature was investigated.
For each test, a mixture of 5 g oxidation roasted calcine and 20 g sand was roasted in 1 L/min of
the sulfation gas (5% SO2, 19.95% O2, and 75.05% N2) at different temperatures, for 30
minutes. The sulfate formation results are shown in Figure 5.4. It is evident that the temperature
dependency varies for elements. Formation of CuSO4 shows a decreasing trend with the increase
of sulfation roasting temperature from 660 °C up to 740 °C. CoSO4 formation reaches a peak at
700 °C, while NiSO4 does so at 720 °C. No appreciable iron sulfate formation is observed in the
temperature range of investigation. The effect of the roasting temperature on sulfate formation is
two-fold. On one hand, the temperature increase would enhance the kinetics for sulfate
formation, especially for the formation of NiSO4 which has been shown to be slow. On the other
hand, the sulfates are less stable at higher temperatures, indicating that higher temperature
favors more oxidation rather than sulfation from a thermodynamic point of view. As a result, the
temperature preferable for the sulfation roasting would be the ones high enough to render fast
kinetics for the sulfates formation but not excessive as to cause substantial oxidation. High
temperature is not preferable for the sulfation of nickel species due to the higher tendency for
formation of nickel ferrite. The optimum sulfation roasting temperature based on this series of
experiments is 700 °C, which was adopted for further sulfation roasting experiments.
136
Figure 5.4. Effect of the sulfation roasting temperature on the formation of sulfates.
137
addition for the sulfation roasting under the current experimental conditions would be 10% of
the weight of the calcine.
The calcines were analyzed with SEM/EDS. Figure 5.6 gives the BSE images of the sulfation
roasted calcines with the weight ratios of Na2SO4 to calcine of 0.02, 0.05 and 0.10. The mixture
of the oxidation roasted calcine and 10% Na2SO4 prior to sulfation roasting is shown on the top
left corner for comparison. Nickel sulfide cores in the images can be distinguished by their
brightness. The sulfide cores are monosulfide solid solution (mss, (Ni,Fe)1-xS) with
approximately 2 at% Fe before sulfation roasting (top-left image). This phase will further be
represented by Ni1-xS for simplification since there is very little Fe in the phase. After sulfation
roasting, the sulfide cores are primarily a non-stoichiometric sulfur-deficient phase which can be
represented by Ni7S6 on average. This phase also has around 2 at% Fe. By comparison, it is
obvious that more nickel sulfide cores were converted to NiSO4 with the addition of higher
amounts of Na2SO4. In the bottom right image which shows the sulfation roasted calcine with
10% Na2SO4 addition, the conversion of nickel sulfide cores to NiSO4 is near completion.
Figure 5.7 shows a sulfation roasted pentlandite particle with incomplete conversion of the
138
sulfide core to NiSO4 (left) and one with complete conversion (right). Each NiSO4 core is
composed of large amounts of small grains, in which no Na2SO4 was detected. This suggests
that the effect of the addition of Na2SO4 on the conversion of nickel sulfide to NiSO4 must be
indirect, meaning it takes place without physical contact. One possible mechanism is that the
Na2SO4 acts as reservoir of SO3 by forming Na2S2O7 which is a strong sulfating agent,
represented by Reaction (1.19). If the Na2SO4 in the fluidized bed happened to have physical
contact with other sulfates, they would preferably form solid solution and/or binary sulfates,
resulting in a lowered melting temperature. Such phases could melt in the fluidized bed if their
melting point is below the roasting temperature, leading to the agglomeration of particles. For
example, Figure 5.8 illustrates a cluster of particles agglomerated by a complex mixture of
binary sulfates and sulfate solid solutions (Na, Ni, Mg, Cu, and Fe).
Figure 5.6. SEM images of the oxidation roasted calcine blended with 10 wt% Na2SO4 (top-
left), and sulfation roasted calcines with the addition of Na2SO4 of 2% (top-right), 5% (bottom-
left), and 10% (bottom-right) with regard to the weight of the calcine (mss: monosulfide solid
solution).
139
Figure 5.7. Incomplete (left) and complete (right) conversion of nickel sulfide cores to NiSO4
with the addition of Na2SO4.
Figure 5.8. Cluster of calcine particles agglomerated by sulfate mixtures with 5% Na2SO4
addition.
Figure 5.9. Effect of the sulfation roasting time on the formation of sulfates.
141
Figure 5.10. Calculated relative amount of Ni existing as sulfide and oxide (%) in the leach
residue against sulfation roasting time.
Figure 5.11. Partly oxidation roasted pentlandite particles under 500 °C, 550 °C and 600 °C.
143
Figure 5.12. Effect of the temperature of the oxidation roasting stage on the formation of
sulfates in the sulfation roasting stage at 700 °C.
The sulfation roasted calcines were examined with SEM/EDS. Results indicate the near-
complete conversion of nickel sulfide cores to NiSO4. However, a substantial amount of non-
stoichiometric nickel ferrite with varying Fe/Ni ratio was found in all samples. Since no
appreciable nickel ferrite was formed during the oxidation roasting stage at relatively low
temperature, it must have been formed in the sulfation roasting stage along with the formation of
sulfates, given a long roasting time (150 min) and relatively high temperature (700 °C). Figure
5.13 shows the morphologies of the sulfation roasted pentlandite particles. The top-left image
shows the calcine produced by direct sulfation of the Raglan concentrate, in which the iron
nickel oxide is evenly spread in the nickel sulfate matrix. This morphological feature is much
different from that of the other three images, in which the oxide and the sulfate are concentrated
in different regions with a defined boundary. This morphological difference is caused by the
effect of the first oxidation roasting stage. Most of the iron would be preferentially oxidized to
form layer(s) of iron oxides during the oxidation roasting stage, resulting in the physical
separation of Fe from Ni through chemical reactions and ion diffusion. On the other hand, one-
step direct sulfation roasting of the concentrate results in simultaneous formation of oxide and
144
sulfate, hence their inter-mixed distribution. The side effect is formation of a larger amount of
nickel ferrite. This explains why the nickel sulfate formation is relatively low by direct sulfation
roasting as shown in Figure 5.12.
Figure 5.13. Sulfation roasted pentlandite particles: single stage sulfation roasting at 700 °C
(top-left); two-stage roasting with the oxidation roasting temperature of 500 °C (top-right), 550
°C (bottom-left) and 600 °C (bottom-right).
145
plotted in Figure 5.14. Uncertainty limits based on 95% confidence level as well as the mean
values (average) are reported in Table 5.1.
100
Fe
90 Ni
Cu
80
Co
Sulfate formation (wt%)
70 S
Mg
60 Al
Si
50
40
30
20
10
0
1 2 3 4 5
Test #
Figure 5.14. Repeated sulfation roasting tests conducted under optimized conditions.
Table 5.1. Sulfate formation (%) and uncertainty limits for sulfation roasting tests under
optimized conditions.
Elements Sulfates formation mean value Uncertainty limits (95% confidence level)
Fe 4.66% ±1.22%
Ni 78.6% ±3.22%
Cu 90.9% ±1.54%
Co 91.0% ±4.48%
S 97.3% ±0.48%
Mg 87.3% ±1.74%
Al 41.5% ±4.27%
Si 7.59% ±1.22%
146
5.3.7 Mechanism of Sulfation
It is evident from the above results that the addition of Na2SO4 enhances the sulfation. Other
researchers have shown that the enhancement of sulfation results from the destruction of the
impervious NiSO4 film by Na2SO4, thus creating new reaction sites for sulfate formation [2]. In
the present study, no Na2SO4 was present in the sulfate cores of the sulfation roasted pentlandite
particles, based on the SEM/EDS analysis. Instead, the sulfate cores were almost pure NiSO4.
Na2SO4 was only observed in the sulfate mixture as the outer layer of each particle, which is
peripheral to the iron oxide layer that was formed during the oxidation roasting stage, as can be
seen in Figure 5.13. Since the way the Na2SO4 was mixed with the oxidation roasted calcine was
by making a slurry, the small amount of sulfates (NiSO4, MgSO4) that was formed during the
oxidation roasting stage was probably dissolved into the Na2SO4 solution and formed a sulfate
mixture of low melting temperature after drying during the sample preparation. At the initial
stage of sulfation roasting, this sulfate mixture would melt immediately and dissolve any NiSO4
that is formed. However, during the progression of the sulfation roasting, it would eventually
reach a saturation point beyond which the sulfate melt could no longer dissolve more NiSO4
(e.g. once the composition crossed the liquidus line with a negative slope in the NiSO4-Na2SO4
phase diagram, Figure 1.12). The formation of NiSO4 beneath the oxide layer of each roasted
pentlandite particle requires SO3 as a reactant. SO3 has to first diffuse though the sulfate melt on
the surface of each particle via the S2O72--SO42- exchange reaction and the counter diffusion of
these two ions, then passing through the porous oxide layer. As could be clearly seen from
Figure 5.7 and Figure 5.13, the sulfate cores formed within the particles are composed of large
amounts of micro grains with cracks and crevices which provide channels for quick gas
transport. As a result, through these cracks and crevices, the SO3 could reach the surface of the
sulfide core where the sulfation reactions would take place.
Figure 5.15 shows the BSE image of another partly roasted pentlandite particle with a nickel
sulfide core. This sulfide core is composed of two phases, which can be seen as areas with two
gray levels after contrast enhancement as shown in the top-right corner. The darker inner core is
sulfur-rich and nickel-deficient monosulfide solid solution (Ni1-xS with 2 at% Fe), while the
brighter outer phase is Ni7S6. This indicates that the conversion of the sulfide core from Ni1-xS
147
to Ni7S6 accompanies NiSO4 formation, which could be represented by Reaction (5.1). In the
previous chapter, it has been shown that the sulfide cores of the partly roasted pentlandite
particles in the calcines are more sulfur-deficient at higher roasting temperatures. Therefore, the
driving force for Reaction (5.1) is likely the higher sulfation roasting temperature (700 °C)
compared with the oxidation roasting temperature (650 °C). Sulfur elimination from sulfide as
represented by Reaction (5.1) is achieved by first migration of sulfur ions from the interior of
the sulfide to the surface [1] where it could combine with O2. During the conversion of the
sulfide cores to Ni7S6, the sulfation reaction would proceed according to Reaction (5.2). The
SO2 required for the sulfation reaction could be provided from Reaction (5.1). Furthermore,
Reaction (5.1) would also result in the shrinkage of the sulfide core, which provides more space
for the formation of NiSO4 on the sulfide surface beneath the oxide layers. Evidence for this
could be found in Figure 5.15, in which the large shrinkage gaps between the sulfide core and
the oxide layers are filled with NiSO4. The oxygen required for the sulfation reactions could be
supplied by either the decomposition of SO3 or the inward diffusion of O2 from the gas. The
overall sulfation mechanism by diffusion could be schematically represented by Figure 5.16.
Figure 5.15. Sulfation roasted pentlandite particle with 20% addition of Na2SO4 at 700 °C for
30min. The sulfide core was isolated and shown in the top-right corner with enhanced contrast
illustrating the presence of two sulfide phases.
As represented by Reactions (5.1) and (5.2), the sulfation reactions are straightforward and do
not involve oxide(s) as intermediate product(s) based on the observation by SEM. This is
contradictory to the mechanism proposed by earlier researchers [2, 3]. In order to further
investigate this, another series of tests was performed. Two samples of nickel sulfide (main
phases: NiS, Ni3S2) were ground and polished to make flat surfaces. An appropriate amount of
Na2SO4 solution was then sprayed onto the surface of one sulfide sample and dried to have a
layer of Na2SO4 deposited on the sulfide with an average thickness of 10–20 μm. For
comparison, the other sulfide sample was not coated with Na2SO4. Both of these specimens
were suspended with the flat surfaces facing downwards in the roaster at 700 °C. Sulfation gas
(5% SO2, 95% air) of 3 L/min was introduced to the flat surfaces. Both of these two sulfide
samples were pulled out from the roaster after 30 min. They were then cut and prepared for
observation under SEM/EDS. Their morphologies are shown in Figure 5.17. As could be seen,
without the Na2SO4 coating, NiO was formed on the surface and propagated into the sample for
hundreds of micrometers. The black areas in the sulfide sample are crevices believed to have
formed due to the volume reduction during the oxidation of nickel sulfide. No NiSO4 was
formed, indicating that NiSO4 is thermodynamically unstable under the current sulfation
149
roasting conditions. For comparison, NiSO4 was formed on the surface of the sulfide cores
beneath the oxide layer as shown in Figure 5.2 during the sulfation roasting with the same
sulfation gas composition and temperature. This indicates that the presence of the porous oxide
layer maintains a stronger sulfation gas environment, i.e. higher partial pressure of SO3, beneath
the oxide layer. And the presence of the Na2SO4-NiSO4 sulfates melt as an outer layer as shown
in Figure 5.6 and Figure 5.7 further enhanced this effect. The other three images in Figure 5.17
illustrate the direct interaction between Na2SO4 and the sulfide sample under sulfation
conditions. In the top-right and bottom-left images, only a layer of sulfate(s) with varying Na/Ni
ratio was formed on the sulfide sample. There exists a layer of Ni2O3 in between the sulfide core
and the sulfates layer in the bottom-right image. The formation of Ni2O3 with a valence state of
Ni3+ could not possibly result from the oxidation of nickel sulfides, as it is thermodynamically
unstable under the present experimental conditions. The oxidation product would be NiO rather
than Ni2O3, if any. Misra and Whittle [4] studied the effects of SO2 and SO3 on the Na2SO4
induced hot corrosion of nickel. They also found the formation of a layer of Ni2O3 beneath the
porous NiO scale in the presence of Na2SO4 at 750 °C. They proved the formation of the
metastable Ni2O3 was from the decomposition of the NiSO4 component of the Na2SO4-NiSO4
melt, which was then slowly converted to NiO. As a result, it is believed that in the present
study as shown in the bottom-right corner of Figure 5.17, the layer of Ni2O3 was also from the
decomposition of the NiSO4 component of the outermost layer of Na2SO4-NiSO4 melt. It is
therefore evident that the sulfation of NiS does not involve the preliminary formation of NiO as
an intermediate compound. The NiS could be directly converted to NiSO4 by the sulfation gas
given favorable sulfation conditions. The role of the Na2SO4 in the sulfation is to quickly
dissolve the NiSO4 formed to form a sulfate melt and expose new sites on the sulfide for
sulfation. Also it accelerates the transportation of the sulfation gas (SO3) to the sulfide with the
S2O72--SO42- equilibrium in the sulfate melt.
