XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1933
magnetitic concentrate. Connelly and Yan (2009) pre-
sented a few flowsheets for treating different magnetite
ores in Australia and mentioned some important trends
in magnetite processing, such as, the use of mineral lib-
eration analysis to understand the mineral behaviour, con-
sider coarse cobbing as a critical upgrading and application
of high-pressure grinding rolls (HPGR) to save grinding
energy.
The current evaluation aims to increase even further the
iron content in the concentrate, aligned with the sustain-
able goals to produce green steel. To accomplish the goal,
the work was developed in three stages: i) perform mineral
characterization of the ore ii) test in bench lab scale a pos-
sible flowsheet iii) investigate opportunities for optimizing
the beneficiation route.
DEVELOPMENT
Two sample were collected in the industrial plant before
(i.e., feed sample) and after the dry magnetic separation
(i.e., cobbing concentrate). The material presented a top
size of 40 mm and it was crushed in a lab jaw crusher mill
at a top size of 8 mm.
Powder X-ray diffraction (XRD) acquisitions were per-
formed using a Bruker D2 Phaser diffractometer, which
features a goniometer in Bragg-Brentano geometry, a Co
Kα source (1.789 Å), and a Lynxeye 1D silicon-drift detec-
tor. Prior to the acquisition, the samples were ground to
a top size of 63µm using a ring mill. Diffractograms were
acquired as coupled 2θ/θ scans on the 7–80°2θ range, with
a step size of 0.02° 2θ, a dwell time of 1.0 to 1.5 sec. per
step. The results are presented in Table 1. These results led
to some important observations: i) iron is mostly associ-
ated to magnetite, but also occurs in some gangue min-
erals (e.g., garnets, amphiboles and pyroxenes), indicating
that iron recovery will be affected ii) pyrrhotite increases
significantly after the pre-concentration step due to its high
magnetic susceptibility, indicating that it is from the mag-
netic type as explained by Rezvani et al. (2024).
Mineral liberation analysis were conducted using a
Zeiss Mineralogic system, which combines a Zeiss Sigma
300 SEM-FEG with the Zeiss Mining Mineralogic soft-
ware to control and setup the analysis strategy. For that, the
feed sample was divided into 6 size fractions and mounted
into polished epoxy blocks: 45–53 µm, 75–106 µm, 150–
212 µm, 300–425 µm, 600–850 µm and 1.18–1.65 mm.
The results in Figure 1 show that magnetite is very well
liberated, especially in the 300–425μm size range and finer
ones, with more than 90% of liberated magnetite particles.
This confirms the potential for a concentration at a smaller
size fraction than the one used in the current pre concen-
tration process. However, gangue minerals (i.e., quartz,
garnet, and calcite) show poorer liberation as compared to
magnetite, most likely because these gangue minerals are
commonly associated in fine-grained textures.
To study the effect of ore fineness/grinding on the
performance of magnetic separation, the feed sample was
ground using ring lab mills commonly used to pulverize