150
Figure 5.17. Nickel sulfide samples uncoated (top-left) and coated with Na2SO4 (the other
three), sulfation roasted at 700 °C for 30 min.
One important observation of the present study is the formation of micro grains of NiSO4 with
cracks and crevices in the pentlandite particles which facilitate the formation of NiSO4 by
providing channels for the access of the sulfation gas. Those cracks and crevices are suggested
to have formed due to the combination of several factors: 1) The high temperature (700 °C) that
results in high kinetics for reactions; 2) The double sulfates melt-porous oxide layers
surrounding the sulfide core that maintain a very favorable gas environment for sulfate
formation; 3) The volume increases by approximately 150% as nickel sulfide (NiS) is converted
to sulfate, which introduces significant stresses into the sulfate formed; 4) Quick formation of
NiSO4 due to the easy access of sulfation gas to the sulfide surface through the already-formed
151
cracks and crevices so that more stress is added to the sulfate particle before the stress in the
particle is relieved by atomic rearrangement, resulting in crack formation; 5) Such cracks
provide more channels for the migration of sulfation gas to the surface of the sulfide. As a
result, the sulfation is self-sustained by the development of cracks and crevices.
152
produced from the ores mined from the Raglan mine located in the Nunavik region of northern
Quebec, Canada. It also contains PGM values as well as other precious metals such as rhenium
(Re), gold (Au) and silver (Ag) which are economically beneficial to recover as by-products. As
a result, it is essential to study whether these metals can be recovered into the leach solution
after water-leaching of the sulfation roasted calcine or they prefer to stay in the leach residue.
This is performed by analyzing several solid samples from the experiments by Inductively
Coupled Plasma Mass Spectrometry (ICP-MS). Three samples were analyzed, which are the
Raglan concentrate, the sulfation roasted calcine, the leach residue from the water-leaching of
the sulfation roasted calcine. Each powder sample was compressed with a die to form a pellet
before analysis. Figure 5.18 illustrates the results (counts per second) of a line scan on the
Raglan concentrate for different elements. The X-axis of the graphs in Figure 5.18 represents the
relative location on the scanned line. The large flutuation of the curves suggests the
inhomogeneity of the concentrate, meaning these elements are concentrated in some of the
particles in the concentrate. This indicates that the measurements should be repeated on different
locations of each sample for a large number of times and averages should be taken to represent
the bulk concentrations. Due to limited time available, four (4) replicates were performed on
different locations of each sample with the results plotted in Figure 5.19. The scattered data for
each replicate also indicates the inhomogenious nature of the sample. Averages of the repeated
measurements are also shown in the diagonal band for each element and each sample. The
analysis for Ru, Os and Ir failed and their results are not shown. The mass of solid increases by
30% when the concentrate is converted to sulfation roasted calcine (See Section 7.1. Mass and
Heat Balance), meaning a slight decrease by 23% in the concentrations of these elements in the
calcine is expected. The mass of the leach residue is only 60% of the original concentrate (See
Section 7.1. Mass and Heat Balance), meaning the concentrations of the precious metals should
increase by 67% if they all remain in the residue. From the results shown in Figure 5.19, it is
inconclusive to determine how much of the precious metals are leached into solution due to the
poor reproducibility of the measurements. However, the results could suggest a substantial
portion of the precious metals remain in the leach residue. The portion of these precious metals
that was leached into solution could be concentrated in the precious metal sludge during the
electrowinning process, which facilitates further recovery in the refineries. The precious metals
in the leach residue also necessitates further possible recovery.
153
Figure 5.18. ICP-MS measurements of some of the precious metals in the Raglan concentrate by
line scan.
154
concentrate,
calcine,
10
E S 0.093 residue,
AG 0.030
E R 0.38 0.090
AV 0.46
Concentrations (ppm)
0.13 1.0
1 0.052
15 0.20
7.0
2.2
2.9
1.1
0.1 1.2 6.1
1.1
1.2
Figure 5.19. Four (4) repeated measurements of the concentrations of some precious metal
isotopes by ICP-MS in three samples: Raglan concentrate (black square); sulfation roasted
calcine (red cross); and leach residue of the sulfation roasted calcine (blue star). Averages of the
repeated measurements are shown in the diagonal band for each element and each sample.
5.4 Conclusions
A two-stage selective oxidation-sulfation roasting process was investigated as an alternative
nickel production method from nickel sulfide concentrate using a batch fluidized bed roaster.
Several parameters were studied, i.e. the sulfation gas flowrate, sulfation roasting temperature,
the addition of Na2SO4, sulfation roasting time, and the oxidation roasting temperature. The
following conclusions were drawn.
155
(1) Under the optimized conditions of sulfation gas composition (95% air, 5% SO2), temperature
(700 °C), Na2SO4 addition (10 wt%) and time (150 min), the recoveries of valuable metals are
79% Ni, 91% Cu, and 91% Co. Only 5% Fe forms water-soluble sulfate.
(2)The residue from the leaching of calcine in water contains 49% Fe and 10% Ni.
(3) The Na2SO4 forms a melt with other sulfates on the surface of particles during sulfation
roasting which is believed to have maintained a high partial pressure of SO3 within each
particle, thus facilitating sulfate formation.
(4) Conversion of sulfide cores from Ni1-xS to Ni7S6, which resulted from the outer diffusion of
sulfur ions, accompanied the NiSO4 formation. Shrinkage gaps formed from this conversion
provided more space for further formation of NiSO4.
(5) The sulfation of the nickel sulfide is direct without preliminary formation of NiO as an
intermediate product.
(6) The NiSO4 formation is accompanied by the creation of cracks and crevices which allowed
the sulfation to proceed to near completion by providing channels for the inner diffusion of
sulfation gas to the sulfide surface.
(7) A large portion of the precious metals (PGM, Re, Au, Ag) remains in the leach residue.
5.5 References
[1] P.G. Thornhill, L.M. Pidgeon, Micrographic Study of Sulfide Roasting, Journal of Metals, 9
(1957) 989-995.
[2] A.W. Fletcher, M. Shelef, The Role of Alkali Sulphates in Promoting the Sulphation
Roasting of Nickel Sulphides, in: Unit Process in Hydrometallurgy. Group C: Plant Operating
Practice - Economics - General, New York Gordon and Breach Science Publishers, 1963, pp.
946-970.
[3] M.C.B. Hotz, T.R. Ingraham, Fused Sodium Pyrosulphate as a Sulphation Catalyst for NiO
and Co3O4, Canadian Metallurgical Quarterly, 5 (1966) 237-244.
156
[4] A.K. Misra, D.P. Whittle, Effects of SO2 and SO3 on the Na2SO4 Induced Corrosion of
Nickel, Oxidation of Metals, 22 (1984) 1-33.
[5] F.K. Crundwell, M.S. Moats, V. Ramachandran, T.G. Robinson, W.G. Davenport, Smelting
of Laterite Ores to Ferronickel, in: Extractive Metallurgy of Nickel, Cobalt and Platinum-Group
Metals, Elsevier Ltd., Great Britain, 2011, pp. 67-83.
[6] C.R.M. Rao, G.S. Reddi, Platinum group metals (PGM); occurrence, use and recent trends in
their determination, TrAC Trends in Analytical Chemistry, 19 (2000) 565-586.
157
6 Reduction of the Leach Residue
6.1 Introduction
An integrated pyro- and hydro-metallurgical approach to treat nickel concentrates through
sulfation roasting followed by leaching is investigated [1, 2] as an alternative processing route
with potential environmental benefits. In this process, the majority of the non-ferrous valuable
metals are leached with water after sulfation roasting, leaving iron in the leach residue as oxides.
However, approximately 20% of the nickel reports to the leach residue due to the inevitable
formation of nickel ferrites (NiFe2O4) during sulfation roasting [3]. This leach residue, which is
mainly composed of hematite (Fe2O3), nickel ferrite, and siliceous gangue, can be a high-grade
source of Ni (10 wt%) and Fe (50 wt%) for the production of ferronickel. As a result, the
reduction of the leach residue to produce ferronickel becomes an integral part of the sulfation
roasting route and was studied.
The reduction of hematite (Fe2O3) with H2 has been widely studied due to its role in production
of directly reduced iron. The temperature dependence of the reduction reactions has been
investigated extensively, indicating that depending on temperature, the reduction could adopt a
two-step mechanism, which is Fe2O3 → Fe3O4 → Fe [4-9], or a three-step mechanism, which is
Fe2O3 → Fe3O4 → FeO → Fe [7, 8]. Direct formation of Fe from the reduction of hematite
could also take place below 450 ºC [8]. The multi-step reduction mechanism is also affected by
the XH2O/XH2 ratio of the gas phase [10]. The effect of the partial pressure of H2 on the reduction
rate of iron oxides has been studied by previous researchers. It has been reported that for the
reduction of hematite to magnetite, the order of reaction with respect to H2 pressure was 0.036.
The reaction orders were 0.09 and 1.89 for the reduction of magnetite to wustite and wustite to
iron, respectively [11]. The reduction reaction is more likely controlled by interfacial chemical
reactions at low temperatures. An example of this is the reduction from hematite to magnetite
under fluidized bed conditions, which was limited by the gas-solid reaction in the temperature
range of 400–500 ºC [12]. Reduction at high temperatures tends to be governed by diffusion
especially when a substantial amount of reduction product is present. Reduction behavior is
greatly affected by the particle size, crystallinity and the conditions of reduction [7, 13].
Apparent activation energies for the reduction of hematite have been reported in literature,
158
showing that they depend not only on the purity and physical state of the raw material, but also
on the experimental conditions employed [8]. Generally the reduction rate of iron oxide with H2
is higher than that obtained by CO [8, 14].
Bahgat et al. [15] studied the H2 reduction of synthesized NiFe2O4 in the temperature range
between 800 ºC and 1100 ºC for the generation of nanocrystalline Fe-Ni alloy and found that the
reaction was governed by the combined gaseous diffusion and interfacial chemical reaction
mechanisms. The apparent activation energies were reported to be 31.85 and 24.57 kJ/mol for
the initial and latter reduction stages, respectively. NiFe2O4 compacts were sintered at 1000–
1200 ºC and were reduced at 900–1100 ºC by pure hydrogen by Khedr [16]. It was found that a
dense metallic layer (FexNiy) was formed surrounding the wustite layer which prevents gas
diffusion, resulting in incomplete reduction. A conclusion was also made that its reduction rate
was greatly influenced by both the firing and reduction temperature. Multi-step reduction of
synthetic NiFe2O4 was also observed to have a dependence on the reduction temperature [17]. In
the temperature range of 500–700 ºC, the rate controlling step was found to be the chemical
reaction at the metal-oxide interface.
Figure 6.1 displays the stability diagram of iron oxides in CO/CO2 atmosphere, as well as the
CO partial pressure in the CO-CO2-C equilibrium, which can be represented by the Boudouard
reaction (Eq. (6.1)) [18]. This figure illustrates that depending on the reduction temperature, the
reduction of Fe2O3 to metallic Fe could adopt a two-step (<550 ºC) or three-step mechanism
(>550 ºC) with the increase of the partial pressure of CO, admitting that in an actual Fe2O3
reduction process, the reduction mechanism could be largely different from the thermodynamic
prediction and be more complicated due to the contributing factors mentioned above. It has
generally been accepted that the indirect reduction of iron oxides with CO dominates the
reduction in a blast furnace as well as in the direct reduction processes producing sponge iron,
with solid-solid direct reduction playing only a minor role [19]. A conclusion has been made
that for those reduction processes with carbon or carbon monoxide as the reductant, the
reversible Boudouard reaction could exert a strong controlling influence on the reaction kinetics
and mechanism [20-22]. However, there are studies [23-25] in which NiO was reduced by
pyrolytic and natural graphite showing that the direct reduction played a major role, and the
reaction occurred mostly at the edge surface rather than at the cleavage surface of the graphite
159
due to the much higher number of active carbon sites on the former where reduction took place
preferably. This suggests that the nature of the carbon source (graphite, charcoal, coke, etc.)
could also have a significant impact on the kinetics and mechanism of reduction.
1.0
0.9
0.8 Fe
)
(g)=Fe+CO2(g
O
0.7 FeO+C
0.6
FeO
0.5 (g)
P(CO) (atm)
4+CO Fe3O4+C
0.4 e3O (g) O(g)=3Fe
1/4F e+CO2 O+CO2(g
)
F
0.3
=3/4
0.2
Figure 6.1. Stability diagram for Fe-O system under CO-CO2 atmospheres (solid lines) and CO
partial pressure established by the Boudouard reaction (dashed line) (Calculated using
thermodynamic data from HSC Chemistry [26]).
6.2 Experimental
6.2.1 Materials
The leach residue is primarily composed of hematite (Fe2O3 with varying concentrations of Ni
as impurities), nickel ferrite (NiFe2O4) and siliceous gangue. Table 6.1 shows the chemical and
160
mineralogical compositions of the material. Figure 6.2 displays the backscattered electron (BSE)
image of the leach residue [3]. It is comprised of three types of particles: (1) hematite particles
(e.g. particle a in Figure 6.2) originated from the oxidation of the pyrrhotite with its
characteristic columnar structure [27, 28]; (2) ring-shaped particles which are composed of
hematite and nickel ferrite, originated from the oxidation of pentlandite (e.g. particle b in Figure
6.2; Ni3S2 resides within some of the ring-shaped particles shown as bright areas); and (3)
silicate particles containing Fe, Mg, Al (e.g. particle c in Figure 6.2). Sulfur in the residue exists
in the form of nickel sulfide which is shown as the bright area within some of the ring-shaped
particles.
Chemical Fe Ni S Cu Co Mg Al Si O
composition (wt%) 49.3 10.5 1.1 1.0 0.06 1.1 0.9 5.1 30.94
162
6.2.2.3 Reduction with Graphite
Two series of reduction tests were also performed using graphite as the reductant. The graphite
powder was produced by grinding a purified (ash content <50 ppm) graphite rod and the fraction
below 25 µm was used. The powder and the leach residue were blended with the weight ratio of
2:5 to provide excess carbon for complete reduction. 70 mg of the mixture (50 mg residue + 20
mg graphite) was used for each TGA run, whereas the sample size for the DTA run was 21 mg
(15 mg residue + 6 mg graphite). Continuous heating of the mixture was conducted in both TGA
and DTA modes at 15 ºC/min to 1500 ºC in 1 L/min argon flow. The offgas was continuously
analyzed for its CO and CO2 content. The isothermal reduction tests were performed for a
temperature range of 800–1400 ºC in the following manner: The sample mixture was loaded in
an alumina crucible, and was first suspended from the micro balance well above the hot zone of
the TGA chamber. The chamber was then heated to the target temperature at the rate of 15
ºC/min while being purged with argon (1 L/min), to prevent air ingress. After the target
temperature was reached, the sample was quickly lowered into the chamber with continued
purge of 1 L/min argon. Because of the small sample size used, the sample was instantly heated
to the target temperature by radiation received from the chamber and the reduction was initiated.
Trial experimental results showed that due to the quick evolution of gaseous product (CO and
CO2) during the reduction process, samples could be blown out of the crucible by the gas and
cause an unexpected large mass loss. A thin layer of alumina wool was then used to cover the
sample in the crucible to prevent this.
Another issue identified in the trial experiments was that some ambient air was inevitably
introduced along with the sample into the chamber. The oxygen in the introduced air could
oxidize the graphite powder in the TGA crucible during reduction process and lead to
unexpected mass loss. Therefore, the oxygen has to be removed before it oxidizes the graphite
powder. This was performed by modifying the TGA setup by mounting two graphite tubes on
the ceramic tube and above the TGA crucible, which is shown in Figure 6.3. Due to the larger
surface area of the graphite tubes compared with that of the graphite powder in the crucible, the
oxygen in the chamber could be mostly removed by reacting with the tubes forming CO or CO2,
which would then be quickly removed from the chamber with the purge of argon. Isothermal
experiments at 800 ºC were performed to evaluate the effectiveness of the addition of the
graphite tubes in terms of the removal of oxygen from the chamber, the results of which are
163
shown in Figure 6.4. In this figure, test #1 shows the evolution of CO and CO2 in the isothermal
tests with empty crucible. The CO and CO2 peaks are from the removal of O2 by the graphite
tubes. Test #2 is the isothermal test with 20 mg graphite powder in the crucible. When compared
with test #1, the amount of CO2 evolved in test #2 is roughly the same, and there is a slight
increase in the evolution of CO. This indicates that the O2 mostly reacted with the graphite tubes
rather than the graphite powder, which illustrates the effectiveness of the graphite tubes in terms
of O2 removal. Test #3 is an isothermal test with a mixture of 20 mg graphite powder and 50mg
residue in the crucible. Much larger peaks of CO and CO2 appeared from the reduction of the
residue with graphite powder. The relatively small CO and CO2 peaks from the removal of O2
by the graphite tubes indicate that they would have negligible effect on the actual reduction
reactions.
Figure 6.3. Modification of the TGA setup by mounting graphite tubes on the ceramic tube.
164
1200 140
1100 120
#1 100
1000
#2
CO (ppm)
900 80
#3
800 60
40
700
20
600
0
500
-20
400
#1 -40
CO2 (ppm)
300 #2 -60
200 #3 -80
100 -100
0 -120
-100 -140
1 2 3 4 5 6 7 8 9 10
Time (min)
Figure 6.4. CO and CO2 evolution during the isothermal tests at 800 ºC with empty crucible
(#1), 20 mg graphite powder in the crucible (#2), and a mixture of 20 mg graphite powder and
50 mg residue in the crucible (#3), after the modification of the TGA setup.
165
6.3 Results and Discussion
6.3.1 Reduction with H2
166
0.4 90 8
2 Heat flow
0 0.2 Mass change 80 6
-2
0.0 4
Rate of mass change (wt%/min)
-4 70
-6 -0.2
Exo
2
Mass change (wt%)
-8 60
-18 -4
30
-20 -1.0
-6
-22 Rate of mass change 20
-1.2
-24 -8
-26 -1.4 10
-28 -10
-1.6 0
200 400 600 800 1000 1200 1400
o
Temperature ( C)
Figure 6.5. TGA/DTA results of the continuous heating tests at 15 ºC/min for the reduction with
15% H2.
167
dependence of rate on gas pressure can be expressed in the form of the power law, which is
shown as Eq. (6.7) [29]. Substituting Eqs. (6.6) and (6.7) in Eq. (6.5) yields Eq. (6.8).
dα
= k(T) ∗ f(α ) ∗ h(P) (6.5)
dt
- Ea
k(T) = A ∗ exp( ) (6.6)
RT
h(P) = P n (6.7)
dα -E
= A ∗ exp( a ) ∗ f(α ) ∗ P n (6.8)
dt RT
For the complete reduction of the oxides in the leach residue, the corresponding mass loss
should be 25.5 wt%. As can be seen from Figure 6.6 and Figure 6.7, the overall reduction rate
keeps increasing all the way from 350 ºC up to 1100 ºC, indicated by the larger slope of the
mass loss curve at higher temperatures. For the isothermal tests at 350 ºC, 400 ºC, 450 ºC and
500 ºC in Figure 6.6, a linear region appears right after the initial quick mass loss, indicating
that the reduction rate is constant and independent of the extent of reaction. This suggests that
the reduction rate is controlled by the zero-order chemical reaction. The differential form for the
zero-order reaction model is f(α)=1 [30]. From these isothermal reduction data, the apparent
activation energy was calculated to be 64.7 kJ/mol. Figure 6.8 illustrates the morphology of the
residue after 22% reduction at 400 ºC. It was observed that there was no apparent morphological
change compared to the leach residue before reduction. Little morphological change suggests a
non-topochemical reaction mode for the reduction at low temperatures.
168
0
-5 o
350 C
-15
450oC
-20
500oC
-25
0 20 40 60 80 100 120 140 160
Time (min)
Figure 6.6. Isothermal H2 (15%) reduction of the leach residue at 350 ºC, 400 ºC, 450 ºC and
500 ºC.
0
-2 400
-4
-6 500
Mass change (wt%)
-8
-10
-12 600
700
-14
-16 1200
-18
800
-20 1300
-22 1100 1500
-24 1000900
-26
0 5 10 15 20 25 30
Time (min)
Figure 6.7. Mass change (wt%) during the isothermal reduction of the leach residue by 15% H2
at various temperatures.
169
Figure 6.8. BSE image of the residue partly reduced (22%) at 400 ºC with 15% H2.
No linear region could be observed for the plots above 600 ºC in Figure 6.7, indicating that the
reaction rate changed with the progress of reduction. The overall reaction rate increased with the
increase of temperature until a transition temperature of 1200 ºC is reached. For the isothermal
test at 1200 ºC, the initial reduction rate is comparable to that at 1100 ºC. However, the reaction
slowed down dramatically after reduction for 90 seconds, compared with the reduction at lower
temperatures. For the isothermal tests at 1300 ºC and 1500 ºC, the initial quick mass loss can
also be observed, and the reduction after the initial period is faster compared to the one at 1200
ºC. The effect of the temperature on the reduction rate could be more easily observed by plotting
the time required for 90% reduction of the leach residue as a function of temperature, which is
illustrated in Figure 6.9. As can be seen, the time needed for 90% reduction continued to
decrease with the increase of temperature before the temperature reached 1200 ºC. 1200 ºC is
the transitional temperature at which the time needed for 90% reduction reached a maximum.
170
100
90
80
70
Time (min) 60
50
40
30
20
10
0
500 600 700 800 900 1000 1100 1200 1300 1400 1500
o
Temperature ( C)
Figure 6.9. Time required to reach 90% reduction of the leach residue with 15% H2 as a function
of temperature.
In order to determine the rate controlling step for the isothermal tests in the temperature range
between 600 ºC and 1100 ºC, a model fitting method was adopted. According to Eq. (6.5), the
rate of reaction dα/dt should have a linear relationship with f(α), since kT and hP are constant for
a particular isothermal reduction test with fixed H2 concentration. This linear relationship is only
observed when applying the 2D diffusion model, which is expressed as Eq. (6.9) [30] and
plotted in Figure 6.10. The trend of the reaction rate for the initial quick reduction within 90
seconds at 1200 ºC also shows a linear relationship in Figure 6.10. The slopes of these plots give
kT·hP according to Eq. (6.5). Eq. (6.10) was derived from Eqs. (6.5)–(6.8). According to this
equation, a linear trend would be observed by plotting ln(kT·hp) vs. the reciprocal of
temperature, which is shown in Figure 6.11. As can be seen from Figure 6.11, the Arrhenius plot
for 600 ºC deviated from the linear trend. This is possibly because the reduction at 600 ºC was
in the transition stage, i.e. reduction rate could be controlled by a combination of chemical
reaction and diffusion. Apparent activation energy could be calculated from the slope of this
linear trend, which is 34.1 kJ/mol. This value is much smaller compared with the apparent
activation energy for the zero–order chemical reactions below 600 ºC, i.e. 64.7 kJ/mol. The
171
values are however in good agreement with the activation energies measured by Shimakage et
al. [17]. They studied the hydrogen reduction of synthetic nickelferrite pellets and reported
activation energies of 66.1 kJ/mol below 600 ºC and 38.1 kJ/mol above 600 ºC.
Ea
ln(k T ∗ h P ) = ln (A ∗ h P ) − (6.10)
RT
o
1200 C
0.6
0.5 o
1100 C o
1000 C
dα/dt (min-1)
0.4
900oC
0.3 800oC
0.2 o
700 C
0.1
o
600 C
0.0
0.5 1.0 1.5 2.0 2.5 3.0
[-ln(1-α)]-1
Figure 6.10. Linear relations between the rate of reaction and the differential form of the 2D
diffusion model which is f(α)=[-ln(1-α)]-1 for the isothermal tests from 600 ºC to 1200 ºC.
172
-1.0 o
1200 Co
-1.5 1100 Co Ea=34.1kJ/mol
o 1000 Co
-2.0 1500 C 900 C o
-2.5 o
1300 C 800 C
o
-3.0 700 C
-3.5
ln(kThP)
-4.0 600oC
-4.5 o
500 C
-5.0 450oC
-5.5
-6.0
Ea=64.7kJ/mol 400oC
-6.5 350oC
-7.0
4 6 8 10 4 -112 14 16 18
1/T x10 (K )
Figure 6.11. Arrhenius plot for the isothermal reduction tests from 350 ºC to 1500 ºC (Data for
1200 ºC was calculated based on the initial reduction period).
Figure 6.12 exhibits the surface morphology of the fully reduced residue produced from the
isothermal tests at 900 ºC. It shows that the sample is still rather porous after reduction. Figure
6.13 shows the XRD pattern of the fully reduced residue, indicating the formation of mainly two
alloy phases, which are Fe3Ni2 and Fe10.8Ni. As mentioned earlier, the rate of reduction fits the
2D diffusion model, thus the gas diffusion through the solid product layer in each individual
particle has to be the rate controlling step for the isothermal tests between 600 ºC and 1100 ºC
as well as the quick initial reduction period of the isothermal test at 1200 ºC. Two other series of
isothermal tests at 800 ºC were performed with the evaluation of the effects of the gas flowrate
and sample size. Figure 6.14 illustrates the effect of the change of the gas flowrate on the
reduction of the residue. It appears that there is little effect on the mass loss of sample by
lowering the reducing gas flowrate from 1000 mL/min to 100 mL/min, meaning that the supply
of H2 to the sample with the flowrate of 1000 mL/min is large enough. The effect of sample size
on the rate of reduction at 800 ºC is illustrated in Figure 6.15. It is evident that the rate of
173
reduction (%) is slower for the larger sample. This indicates that the reduction reactions did not
take place uniformly in the whole sample in the TGA crucible, but a reaction front progressed
from the top to the bottom, which is the consequence of the reaction rate partially governed by
H2 gas diffusion through the sample bed. From the discussion above, the isothermal reduction is
most likely controlled by the combination of the gas diffusion through the pores of the sample
bed and gas diffusion through the solid product layer of individual particles (mixed control).
Figure 6.12. Morphology of the fully reduced residue (900 ºC with 15% H2).
174
1000 Fe3Ni2
Fe10.8Ni
800
Intensity
600
400
200
0
15 20 25 30 35 40 o 45 50 55 60 65
2 Theta ( )
Figure 6.13. XRD pattern for the fully reduced residue (900 ºC with 15% H2).
0
-2 100ml/min
-4 250ml/min
500ml/min
-6
1000ml/min
Mass change (wt%)
-8
-10
-12
-14
-16
-18
-20
-22
-24
-26
0 2 4 6 8 10 12 14 16 18 20
Time (min)
Figure 6.14. Effect of gas flowrate on the isothermal reduction of the leach residue at 800 ºC
(15% H2).
175
Figure 6.15. Effect of sample mass on the isothermal reduction of the leach residue at 800 ºC.
At 1100 ºC, the ferronickel alloy that was produced would partly melt and sinter, illustrated by
the cross section of the alloy particles in Figure 6.16. But it was still rather porous, which
provided channels for the gas diffusion through the sample bed in the TGA crucible. However,
when the temperature reached 1200 ºC, the silicate material melted and sintered with the alloy
that was formed, which is shown in Figure 6.17. The sample became non-porous, thus the
reducing gas had to diffuse through the molten sample bed in order for the reduction reactions to
proceed. This physical transition explains the two different stages occurring during the
isothermal reduction at 1200 ºC. When the temperature was raised to 1200 ºC, the silicate
materials in the sample melted immediately. However, the sample was still rather porous at the
initial stage of reduction because of the presence of a large amount of solid oxides. As a result,
during the initial reduction stage, the rate controlling step was the gas diffusion into the sample
bed through the pores, leading to a quick initial mass loss. As the reduction proceeded, the alloy
and silicates began to melt, leading to substantial shrinkage. The porosity of the sample bed
would eventually reduce to a critical point where gas diffusion became very limited. This would
in turn result in a much slower reduction rate. As a result, the reduction rate decreased
substantially after 90 seconds reduction at 1200 ºC. At the latter stage, the rate controlling step
176
would be the gas diffusion through the molten sample. This transition of the rate controlling
mechanisms was also observed in the isothermal tests at 1300 ºC. A 2D diffusion model was
applied to the isothermal tests at 1300 ºC and 1500 ºC. Arrhenius plots for these two
temperatures were also calculated and shown in Figure 6.11 for comparison with those for other
temperatures. Due to the insufficient data for this high temperature range, calculation of the
apparent activation energy was not possible.
Figure 6.16. Morphology of the residue fully reduced at 1100 ºC with 15% H2.
177
Figure 6.17. BSE image of the fully reduced residue from the isothermal test at 1200 ºC with
15% H2 (white area is the ferronickel alloy, grey areas are silicates with varying Fe contents).
In order to evaluate the order of reaction with respect to the partial pressure of the reducing gas
(n) in Eq. (6.7), one isothermal test was performed at 800 ºC with 40% H2 for comparison with
the isothermal reduction with 15% H2. The results are shown in Figure 6.19. The reduction rate
is much higher with 40% H2. A linear relationship was also obtained by plotting the rate of
reduction (dα/dt) as a function of [-ln(1-α)]-1, meaning that this isothermal reduction was also
controlled by gas diffusion. The order of reaction (n) was calculated to be 1.4. The pre-
exponential factors (A) could also be solved for different temperature ranges after n is obtained.
Accordingly, two empirical correlations for the reaction rates in the two regimes could be
established as shown in Table 6.2.
179
0
-5
-15
15%H2
-20
40%H2
-25
0 5 10 15 20 25 30
Time (min)
Figure 6.19. Comparison of the isothermal reduction with 15% H2 and 40% H2 at 800ºC.
Table 6.2. Overall reaction rate expressions for different temperature ranges.
Temperature Rate-controlling
Rate of reaction equations
ranges mechanisms
Pseudo-zero-order dα - 64673
350 ºC–600 ºC = 4012 ∗ exp( ) ∗ P1.4
chemical reactions dt RT
Mixed control of gas
diffusion through product dα - 34093
600 ºC–1200 ºC layer of individual particles = 66 ∗ exp( ) ∗ [− ln(1 − α )]-1 ∗ P1.4
and gas diffusion through dt RT
pores of sample bed
Gas diffusion through the
1200 ºC–1500 ºC ────────────────
molten sample bed
180
6.3.2 Reduction with CO
0 0 24 3500
22
-1 -5
Rate of mass change 20 3000
-10 18
Exo
-2
Rate of mass change (wt%/min)
16 2500
-15
14
Mass change (wt%)
-3 ∆T
-25 8 1500
-5
-30 Offgas CO2 Heat flow 6
4 1000
-6
-35 2
-7 0 500
-40
-2
-8 -45 -4 0
200 400 600 800 1000 1200 1400
Temperature (oC)
Figure 6.20. Continuous heating (15 ºC/min) of the leach residue in CO.
In order to identify what reduction reactions occurred in these two stages, samples were
collected from intermediate temperatures of 520 ºC and 800 ºC and were subject to XRD
181
analysis. Their XRD patterns are plotted in Figure 6.21 along with the pattern for the original
leach residue for comparison. As can be seen, the phases identified in the original leach residue
are hematite and nickel ferrite. Hematite disappeared after the sample was heated to 520 ºC with
the formation of substantial amount of magnetite (Fe3O4). The formation of an alloy phase
Fe10.8Ni is also observed. It becomes evident that one of the reactions in the first stage is the
reduction of hematite by CO to magnetite, which could be represented by Eq. (6.11). By
calculation, Eq. (6.11) could only cause 1.5 wt% mass loss. This suggests that the first small
CO2 peak at 424 ºC is most likely caused by this reduction reaction. The second reaction which
is represented by Eq. (6.12) is responsible for the formation of the alloy Fe10.8Ni and the
appearance of the second CO2 peak at 455 ºC. Due to the presence of Ni as impurities of varying
concentrations in the original hematite phase, reduction via Eq. (6.11) would result in the
formation of an iron-nickel oxide (NixFe3-xO4) instead of magnetite (Fe3O4). As a consequence,
the reduction product of Eq. (6.12) is an Fe-rich alloy phase (Fe10.8Ni) instead of pure Fe. Iron
oxide reduction is not complete at 520 ºC as suggested by the presence of substantial amount of
magnetite at this temperature, Figure 6.21. Relatively broad peaks for magnetite phase, which
partly overlap with those for NiFe2O4, can be seen at 520 ºC in Fig. 4. This is due to the
presence of various amounts of Ni in this phase (NixFe3-xO4) which caused the distortion of the
lattice structure to varying degrees, thereby shifting and/or broadening the XRD peaks. The
completion of iron oxide reduction is marked by the disappearance of the magnetite peak at 800
ºC, indicating that magnetite reduction proceeds in both reduction stages (400–520 ºC, 520–800
ºC). This is probably caused by the presence of two morphologies of the original hematite phase
which gives rise to the difference in their reduction kinetics. The first type has a rather porous
columnar structure (e.g. Particle a in Figure 6.2), which has easy access to the reduction gas.
The second type is relatively dense and co-exists with nickel ferrite in the ring-shaped particles
(e.g. Particle b in Figure 6.2). The reduction of the latter particles should be slower because of
the dense nature of solid phase(s). As a result, it is tentatively suggested that the reduction of the
porous magnetite occurred at around 455 ºC, and the denser magnetite in the ring-shaped
particles was reduced between 520 and 800 ºC.
182
Fe2O3 NiFe2O4
Fe10.8Ni Fe3O4
Fe0.64Ni0.36 Fe3C
(c)
Intensity
(b)
(a)
20 30 40 50 60 70 80 90
o
2 Theta ( )
Figure 6.21. XRD patterns for the leach residue (a), and samples collected after continuous
heating in CO to 520 ºC (b) and 800 ºC (c).
In the second reduction stage (520–800 ºC), nickel ferrite is also reduced, as indicated by the
disappearance of nickel ferrite and the formation of an alloy phase Fe0.64Ni0.36 in Figure 6.21.
The reduction of nickel ferrite could be represented by Eq. (6.13).
Further increase in temperature above 800 ºC caused little mass loss before 1200 ºC. On the
contrary, a slight mass gain is observed in this temperature range, which is believed to be caused
by the carbon deposition onto the sample and the crucible by the reverse Boudouard reaction
(Eq. (6.1)). The gradual drift of the base of the CO2 curve in Figure 6.20 also indicates the
occurrence of the reverse Boudouard reaction. The base of the heat flow curve in Figure 6.20
183
has drifted, which is caused by the physical changes of the sample during the course of
reduction, i.e. decrease in the heat capacity of the sample and/or increase in its thermal
conduction rate partly due to the shrinkage of the sample bed.
Further mass loss occurred after 1200 ºC, following an endothermic peak. A previous study on
the reduction of the leach residue by H2 showed that the silicate materials in the sample melted
at around 1200 ºC, suggesting that the reduction of the Fe from the molten silicates (Eq. (6.14))
should be responsible for this mass loss. The reduced product Fe should be in the form of an
alloy by merging with the existing alloy phases.
400
-5
Mass change (wt%)
500
-10
1200
-15 1300
600
700
-20 1100
900 800
1000
-25
0 5 10 15 20 25 30
Time (min)
Figure 6.22. TGA isothermal reduction of the leach residue with CO.
The kinetics of the isothermal reduction was evaluated by analyzing the mass change curves in
Figure 6.22. Eq. (6.15) [29] was adopted for this purpose. The Arrhenius equation (Eq. (6.6))
was used to describe the temperature dependence. The dependence of reduction rate on the
extent of reduction is represented by the kinetic model f(α), the form of which is determined by
the rate controlling mechanism. For an isothermal reaction, the integral form of the kinetic
model g(α) has the relationship with temperature (T) and time (t) as expressed by Eq. (6.16)
[29]. Isoconversional (model-free) method was firstly employed to calculate the apparent
activation energies without determining the kinetic models. Eq. (6.17) can be easily derived
from Eq. (6.16), in which tα,i represents the time required to reach a certain extent of reduction
185
(α) at temperature (Ti). The apparent activation energy at certain extent of reduction can be
calculated from the slope of the plot lntα,i vs. 1/Ti. The mass change curves for the isothermal
reduction tests at 700–1100 ºC were evaluated by applying Eq. (6.17) at various extents of
reduction (α, 0.2–0.9). The evaluation was not performed for α<0.2 because of the relatively
large error arising from the ambiguity in determining the starting time of the reduction from the
TGA curves. The apparent activation energies were calculated and plotted against α in Figure
6.23. The plot of lntα,i vs. 1/Ti is also shown in Figure 6.23 as an inset. As can be seen, the
apparent activation energy increases from 6.7 kJ/mol to 41.1 kJ/mol with the progress of the
reduction. This large variation of apparent activation energy suggests that these Ea are
composite values resulting from the tangled interplay of different steps/processes. These
steps/processes could be species-dependent chemical steps, gaseous reactants/products diffusion
through a solid product/reactant, adsorption–desorption of gaseous products/reactants on the
surface of the reacting solid, etc. [32, 33]. The apparent activation energies are relatively low,
indicating the rate controlling mechanism is very likely diffusion control rather than chemical
control. In order to determine whether the diffusion of CO through the pores of the sample bed
is the rate-controlling factor, two series of TGA experiments were performed. In the first series
of experiments, 50 mg leach residue was isothermally reduced at 800 ºC with varying flowrates
of CO. The results are plotted in Figure 6.24. As can be seen, three mass change curves overlap,
indicating that the transfer of CO to the surface of sample bed at 1 L/min was high enough and
did not adversely affect the reduction rate. The variation of sample size was then evaluated in
the second series of isothermal experiments at 800 ºC, the results of which are shown in Figure
6.25. A higher reduction rate can be observed with smaller sample size especially within the
first minute of reduction, indicated by the varying slopes of the mass change curves. This
suggests that the reduction did not take place uniformly in the sample bed, which resulted from
the limited mass transfer (gaseous transfer) in the pores of the sample bed during reduction. The
effect of this rate controlling mechanism gradually diminished with the progress of reduction, as
suggested by the parallel curves after approximately 1 minute.
dα
= k(T) ∗ f(α ) (6.15)
dt
α dα t − Ea − Ea
g(α ) ≡ ∫ =A * ∫ exp( )dt = A * exp( )*t (6.16)
0 f(α ) 0 RT RT
186
g (α ) E a
lnt α ,i = ln + (6.17)
Aα RTi
Figure 6.23. Variation of apparent activation energies as a function of extent of reduction (α) for
the isothermal reduction tests. lnt vs.1/T ×104 (isoconversional method) is also plotted as an
inset.
187
Figure 6.24. Effect of the variation of the flowrates of CO on the isothermal reduction of the
leach residue.
Figure 6.25. Effect of the variation of sample sizes on the isothermal reduction of the leach
residue.
188
Model fitting method was also used to further determine the rate controlling mechanism. Linear
trends are obtained between the reduction rate and f(α) for the isothermal tests in the
temperature range 500–1100 ºC by employing the 2D diffusion model, which is represented by
Eq. (6.9) [29]. These linear trends are exhibited in Figure 6.26 as an inset. For each individual
test, the 2D diffusion model applies only when α is higher than a certain value, which is shown
in Figure 6.26. This is because in the lower range of α, the reduction rate is predominantly
controlled by the gaseous diffusion in the pores which is discussed earlier. Since the 2D
diffusion model represents the diffusion in the solid product layer [34], this indicates that in the
higher range of α where the 2D diffusion model applies, diffusion of reactive species through
metallic alloys in individual particles controls the reduction rate. An Arrhenius plot is also
shown in Figure 6.26 yielding the apparent activation energy of 31.8 kJ/mol based on the slope
of the line for the temperature range 600–1100 ºC. This value is very close to that obtained with
hydrogen as the reductant, which is 34.1 kJ/mol for 600–1200 ºC. The plot for 500 ºC deviated
from the linear trend in Figure 6.26. This deviation is possibly caused by the partial reduction at
low temperatures due to the stepwise reduction mechanism (Eqs. (6.11)–(6.13)), whereas at
other higher temperatures the reduction is near completion. From the above discussion, gaseous
diffusion through the pores of the sample bed mainly controls the reduction rate at the lower
range of α, and the rate controlling factor gradually shifts to diffusion through the layer of alloy
products in individual particles at a higher range of α. A large variation of activation energies
with α in Figure 6.23 is a result of the combining effect of the two rate-controlling mechanisms
as well as their varying contributions to the overall reduction rate.
189
Figure 6.26. Arrhenius plot for the isothermal reduction tests between 500 and 1100 ºC. The
relationship between the reduction rate and the 2D diffusion model (f(α)=[-ln(1-α)]-1) is also
plotted as an inset.
The reduction products from the isothermal tests were examined by SEM/EDS. Figure 6.27
illustrates the microstructures of the porous hematite particles after reduction by CO at various
temperatures. The hematite was not reduced at 400 ºC and its columnar porous structure was
preserved (top-left). At 500 ºC, the hematite was reduced to form highly porous Fe particles
with submicron grains (top-right). The grains grow coarser and further apart when reduced at
higher temperatures (bottom-left and bottom-right). Some sintering could be observed in the
reduced hematite particles at 900 ºC.
190
Figure 6.27. Isothermal reduction of the porous hematite particles with CO at various
temperatures.
The change of morphologies of the ring-shaped particles reduced at various temperatures was
also examined and is presented in Figure 6.28. Nickel sulfide resided within some of the ring-
shaped particles, which can be seen in Figure 6.28. The particle at 400 ºC in Figure 6.28 was not
reduced, and the oxide ring is much denser compared with the hematite particle in Figure 6.27.
Reduction was obvious when the temperature reached 500 ºC and the ring formed Fe-Ni alloy
with micron-sized pores. At 600 and 700 ºC, the alloy rings were still rather porous, and the
residual nickel sulfide was partly converted to nickel-rich Ni-Fe alloy. This reaction is
tentatively suggested as the reduction of the nickel sulfide to form alloy with carbonyl sulfide
(COS) as the gaseous product, which is represented by Eq. (6.18). Crowe and Utigard [35] have
reported this reaction as being responsible for the reduction of Ni, Cu, and Co sulfides by CO.
Above 800 ºC in Figure 6.28, substantial sintering took place, and the degree of sintering is
greater at higher temperatures. For the sample reduced at 900 ºC, some Fe3O4 was found locked
in the alloy phase. This was caused by the immediate sintering of the alloys that were formed on
191
the surface of the oxide which limited the reduction rate of the oxide. The residual nickel sulfide
was fully reduced to nickel-rich alloy which sintered on the alloy ring. At 1100 ºC, the reduced
ring was adequately densified and formed a uniform alloy particle.
The silicates melted at 1200 ºC and all pores were removed. The alloy phase formed during this
process was distributed in the silicate melt as small particles of various shapes. Two silicates
were formed from the exsolution of the melt upon cooling, i.e. pyroxene which is rich in Fe and
Al, and olivine which is rich in Fe and Mg. The silicates phases formed were found to have
higher Fe content than the original silicates in the leach residue before reduction. This suggests
that part of the iron oxide in the leach residue must have dissolved into the silicate melt. This is
further supported by the fact that a large number of small regular-shaped Fe2O3 particles were
exsolved from the silicate melt upon cooling, which could be seen in the BSE image at 1200 ºC
in Figure 6.28. The partial dissolution of the hematite into the silicate melt which substantially
lowers the activity of hematite might be another critical factor that leads to the slow reduction
rate at 1200 ºC (Figure 6.22). In this case, the reduction is better represented by Eq. (6.14) in
which Fe is formed from the reduction of the silicate melt and merges with the existing alloy
particles. At 1300 ºC, more alloy was formed and the two silicate phases (pyroxene and olivine)
which were exsolved from the melt formed a laminar structure with no exsolution of Fe2O3
particles. This is probably caused by the greater degree of reduction at higher temperature which
results in an iron-deficient silicate melt. During reduction, gas bubbles of CO2 are formed,
which on solidification leave behind features such as those seen in Figure 6.29. Upon cooling
the melt, the gas bubbles shrink in volume, drawing the still-liquid silicate phase from the
already sintered alloy particles. On complete solidification, a spherical void and an alloy-free
halo remain at the bubble site. Because the exsolved pyroxene phase has a higher freezing
temperature than the olivine phase, the pyroxene solidifies before olivine, creating the unique
laminar structure with parallel plates of olivine separated by a gap, Figure 6.29.
192
Figure 6.28. Microstructure of particles reduced by CO under isothermal conditions (Px:
Pyroxene, silicate containing Fe, Al, Na; Ol: Olivine, silicate containing Fe, Mg).
193
Figure 6.29. Laminar structure of the silicates formed upon cooling the reduced residue from
1300ºC with a gas bubble formed in the melt (left) and interior of the bubble (right).
The Boudouard reaction, Eq. (6.1), was also expected to take place. The speculation is made
based on the fact that in Figure 6.30, the formation of CO2 as the gaseous reduction product
predominates below around 1000 ºC, above which the formation of CO prevails. This trend is in
accordance with the equilibrium of Boudouard reaction (Figure 6.1), although the offgas
CO/CO2 in Figure 6.30 did not reach equilibrium because of the non-equilibrium conditions
applied. The reduction slowed down until the temperature reached 1200 ºC, above which a
second stage quick reduction occurred. It peaked at 1300 ºC and ended at 1340 ºC. This second
stage of reduction results in the formation of CO as the gaseous product and corresponds to an
endotherm peak at 1235 ºC. This is believed to be induced by the melting of the silicates in the
sample which brings about the reduction of the molten silicate by the graphite, which could be
represented by Eq. (6.25).
195
0 0 4000 20
-2 -5
Mass change Rate of mass change 3500 10
-4 -10
Rate of mass change (wt%/min)
-6 -15 3000 0
-8
Exo
Mass change (wt%)
1500 -30
Offgas CO
-40
-18
-20 -45 1000 -40
-22 -50
500 -50
-24 -55 Offgas CO2
Figure 6.30. Continuous heating of the leach residue with graphite in TGA and DTA.
196
6.3.3.2 Isothermal Reduction
In order to study the reduction kinetics at different temperatures, isothermal reduction tests were
performed in the temperature range 800–1400 ºC with the results shown in Figure 6.31. The
reduction products from these isothermal runs were examined by SEM/EDS, EPMA, and XRD
to elucidate the reduction mechanisms. As can be seen in Figure 6.31, at low temperatures of
800 and 900 ºC, the reduction rate is slow and very limited reduction took place within 30 min.
Effective reduction only occurred when the temperature is higher than 1000 ºC. Two quick
reduction stages can be observed at this temperature, which took place during 0–2 min and 4–7
min, respectively. This possibly resulted from the sequential reduction of the porous hematite
particles and the relatively dense ring-shaped particles which are composed of both hematite and
nickel ferrite. The reduction at 1100 ºC was faster and near complete within 3 min. Two stages
of reduction were also observed for the isothermal runs at and above 1200 ºC. In this case, the
first stage is always the quick reduction of the oxides until approximately 25 wt% mass loss is
reached, while the following second stage is the reduction of the molten silicate. It is apparent
from Figure 6.31 that the molten silicate reduction stage is faster at higher temperatures,
indicated by the higher slope of the curves below -25 wt% at higher temperatures.
-5 800oC
900oC
-10
Mass change (wt%)
-15
-20
1000oC
-25
1100oC
-30
o
1200oC
1400 C o
1300 C
-35
-40
0 5 10 15 20 25 30
Time (min)
Figure 6.31. TGA isothermal reduction of the leach residue with graphite.
197
The alloy phases that were formed from these isothermal runs were analyzed by EPMA and the
results are presented in the ternary graph in Figure 6.32. The alloy phases identified at 900 ºC
under the electron probe are all nickel-rich Ni-Fe alloys, which are formed from the reduction of
the nickel sulfide. Previous study showed that the nickel sulfide (Ni3S2) melted at 813 ºC [27].
As a result, it was in its molten state during isothermal reduction at 900 ºC. The reaction could
be represented by Eq. (6.26). Figure 6.33 exhibits a partly reduced nickel sulfide particle
forming nickel-rich alloy with a clear boundary between the alloy and sulfide phases. Another
alloy particle is shown in Figure 6.34 as the product from the reduction of the nickel sulfide at
1000 ºC. It is evident from this particular morphology that both the alloy and the nickel sulfide
were in a liquid state during reduction, the sulfide covering the surface of the alloy drop.
60
40
)
ic%
900oC 1000oC
Si
1100oC 1200oC
om
(at
80 1300oC 1400oC
om
20
(at
ic%
Ni
100
)
0
0 20 40 60 80 100
Fe (atomic%)
Figure 6.32. EPMA analysis on the alloy particles formed from the isothermal reduction tests by
graphite.
198
Figure 6.33. Partial conversion of the nickel sulfide to alloy at 900 ºC.
Figure 6.34. Alloy surrounded by monosulfide solid solution (Mss, (Ni,Fe)S) reduced at 1000
ºC (Px: Pyroxene, silicate containing Fe, Al).
With further increase of temperature, more iron-rich alloys were formed from the reduction of
the hematite and nickel ferrite, which can be seen from Figure 6.32. Reduction of the silicates
started at 1200 ºC denoted by the presence of around 15 at% Si in the alloy phase. The silicates
after isothermal reduction were also analyzed by EPMA to reveal the progress of reduction,
which is plotted in Figure 6.35. The low concentration of FeO in the silicates at 1200 ºC
indicates that Fe is also reduced along with Si. The reduction of Fe from the silicate is more
complete at 1200 ºC and higher. The SiO2 content of the silicates also dropped with the increase
199
of temperature, indicating a larger extent of reduction at higher temperatures in terms of Si
removal. Figure 6.36 exhibits the nucleation and growth of Fe–Si alloy particles within the
silicate melt. The sizes of the particles range from nanometers to microns. The exterior Fe-Ni
alloy gradually absorbs some of these Fe-Si particles, particularly from the edge of the silicate,
to form a Fe-Ni-Si phase. Figure 6.37 shows a Fe-depleted silicate particle covered with a
relatively thick layer of Fe-Ni-Si alloy at 1200 ºC. This suggests that the reduction of silicate
melt could also proceed by the diffusion of carbon or possibly CO through the alloy.
0
100
20
900 oC
80
1000oC
1100oC
40 1200oC
)
ic%
1300oC 60
Fe
om
1400oC
O(
(at
ato
60
2
40
mic
SiO
%)
80
20
100
0
0 20 40 60 80 100
MgO+Al2O3 (atomic%)
200
Figure 6.36. Formation of SixFe alloy particles from the silicate melt at 1200 ºC (Px: Pyroxene,
silicate containing Al, Mg, Fe).
Figure 6.37. Reduction of Fe and Si from a silicate particle forming Ni-Si-Fe alloy at 1200 ºC
(Px: Pyroxene, silicate containing Al, Mg, Na).
Thermodynamic evaluation of the formation of the Fe-Si alloys from the reduction of the silicate
by graphite was performed using HSC Chemistry [26]. Changes in Gibbs free energy (ΔG) for
the reduction of Fayalite (Fe2SiO4) by graphite to form various alloys and gaseous CO were
201
calculated. Figure 6.38 depicts the minimum temperature to initiate the reduction (ΔG=0) as a
function of the molar ratio of Si/(Fe+Si) of the alloy phase formed. The solid products are
indicated near each point in the graph. As can be seen, the silicate could be reduced to form Fe
at temperature as low as 731 ºC from a thermodynamic point of view. Higher temperature is
required to form an alloy with a higher Si content. If pure Si and Fe are formed without the
formation of Fe-Si alloy, the minimum temperature to initiate the reduction is 1217 ºC. The
reduction of the silicates in this study suggests 1200 ºC as the minimum temperature to form
alloy phases. This is because the melting of the silicates at about 1200 ºC greatly enhances the
kinetics for reduction. Below 1200 ºC, the reduction of silicates is limited due to the slow
kinetics of the solid-solid reduction reactions.
1300
1200
Si,Fe
FeSi2,Fe
Temperature (oC)
1100 FeSi,Fe
Fe5Si3,Fe
1000
Fe3Si,SiO2
900
800
700 Fe,SiO2
Figure 6.38. Temperatures at which ΔG=0 for the reduction of FeSiO4 by graphite as a function
of the molar ratio of Si/(Fe+Si) of the alloy product.
With the increase of the isothermal temperature from 1200 to 1400 ºC, the composition of the
alloy particles becomes more uniform, suggested by the converging trend of the plots in Figure
6.32. This indicates that alloy particles have a greater tendency to merge and form larger and
more homogenized particles at higher temperatures, which is supported by the evidence from
both the SEM and XRD analysis. Figure 6.39 shows the spherical alloy particles produced at
1300 ºC. These particles could be as large as 100 µm in diameter. Figure 6.40 shows an alloy
particle 175 µm in diameter. Two phases were exsolved from the alloy melt showing different
202
grey levels. The interior brighter phase has more Fe and less Si. XRD analysis was conducted on
the samples produced from the isothermal runs at 1000 ºC, 1200 ºC and 1400 ºC with the results
shown in Figure 6.41. At 1000 ºC, the main alloys are Fe0.64Ni0.36 and Fe10.8Ni which were
produced from the reduction of the nickel ferrite and hematite, respectively. Due to the
reduction of molten silicate, Fe-Ni-Si alloy was formed as the major alloy phase at 1200 ºC with
little Fe0.64Ni0.36. The alloy produced at 1400 ºC is a relatively homogeneous phase of Ni-Si-Fe,
showing enhanced merging of the alloy particles.
203
Figure 6.40. An NixSiyFe alloy particle formed from the reduction at 1400 ºC.
Graphite
Fe0.64Ni0.36
Fe10.8Ni
FeNi
Fe2SiO4
NixSiyFe
1400oC
Intensity
1200oC
1000oC
40 45 50 55 60 65 o
70 75 80 85 90
2 Theta ( )
Figure 6.41. XRD analysis on the product from the reduction at 1000 ºC, 1200 ºC and 1400 ºC.
204
6.4 Conclusions
Reduction of the leach residue by hydrogen, carbon monoxide and graphite was conducted using
TGA and DTA. The reduced products were analyzed by SEM/EDS, EPMA and XRD. Effective
reduction of the residue by H2 could only be achieved with the temperature above around 350
ºC. At low temperatures (350–600 ºC), the reduction rate is low and is controlled by the
chemical reactions. Reduction occurred in a non-topochemical mode. Above 600 ºC and up to
1200 ºC, the reduction is faster and is controlled by the combination of gas diffusion through the
solid product layer of individual particles and gas diffusion through the pores of the sample bed.
The formed ferronickel alloy was sintered at around 1100 ºC. The silicate materials in the
sample melted at 1200 ºC and greatly reduced the porosity of the sample, resulting in substantial
reduction of the reaction rate. The reduction could be accelerated by increasing the partial
pressure of H2 with the reaction order (n) of 1.4. Apparent activation energies of the two
reaction regimes were 64.7 kJ/mol below 600 ºC and 34.1 kJ/mol for 600–1200 ºC.
Reduction by CO initiated at around 400 ºC during the continuous heating tests. The reduction
of the hematite phase took place in two stages with magnetite as the intermediate products.
Nickel ferrite was reduced mostly in between 520 and 800 ºC resulting in the formation of an
alloy phase Fe0.64Ni0.36. Significant sintering of the alloys was observed above 800 ºC, and the
extent of the sintering was greater at higher temperatures. Below 1200 ºC, the isothermal
reduction was firstly controlled by the gaseous diffusion in the pores of the sample bed. More
alloy products formed with the progress of reduction, resulting in the gradual shift of the rate-
controlling mechanism to the diffusion through the alloy products in individual particles. This
change of rate-controlling mechanism led to the increase of the apparent of activation energies
from 6.7 kJ/mol (α=0.2) to 41.1 kJ/mol (α=0.9). The siliceous gangue melted at 1200 ºC, which
caused the partial dissolution of hematite into the silicate melt and removal of pores from the
sample bed. This resulted in slow reduction kinetics.
Substantial reduction of the leach residue by graphite powder took place only above 1000 ºC.
The formation of CO or CO2 as the main gaseous product is largely dependent on temperature,
and is suggested to be dominantly influenced by the Boudouard reaction. The Fe and Si
205
components of the silicates could be reduced by graphite above their melting temperature (1200
ºC) forming an Fe-Si alloy. The alloys formed from the reduction of hematite, nickel ferrite or
silicates tend to be more homogeneous at higher temperatures resulting from the enhanced
merging of the alloy particles.
6.5 References
[1] D. Yu, T.A. Utigard, M. Barati, Fluidized Bed Selective Oxidation-Sulfation Roasting of
Nickel Sulfide Concentrate: Part I. Oxidation Roasting, Metallurgical and Materials
Transactions B, 2013, DOI: 10.1007/s11663-013-9958-x.
[2] D. Yu, T.A. Utigard, M. Barati, Fluidized Bed Selective Oxidation-Sulfation Roasting of
Nickel Sulfide Concentrate: Part II. Sulfation Roasting, Metallurgical and Materials
Transactions B, 2013, DOI: 10.1007/s11663-013-9959-9.
[3] D. Yu, M. Zhu, T.A. Utigard, M. Barati, TGA Kinetic Study on the Hydrogen Reduction of
an Iron Nickel Oxide, Minerals Engineering, 54 (2013) 32-38.
[4] O.J. Wimmers, P. Arnoldy, J.A. Moulijn, Determination of the Reduction Mechanism by
Temperature-Programmed Reduction: Application to Small Fe2O3 Particles, Journal of Physical
Chemistry, 90 (1986) 1331-1337.
[5] A.J.H.M. Kock, H.M. Fortuin, J.W. Geus, The Reduction Behavior of Supported Iron
Catalysts in Hydrogen or Carbon Monoxide Atmospheres, Journal of Catalysis, 96 (1985) 261-
275.
[6] H.-Y. Lin, Y.-W. Chen, C. Li, The mechanism of reduction of iron oxide by hydrogen,
Thermochimica Acta, 400 (2003) 61-67.
[8] A. Pineau, N. Kanari, I. Gaballah, Kinetics of reduction of iron oxides by H2 Part I: Low
temperature reduction of hematite, Thermochimica Acta, 447 (2006) 89-100.
[10] J. Zielinski, I. Zglinicka, L. Znak, Z. Kaszkur, Reduction of Fe2O3 with hydrogen, Applied
Catalysis A: General, 381 (2010) 191-196.
206
[11] B. Weiss, J. Sturn, S. Voglsam, F. Winter, J. Schenk, Industrial fluidised bed direct
reduction kinetics of hematite ore fines in H2 rich gases at elevated pressure, Chemical
Engineering Science, 66 (2011) 703-708.
[13] J.-m. Pang, P.-m. Guo, P. Zhao, C.-z. Cao, D.-w. Zhang, Influence of Size of Hematite
Powder on Its Reduction Kinetics by H2 at Low Temperature, Journal of Iron and Steel
Research, International, 16 (2009) 07-11.
[15] M. Bahgat, M.-K. Paek, J.-J. Pak, Reduction Kinetics and Mechanisms of NiFe2O4 with
Synthesis of Nanocrystalline Fe-Ni Alloy, Materials Transactions, JIM, 48 (2007) 3132-3139.
[16] M.H. Khedr, Isothermal reduction kinetics at 900-1100ºC of NiFe2O4 sintered at 1000-
1200ºC, Journal of Analytical and Applied Pyrolysis, 73 (2005) 123-129.
[18] M.O. Boudouard, Recherches sur les equilibres chimiques, Annales De Chimie Et De
Physique, 24 (1901) 1-85.
[19] R. Haque, H.S. Ray, Role of Ore/Carbon Contact and Direct Reduction in the Reduction of
Iron Oxide by Carbon, Metallurgical and Materials Transactions B, 26B (1995) 400-401.
[20] M.S. Bafghi, M. Fukuda, Y. Ito, S. Yamada, M. Sano, Effect of CO Gas Formation on
Reduction Rate of Iron Oxide in Molten Slag by Graphite, ISIJ International, 33 (1993) 1125-
1130.
[21] J. Moon, V. Sahajwalla, Investigation into the Role of the Boudouard Reaction in Self-
Reducing Iron Oxide and Carbon Briquettes, Metallurgical and Materials Transactions B, 37B
(2006) 215-221.
[22] K. Mondal, H. Lorethova, E. Hippo, T. Wiltowski, S.B. Lalvani, Reduction of iron oxide in
carbon monoxide atmosphere - reaction controlled kinetics, Fuel Processing Technology, 86
(2004) 33-47.
[23] S.K. Sharma, F.J. Vastola, J. P.L. Walker, Reduction of nickel oxide by carbon: I.
Interaction between nickel oxide and pyrolytic graphite, Carbon, 34 (1996) 1407-1412.
[24] S.K. Sharma, F.J. Vastola, J. P.L. Walker, Reduction of nickel oxide by carbon: II.
Interaction between nickel oxide and natural graphite, Carbon, 35 (1997) 529-533.
207
[25] S.K. Sharma, F.J. Vastola, J. P.L. Walker, Reduction of nickel oxide by carbon: III. Kinetic
studies of the interaction between nickel oxide and natural graphite, Carbon, 35 (1997) 535-541.
[26] A. Roine, HSC Chemistry, Outokumpu Research Oy, Pori, Finland, 2007.
[27] D. Yu, T.A. Utigard, TG/DTA study on the oxidation of nickel concentrate,
Thermochimica Acta, 533 (2012) 56-65.
[28] P.G. Thornhill, L.M. Pidgeon, Micrographic Study of Sulfide Roasting, Journal of Metals,
9 (1957) 989-995.
[31] G.C. Ulmer, W.C. Elliott, T. Buntin, J. Edwin S. Erickson, J.J. Friel, Role of Selected
Cations and Gas Speciation on the Reduction of Fayalite at 1300ºC, Journal of the American
Ceramic Society, 75 (1992) 1476-1483.
[32] S. Vyazovkin, Kinetic concepts of thermally stimulated reactions in solids: A view from a
historical perspective, International Reviews in Physical Chemistry, 19 (2000) 45-60.
[33] A.K. Galwey, What is meant by the term ‘variable activation energy’ when applied in the
kinetic analyses of solid state decompositions (crystolysis reactions)?, Thermochimica Acta, 397
(2003) 249-268.
[34] A. Khawam, D.R. Flanagan, Solid-State Kinetic Models: Basics and Mathematical
Fundamentals, The Journal of Physical Chemistry B, 110 (2006) 17315-17328.
[35] C.J.B. Crowe, T.A. Utigard, Kinetics of Carbon Monoxide Reduction of Oxidized Calcines,
Canadian Metallurgical Quarterly, 42 (2003) 447-454.
208
7 Summary and Conclusions
7.1 Mass and Heat Balance
A mass and heat balance calculation was performed using HSC Chemistry [1] for three stages of
the process, i.e. oxidation roasting of the Raglan concentrate, sulfation roasting of the oxidation
roasted calcine, and reduction of the leach residue. The results are shown in Table 7.1, Table 7.2
and Table 7.3, respectively. For the oxidation roasting stage, Raglan concentrate of 100 kg was
used as the starting material. Stoichiometric amount of air was used to preferentially oxidize the
Raglan concentrate at 600 ºC to produce a calcine of a similar chemical composition as that
obtained in the oxidation roasting experiments. Formation of small amounts of sulfates was also
taken into account in the calculation process as shown in Table 7.1. The BALANCE of mass (in
kg) in Table 7.1 is zero, indicating that the mass balance was performed correctly. For the
calculation of heat balance, thermodynamic data for all the species involved is available in the
database of HSC Chemistry [1], except that for pentlandite. The standard enthalpy of formation
for pentlandite of -(837.37±14.59) kJ/mol [2] was used. The heat capacity of pentlandite at
298.15 K is 442.7 J/mol·K [3], which was used for the calculation at high temperatures since no
heat capacity at higher temperature is available. As can be seen, heat of a total of 267.79 MJ is
released from the processing of 100 kg concentrate. This balance of heat should be absorbed to
prevent the overheating of the fluidized bed in an industrial setting. This could be done by
mixing the concentrate with water to make a slurry as the feed to the fluidized bed roaster,
which has been a common practice in the industry. Such a practice could also alleviate the
entrainment of dust into the offgas. By calculation, the amount of heat released from roasting of
100 kg concentrate could form 103.66 kg steam from water at 25 ºC. This suggests that the
slurry of feed could be made with the concentrate to water ratio of 1:1 to achieve a constant
temperature of the fluidized bed. Alternatively, the excessive amount of heat could also be
absorbed by inserting water coolants into the fluidized bed to recover the heat by making steam.
As to the mass balance of sulfur which accounts for 29.39 wt% of the Raglan concentrate, the
degree of sulfur elimination in the oxidation roasting stage is 56.1 wt%, the remaining 43.9
wt% being in the calcine in the form of sulfides and sulfates.
209
Table 7.1. Mass and heat balance for the oxidation roasting of 100 kg Raglan concentrate.
INPUT
Raglan concentrate Temperature Amount Amount Amount Latent H Total H
100kg (ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe4.5Ni4.5S8 25.0 0.070 53.94 0.000 0.00 -58.51
Fe4.5Co4.5S8 25.0 0.001 1.08 0.000 0.00 -1.17
Fe7S8 25.0 0.025 16.11 0.000 0.00 -18.32
CuFeS2 25.0 0.074 13.53 0.003 0.00 -14.04
Mg2SiO4 25.0 0.069 9.77 0.003 0.00 -151.17
Al2SiO5(S) 25.0 0.008 1.24 0.000 0.00 -19.81
SiO2 25.0 0.072 4.33 0.002 0.00 -65.65
Air
O2(g) 25.0 1.147 36.69 25.700 0.00 0.00
N2(g) 25.0 4.313 120.84 96.681 0.00 0.00
Total input N/A 5.779 257.53 122.389 0.00 -328.67
OUTPUT
Temperature Amount Amount Amount Latent H Total H
Calcine
(ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe2O3 600.0 0.284 45.40 0.009 22.40 -211.59
NiS 600.0 0.252 22.83 0.004 9.40 -12.70
CoS 600.0 0.004 0.35 0.000 0.12 -0.26
Cu2S 600.0 0.034 5.40 0.001 1.93 -0.76
Mg2SiO4 600.0 0.049 6.84 0.002 4.28 -101.54
NiSO4 600.0 0.063 9.73 0.002 4.68 -50.21
CoSO4 600.0 0.002 0.38 0.000 0.19 -1.99
CuSO4 600.0 0.006 0.94 0.000 0.45 -4.10
MgSO4 600.0 0.042 5.02 0.002 2.99 -49.59
Al2SiO5(S) 600.0 0.008 1.24 0.000 0.73 -19.07
SiO2 600.0 0.093 5.58 0.002 3.41 -81.21
Offgas
O2(g) 600.0 0.001 0.02 0.015 0.01 0.01
N2(g) 600.0 4.313 120.84 96.681 74.90 74.90
SO2(g) 600.0 0.515 32.96 11.532 14.37 -138.34
Total output N/A 5.664 257.53 108.251 139.87 -596.46
210
The calcine produced from the above oxidation roasting stage was used at 25 ºC as part of the
input for the mass and heat balance calculation of the sulfation roasting stage, as shown in Table
7.2. In reality, calcine produced from the oxidation roasting stage could be directly fed to the
sulfation roasting stage before its temperature drops substantially, for the sake of energy
conservation. Na2SO4 was also added for sulfation roasting with the concentrate to Na2SO4
weight ratio of 10:1. The offgas from the oxidation roasting stage which has 10.7 vol% SO2
could be used as the feed gas for the sulfation roasting stage. In order to achieve a preferable
sulfation condition and to produce an offgas from the sulfation roasting stage with a SO2
concentration high enough for acid making, this offgas is mixed with 96% oxygen (balance
nitrogen) instead of air as the feed gas. Sulfation roasting was assumed to take place at 700 ºC.
Solid product from the sulfation roasting is the calcine containing sulfates, oxides and silicates,
the amounts of which were calculated based on the metal recoveries in the experiments. As to
the balance of sulfur, 17.1 vol% of the SO2 in the feed gas forms sulfates in the sulfation
roasting stage, and 46.5 wt% of total sulfur remains as SO2 and SO3, the rest being in the
sulfation roasted calcine. The concentrations of O2, SO2, and SO3 in the offgas in the OUTPUT
were calculated based on their thermodynamic equilibrium at 700 ºC. The calculated
concentrations of SO2 and SO3 in the offgas are 4.1 vol% and 4.8 vol%, respectively. This
relatively high concentration of SO3 in the offgas will cause possible operational and
maintenance problems of the offgas handling system, such as plugging and corrosion, which
necessitates the special design of the offgas handling system to cope with the possible issues. In
terms of heat balance, total heat of 182.37 MJ is produced from the sulfation roasting stage
when 100kg concentrate is processed. This heat could be absorbed either by making slurry as
the solid feed or by applying water coolants in the fluidized bed.
Table 7.2. Mass and heat balance for the sulfation roasting stage.
INPUT
Calcine from oxidation Temperature Amount Amount Amount Latent H Total H
roasting (ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe2O3 25.0 0.284 45.40 0.009 0.00 -234.00
NiS 25.0 0.252 22.83 0.004 0.00 -22.10
CoS 25.0 0.004 0.35 0.000 0.00 -0.38
Cu2S 25.0 0.034 5.40 0.001 0.00 -2.70
Mg2SiO4 25.0 0.049 6.84 0.002 0.00 -105.82
NiSO4 25.0 0.063 9.73 0.002 0.00 -54.90
211
CoSO4 25.0 0.002 0.38 0.000 0.00 -2.18
CuSO4 25.0 0.006 0.94 0.000 0.00 -4.55
MgSO4 25.0 0.042 5.02 0.002 0.00 -52.57
Al2SiO5(S) 25.0 0.008 1.24 0.000 0.00 -19.81
SiO2 25.0 0.093 5.58 0.002 0.00 -84.62
Addition of Na2SO4
Na2SO4 25.0 0.073 10.37 0.004 0.00 -101.34
Offgas from oxidation roasting
O2(g) 600.0 0.001 0.02 0.015 0.01 0.01
N2(g) 600.0 4.313 120.84 96.681 74.90 74.90
SO2(g) 600.0 0.515 32.96 11.532 14.37 -138.34
96% oxygen to be mixed with the offgas for sulfation roasting
O2(g) 25.0 0.772 24.70 17.298 0.00 0.00
N2(g) 25.0 0.032 0.90 0.721 0.00 0.00
Total input N/A 6.541 293.49 126.273 89.28 -748.38
OUTPUT
Temperature Amount Amount Amount Latent H Total H
Sulfation roasted calcine
(ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe2O3 700.0 0.204 32.54 0.006 19.53 -148.18
NiFe2O4 700.0 0.067 15.77 0.000 0.00 -72.75
CuO 700.0 0.007 0.53 0.000 0.23 -0.82
CoO 700.0 0.001 0.04 0.000 0.02 -0.11
Mg2SiO4 700.0 0.009 1.24 0.000 0.93 -18.27
Al2SiO5(S) 700.0 0.004 0.73 0.000 0.51 -11.07
SiO2 700.0 0.136 8.16 0.003 5.92 -117.84
Fe2(SO4)3 700.0 0.013 5.30 0.002 3.21 -31.04
NiSO4 700.0 0.247 38.25 0.010 22.09 -193.65
CuSO4 700.0 0.067 10.70 0.003 6.09 -45.61
CoSO4 700.0 0.006 0.89 0.000 0.55 -4.53
MgSO4 700.0 0.121 14.59 0.005 10.48 -142.51
Al2(SO4)3 700.0 0.003 1.09 0.000 0.79 -10.14
Na2SO4 700.0 0.073 10.37 0.004 9.33 -92.00
Offgas from sulfation roasting
O2(g) 700.0 0.019 0.60 0.417 0.40 0.40
SO2(g) 700.0 0.198 12.68 4.437 6.61 -52.15
SO3(g) 700.0 0.228 18.28 5.118 10.42 -79.95
N2(g) 700.0 4.346 121.74 97.402 89.46 89.46
Total output N/A 5.749 293.49 107.407 186.58 -930.75
212
The water-insoluble part of the sulfation roasted calcine produced from above was assumed as
the water-leach residue for the mass and heat balance calculation of its reduction by CO. A
stoichiometric amount of CO was used for reduction. The reduction temperature was assumed as
1250 ºC which is 33 ºC higher than the melting point of fayalite (*FeO*SiO2). The grade of Fe-
Ni alloy formed would be higher because part of the iron oxide would form molten slag phase as
observed in the experiments. The reduction requires an energy input of 93.62 MJ per 100 kg
concentrate processed. This energy input could be in the form of incomplete combustion of fuel
(e.g. oil, coke) which provides both heat and reductant for the reduction process.
Table 7.3. Mass and heat balance for the reduction of the leach residue with CO.
INPUT
Leach Temperature Amount Amount Amount Latent H Total H
residue (ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe2O3 25.0 0.217 34.66 0.007 0.00 -178.62
NiFe2O4 25.0 0.067 15.77 0.000 0.00 -72.75
CuO 25.0 0.007 0.53 0.000 0.00 -1.05
CoO 25.0 0.001 0.04 0.000 0.00 -0.13
Mg2SiO4 25.0 0.009 1.24 0.000 0.00 -19.20
Al2SiO5(S) 25.0 0.004 0.73 0.000 0.00 -11.59
SiO2 25.0 0.136 8.16 0.003 0.00 -123.76
Reductant
CO(g) 25.0 0.656 18.37 14.699 0.00 -72.49
Total input N/A 1.097 79.51 14.709 0.00 -479.58
OUTPUT
Temperature Amount Amount Amount Latent H Total H
Solid
(ºC) (kmol) (kg) (Nm3) (MJ) (MJ)
Fe 1250.0 0.297 16.58 0.002 13.83 13.83
Ni 1250.0 0.067 3.95 0.000 2.69 2.69
Cu 1250.0 0.007 0.43 0.000 0.32 0.32
Co 1250.0 0.001 0.03 0.000 0.03 0.03
Mg2SiO4 1250.0 0.009 1.24 0.000 1.80 -17.39
Al2SiO5(S) 1250.0 0.004 0.73 0.000 1.01 -10.57
*2FeO*SiO2 1250.0 0.136 27.69 0.006 42.80 -158.17
Offgas
CO2(g) 1250.0 0.656 28.86 14.699 41.37 -216.69
Total output N/A 1.176 79.51 14.708 103.85 -385.96
213
7.2 Conclusions
Selective oxidation and sulfation roasting followed by leaching was proposed and investigated
as an alternative route to the conventional roasting–electric furnace smelting–converting route
for nickel processing. The oxidation roasting stage of the innovative process was intended to
preferentially oxidize the iron species from the nickel concentrate forming iron oxides. The
sulfides of non-ferrous metals (Ni, Cu and Co) were then selectively converted to water-soluble
sulfates for further leaching either by water or acid.
In order to achieve preferential oxidation of the iron species, the roasting mechanism of the
nickel concentrate was firstly studied by means of TGA and DTA. Roasting products from
intermediate temperatures were analyzed by various techniques, such as SEM/EDS, XRD,
EPMA, and ICP, to help elucidate the reaction scheme. It was found that due to the complex
mineralogy of the nickel concentrate and the heterogeneous nature of the roasting reactions,
various reactions involving a large variety of intermediate compounds as well as phase changes
could take place, such as the low temperature thermal decomposition of pentlandite (350–550
ºC), preferential oxidation of iron sulfide species (<700 ºC), transformation of nickel sulfide
core from Ni1-xS to Ni3±xS2 (730 ºC), melting of nickel sulfide core (813 ºC), and sulfate
formation and decomposition at various temperatures. The reaction rate was found to be
controlled by the gas diffusion downward through the sample bed.
In order to gain more understanding of a possible industrial roasting practice, a laboratory scale
fluidized bed roster which is operated in a batch mode was constructed for oxidation and
sulfation roasting experiments. It was found that the oxidation roasting in a fluidized bed starts
with the preferential oxidation of iron sulfide species forming iron oxides until the content of Fe
in the pentlandite sulfide cores drops to ~2 at%. The remaining nickel sulfide core in the
pentlandite particles is either NiS or Ni3S2 depending on the roasting temperature. Formation of
sulfates of Ni and Co occurs after the oxidation of iron sulfide. Low temperature (e.g. 650 °C) is
favorable for the preferential oxidation of iron sulfide species while minimizing the formation of
nickel ferrite. Unlike the oxidation of iron sulfide, the oxidation of Co species through diffusion
214
is a slow process. The oxidation of the nickel sulfide core Ni3S2 is the last step if the roasting
temperature is high enough.
Selective sulfation roasting of the oxidation roasted calcine was further studied in the fluidized
bed roaster with the evaluation of several parameters. It was concluded that among the
parameters evaluated, the addition of Na2SO4 as catalyst was most effective in enhancing sulfate
formation. Under optimized conditions of sulfation gas composition (95% air, 5% SO2),
temperature (700 °C), Na2SO4 addition (10 wt%) and time (150 min), 79% Ni, 91% Cu, and
91% Co were converted to water-soluble sulfates. Iron sulfate formation was only 5%. A
sulfation mechanism was suggested. A high partial pressure of SO3 is maintained within each
particle due to the formation of sulfate melt on the surface, favoring sulfate formation. The
conversion of the nickel sulfide core to sulfate is direct without preliminary formation of NiO as
an intermediate product. Another important finding was that the sustained conversion to NiSO4
is due to the unique morphology of the NiSO4 formed, which is the micro-grain structure with
cracks and crevices which allowed the sulfation to proceed to near completion by providing
channels for the inner diffusion of sulfation gas to the sulfide surface.
Leach residue from the optimized oxidation–sulfation roasting test is mainly composed of
hematite (Fe2O3) and nickel ferrite (NiFe2O4), which could be a superior feedstock for the
production of ferronickel. The feasibility of producing ferronickel by pyro-reduction of the
leach residue with H2, CO and graphite was evaluated by means of TGA and DTA. It was found
that at low temperatures (350–600 ºC), the reduction rate is governed by chemical reactions
when using H2 as reductant. The rate controlling step becomes mixed control of gas diffusion
through the solid product layer of individual particles and gas diffusion through the sample bed
at higher temperatures. The formed alloys experienced significant sintering at temperatures
higher than 1100 ºC. When using CO as the reductant, sintering of the products took place at
temperatures as low as 800 ºC. A shift of the rate controlling mechanism from gas diffusion
through the pores of the sample bed to gas diffusion through the product layer in individual
particles were observed for the isothermal reduction tests below 1200 ºC, resulting in a variation
of apparent activation energies with the extent of reduction. Melting of the siliceous materials in
the residue at 1200 ºC greatly hindered the reduction process by substantially reducing the
porosity of the sample bed. Temperatures higher than 1000 ºC are required to initiate effective
215
reduction by graphite. The molten silicate (>1200 ºC) can be reduced by graphite to form Si-
containing alloys.
Calcines were also leached by hot water and HCl to study their leaching behavior. It was
concluded that hot water leaching at 90 ºC for 30 minutes is effective to dissolve all sulfates
from the calcines. In order to increase of the recovery of the non-ferrous metals, acid leaching is
beneficial. However, less selectivity is observed with acid leaching because substantial iron
oxides can also be dissolved.
7.3 References
[1] A. Roine, HSC Chemistry, Outokumpu Research Oy, Pori, Finland, 2007.
[2] L. Cemič, O.J. Kleppa, High temperature calorimetry of sulfide systems, Phys Chem
Minerals, 14 (1987) 52-57.
[3] G.A. Berezovskii, V.A. Drebushchak, T.A. Kravchenko, Low-temperature heat capacity of
pentlandite, American Mineralogist, 86 (2001) 1312-1313.
216
8 Proposed Flow Sheet
Based on this study, a flow sheet was proposed for the two-stage fluidized bed selective
oxidation and sulfation roasting of nickel sulfide concentrate, which is illustrated in Figure 8.1.
It involves two fluidized bed roasters in series, one for oxidation and the other for sulfation. As
suggested from the Mass and Heat Balance Section (Section 7.1), offgas from the oxidation
roasting stage could be used after mixing with 96% oxygen as the feed gas for the sulfation
roasting stage. And the offgas from the sulfation roasting stage could be used for acid making.
Na2SO4 solution which is recycled from the hydrometallurgical stream should be mixed with the
oxidation roasted calcine to make a slurry before being fed to the sulfation roaster. The latent
heat of the hot calcine produced from the oxidation roasting stage (600 ºC) could be utilized to
concentrate the diluted Na2SO4 solution by vaporizing some of the water content, so that the
water content of the slurry is just enough to absorb the heat released from the sulfation roasting
stage (Please refer to the Mass and Heat Balance Section 7.1). Heat scavenged from the
oxidation roasting stage could also be utilized for the concentration of the Na2SO4 solution. The
hot calcine produced from the sulfation roasting stage can be quenched by water to render quick
dissolution of sulfates. After leaching, valuable metals could be separated and recovered by
standard hydrometallurgical methods, e.g. ion exchange, solvent extraction, and electrowinning,
after Fe removal. The portion of the precious metals that is leached into water could form
concentrated precious metal sludge during electrowinning. The method for the recovery of these
precious metals from the sludge has been well established in the industry. The diluted sulfuric
acid produced from the electrowinning process could be neutralized by adding limestone to
make gypsum (CaSO4·2H2O). Na2SO4 could be recovered from the solution and recycled back
to the sulfation roasting stage. The leach residue which contains mainly hematite and nickel
ferrite is used for the production of ferronickel alloys by pyro-reduction. One possible way of
recovering the precious metals from the leach residue is by partial reduction of the leach residue
so that the precious metals could be enriched in the metallic part.
217
Air Nickel Concentrate
Oxidation Roasting
Na2SO4 SO2 O2
Sulfation Roasting
SO2&SO3
Acid Making
Water Leaching
H2SO4
Concentration
Leachate Residue
Solvent Extraction/
Pyro-Reduction
Ion Exchange
CaCO3 H2SO4 Ni Cu Co
CaSO4
Figure 8.1. Proposed flow sheet for the selective oxidation and sulfation roasting of nickel
concentrate.
In the conventional pyrometallurgical process to treat nickel concentrate, the iron species end up
in molten slags in various stages. In comparison, the iron component of the nickel concentrate is
utilized to form final product, i.e. ferronickel alloy, in this proposed process. Thus maximum
218
utilization of the raw material is achieved, which is of great significance in terms of both the
economic aspect and approaching sustainability of the mining and metallurgical industries. SO2
emission from this process could be minimal since offgas with strong SO2 is produced from the
sulfation roaster, which makes it suitable for acid making. The consumption of electric energy
could also be greatly reduced. Pyro-reduction of the leach residue requires energy input, either
from electric energy or from combustion of fossil fuels. However, this energy input should be
much lower compared with the energy requirement of an electric furnace (EF) in the
conventional roasting–electric furnace smelting–converting process. This is because it can be
operated at much lower temperatures (~800 ºC) compared with the conventional EF smelting
practice (~1500 ºC). In addition, for the processing of one tonne of nickel concentrate in both
the conventional route and the proposed selective oxidation-sulfation roasting route, the mass of
the leach residue for reduction is much smaller than that of the feed to the EF furnace, meaning
much less energy is required.
By directly addressing the sustainability issues in three aspects, i.e. materials, environment, and
energy, this innovative process should be more sustainable compared with the conventional
pyrometallurgical route for nickel production.
219
Appendices
Appendix 01: C Code for the Calculation of Enthalpy
Change during the Roasting of the Raglan Concentrate as a
Function of Temperature
/*
This program performs the following functions:
1. Read the csv file which contains one column of numerical values and convert them to
double values;
2. Process the data according to the function, which calculate the enthalpy change for the
following reactions:
1) 1Fe4.5Ni4.5S8 + 8.375O2 = 2.25Fe2O3 + 1.5Ni3S2 + 5SO2
2) 1Fe7S8 + 13.25O2 = 3.5Fe2O3 + 8SO2
3) 1CuFeS2 + 3.25O2 = 0.5Fe2O3 + 1CuO + 2SO2
According to the mineralogy of the Raglan concentrate, the molar ratios of these three
minerals are:
Pn:Py:Cpy=1 : 0.349180273 : 1.034034368
The molar ratio of the total SO2 formed to the minerals are:
SO2:Pn:Py:Cpy=(1*5+0.349180273*8+1.034034368*2) : 1 : 0.349180273 : 1.034034368
=9.86151092: 1 : 0.349180273 : 1.034034368
#include <string.h>
#include <stdlib.h>
#include <math.h>
#include <stdio.h>
#include <graphics.h>
#include <conio.h>
double function(double a);
double pn(double T);
double o2(double T);
double fe2o3(double T);
double ni3s2(double T);
double so2(double T);
double py(double T);
double cpy(double T);
double cuo(double T);
void main()
{
FILE *fp, *fp2;
int i=5, indicator_row=0, indicator_column=0;
double buff;
220
char t, temp[50], pathread[50], pathwrite[50], *p, *end;
printf("This program performs the following functions:\n1. Read the csv file which contains one
column of numerical values and convert them to double values;\n2. Process the data according
to the function;\n3. Write the data produced into a new csv file.\n--Dawei Yu, March 3rd,
2012\n");
looppath:
printf("Please input the path of the source file here.\n You have %d time(s):",i);
scanf("%s",pathread);
clrscr();
fp=fopen(pathread,"r");
if(fp!=NULL)
printf("\n\n\nFile is opened successfully!\n");
else
{i--;
printf("File is not open! Please check the path!\n");
if(i>=1) goto looppath;
else goto end;
}
printf("Please input the directory of the csv file which contains the results:");
scanf("%s",pathwrite);
fp2=fopen(pathwrite,"w");
FLAG:
for(i=0,t=fgetc(fp);!feof(fp)&&i<50;t=fgetc(fp),i++)
{
if(t!=','&&t!='\n') temp[i]=t;
if(t==',') {indicator_column=1; break;}
if(t=='\n') {indicator_row=1; break;}
buff=function(strtod(temp,&end));
sprintf(temp,"%lf",buff);
221
p=temp;
while((*p!='\0')&&fputc(*(p++),fp2)!=EOF);
if(indicator_column==1) fputc(',',fp2);
else if(indicator_row==1) fputc('\n',fp2);
indicator_column=0;
indicator_row=0;
end: fclose(fp);
fclose(fp2);
printf("\nDONE!\n");
getch();
}
double function(double T)
{
double Enthalpy;
/*
1) 1Fe4.5Ni4.5S8 + 8.375O2 = 2.25Fe2O3 + 1.5Ni3S2 + 5SO2
2) 1Fe7S8 + 13.25O2 = 3.5Fe2O3 + 8SO2
3) 1CuFeS2 + 3.25O2 = 0.5Fe2O3 + 1CuO + 2SO2
Pn:Py:Cpy=1 : 0.349180273 : 1.034034368
*/
Enthalpy=1*(2.25*fe2o3(T)+1.5*ni3s2(T)+5.0*so2(T)-1.0*pn(T)-
8.375*o2(T))+0.349180273*(3.5*fe2o3(T)+8.0*so2(T)-1.0*py(T)-
13.25*o2(T))+1.034034368*(0.5*fe2o3(T)+1.0*cuo(T)+2*so2(T)-1.0*cpy(T)-3.25*o2(T));
return Enthalpy;
}
double pn(double T)
{
double En;
En=-837.37+0.4427*(T-298.15);
return En;
}
double o2(double T)
{
222
double T2, En, a[2][6]={{22.060,20.887,1.621,-8.207,298.150,700.0},{29.793,7.910,-6.194,-
2.204,700,1200}};
int i, flag=0;
En=0.00;
for(i=0;i<2;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
return En;
}
double so2(double T)
{
double T2, En, a[2][6]={{29.134,37.222,0.058,-2.885,298.15,500},{54.779,3.350,-24.745,-
0.241,500,5000}};
int i, flag=0;
En=-296.813;
for(i=0;i<2;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
return En;
}
double fe2o3(double T)
{
double T2, En, a[4][6]={{143.556,-36.323,-31.433,71.792,298.15,700},{637.809,-963.532,-
447.383,560.951,700.0,950.0},{-220672.038,290104.570,378928.406,-
107181.103,950.0,1050.0},{80.217,55.974,167.385,-12.403,1050.0,1812}};
int i, flag=0;
223
En=-823.00;
for(i=0;i<4;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
return En;
}
double ni3s2(double T)
{
double T2, En, a[3][6]={{113.202,48.626,-8.858,-
0.008,298.15,829.0},{188.615,0.0,0.0,0.0,829.0,1062},{191.790,0.0,0.0,0.0,1062.0,3800.0}};
int i, flag=0;
En=-216.313;
for(i=0;i<3;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
if(T>1062) En=En+19.748+56.233;
else if(T>829) En=En+56.233;
return En;
}
double py(double T)
{
double T2, En, a[1][6]={{398.568,0.0,0.0,0.0,298.150,1500}};
int i, flag=0;
En=-736.384;
for(i=0;i<1;i++)
{
224
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
return En;
}
double cpy(double T)
{
double T2, En, a[3][6]={{86.985,53.555,-5.607,0.0,298.15,830},{-
1441.974,1844.977,0.0,0.0,830,930},{172.464,0.0,0.0,0.0,930.0,1200.0}};
int i, flag=0;
En=-190.372;
for(i=0;i<3;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
if(T>830.0) En=En+10.083;
return En;
}
double cuo(double T)
{
double T2, En, a[1][6]={{48.591,7.198,-7.500,0.001,298.15,1500}};
int i, flag=0;
En=-155.800;
for(i=0;i<1;i++)
{
if(T<=a[i][5])T2=T, flag=1;
else T2=a[i][5];
225
En=En+1e-3*((a[i][0]*T2+a[i][1]*1e-3/2*pow(T2,2)-a[i][2]*1e5/T2+a[i][3]*1e-
6/3*pow(T2,3))-(a[i][0]*a[i][4]+a[i][1]*1e-3/2*pow(a[i][4],2)-a[i][2]*1e5/a[i][4]+a[i][3]*1e-
6/3*pow(a[i][4],3)));
if(flag==1) break;
else ;
}
return En;
}
226
Appendix 02: C Code for the Calculation of the Heat
Transfer Rate through the Quartz Tube from the Air in the
Electric Furnace to the Fluidized Bed
/*C code for the calculation of the heat transfer rate through the quartz tube from the air in the
electric furnace to the fluidized bed.*/
/*
Conditions: The temperature of the air in the furnace is known to be 973.15K. And the
temperature of the fluidized bed is also known at specific time to be represented by Tb. There
are three layers to be considered: 1. the natural convection of the air near the outer surface of the
quartz tube; 2. the heat conduction in the quartz; 3. the heat transfer from the fluidized bed to the
inner wall of the quartz tube.
2. For the calculation of the heat transfer coefficient for natural convection, the following
equations are used:
Tf=1/2*(Toq+973.15) where Toq is the temperature of the outer surface of the quartz tube.
Pr=nv/alfa, where Pr is Prandtl number, nv is kinematic viscosity of air, alfa is thermal
diffusivity of air.
Gr=9.81*pow(L,3)*pow(rho,2)*beta*(Tf-973.15)/pow(nv,2), where L is the height of the
fluidized bed; rho is the density of the air at the temperature Tf, beta is 1/Temperature for ideal
gases.
Nu=0.902*sqrt(Pr)*pow(Gr/4,1/4)/pow(0.861+Pr,1/4), where Nu is the Nusselt number.
Nu=ha*L/k, where ha is the heat transfer coefficient, L is the height of the fluidized bed, k is the
thermal conductivity of the air at temperature Tf.
3. For the calculation of the heat transfer coefficient for the fluidized bed, the following
equations are used:
Radiation of the fluidized bed can be neglected since the quartz tube is transparent.
Tf2=(Tb+Tiq)/2, where Tf2 is the average, Tb is the temperature of the fluidized bed, Tiq is the
temperature of the inner wall of the fluidized bed.
Re=D*V*rho/nv
227
where D is the mean diameter of the fluid particle, V is the superfacial velocity of the feeding
gas,rho is the density of the air,nv is the kinematic viscosity of the air at Tf2.
Pr=nv/alfa, where Pr is the Prandtl number, nv is kinematic viscosity of the air at Tf2, alfa is the
thermal diffusivity of air.
Nu=0.85*pow(Ar,0.1)*pow(2872/rho,0.14)*pow(Cp/Ca,0.24)*pow(1-
omega,2/3)+0.046*Re*Pr*pow(1-omega,2/3)/omega,
where rho is the density of air at Tf2, Cp is the heat capacity of sand(quartz) at Tb, Ca is the
heat capacity of air at Tb, omega is the porosity of the fluidized bed, which is assume to be 0.7.
Nu=hb*D/k, where k is the thermal conductivity of air at Tf2, hb is the heat transfer coefficient
of the fluidized bed, D is the mean particle size.
*/
#include<string.h>
#include<math.h>
#include<stdio.h>
#include<conio.h>
double kq(double T);
double Cp(double T);
double nv(double T);
double alfa(double T);
double rho(double T);
double k(double T);
double Ca(double T);
void main()
{
Start:
int i;
char a;
double Flag1=1, Flag2=1, Gr, Nu, Re, Pr, Tf, Tf2, Ar, ha, hb, Tb, Tiq=1.0, Tiq2=1.0, Toq=1.0,
Toq2=1.0, D, V, r0=0.018, r1=0.020, omega=0.7, Q1=1, Q2, Q10, Q20, L=0.0294, beta;
D=(150.0+212.0)/2.0*1e-6;
for(Q2=2.0*Q1, i=0;Flag2*Flag1>=0;i++)
{
if(i>0) Flag2=Flag1;
else ;
for(;;)
{
Tf2=(Tiq+Tb)/2.0;
Ar=9.81*pow(D,3.0)*rho(Tf2)*(2872.0-rho(Tf2))/pow(nv(Tf2),2.0);
V=3.0*1e-3/60.0/(3.141592653*r0*r0)*Tf2/298.15;
Re=D*V*rho(Tf2)/nv(Tf2);
Pr=nv(Tf2)/alfa(Tf2);
Nu=0.85*pow(Ar,0.1)*pow(2872.0/rho(Tf2),0.14)*pow(Cp(Tf2)/Ca(Tf2),0.24)*pow(1.
0-omega,2.0/3.0)+0.046*Re*Pr*pow(1.0-omega,2.0/3.0)/omega;
hb=Nu*k(Tf2)/D;
Q1=2.0*3.141592653*L*(Toq-Tb)/(1.0/(r0*hb)+log(r1/r0)/kq(Tf2));
Tiq2=Toq-Q1*log(r1/r0)/kq(Tf2)/2.0/3.141592653/L;
if(fabs((Tiq2-Tiq)/Tiq)>=1e-9) Tiq=Tiq2;
else break;
}
for(;;)
{
Tf=1.0/2.0*(Toq+973.15);
beta=1.0/Tf;
Pr=nv(Tf)/alfa(Tf);
Gr=9.81*pow(L,3.0)*pow(rho(Tf),2.0)*beta*fabs(Tf-973.15)/pow(nv(Tf),2.0);
Nu=0.902*sqrt(Pr)*pow(Gr/4.0,1.0/4.0)/pow(0.861+Pr,1.0/4.0);
ha=Nu*k(Tf)/L;
229
Q2=2.0*3.141592653*L*(973.15-Toq)/(1.0/(r1*ha));
Toq2=Q2*log(r1/r0)/kq(Tf)/2.0/3.141592653/L+Tiq;
if(fabs((Toq2-Toq)/Toq)>=1e-9) Toq=Toq2;
else break;
}
Flag1=Q1-Q2;
if(i=0) Flag2=Flag1;
else ;
231
Appendix 03: Photos of the Fluidized Bed Experimental
Setup
232
Figure A 2. Sample feeding and collection systems of the fluidized bed setup.
233
Figure A 3. Brass cap and its fittings on the quartz tube.
234
Figure A 4. Interior of the brass cap. The gas outlet is covered with a piece of stainless steel
mesh to filter the dust from the offgas. A thin steel blade is fixed on the rotatable ceramic tube
by cement. During roasting, the dust accumulated on the mesh is periodically scraped by the
rotating blade and falls into the fluidized bed.
235
Figure A 5. Upper part of the clear quartz tube showing the SO3 fume formed during roasting
and the condensed SO3 on the inner wall.
236
Figure A 6. Interior of the muffle furnace.
237
Appendix 04: Related Publications
(i) R. Pandher, S. Thomas, D. Yu, M. Barati, T. Utigard, Sulfate Formation and Decomposition
of Nickel Concentrates, Metallurgical and Materials Transactions B, 42 (2011) 291-299.
(ii) D. Yu, T.A. Utigard, TG/DTA study on the oxidation of nickel concentrate, Thermochimica
Acta, 533 (2012) 56-65.
(v) D. Yu, M. Zhu, T.A. Utigard, M. Barati, TGA kinetic study on the hydrogen reduction of an
iron nickel oxide, Minerals Engineering, 54 (2013) 32-38.
(vi) D. Yu, M. Zhu, T.A. Utigard, M. Barati, TG/DTA study on the carbon monoxide and
graphite thermal reduction of a high-grade iron nickel oxide residue with the presence of
siliceous gangue, Thermochimica Acta, 575 (2014) 1-11.
(vii) D. Yu, T.A. Utigard, M. Barati, Leaching behavior of the roasted nickel calcine, in: 51st
Annual Conference of Metallurgists, Niagara Falls, Ontario, Canada, 2012, pp. 275-289.
(viii) D. Yu, M. Zhu, T.A. Utigard, M. Barati, TGA Kinetic Study on the H2 Reduction of an
Iron Nickel Oxide, in: The 3rd International Symposium on Processing of Nickel Ores &
Concentrates, Cape Town, South Africa, 2012.
238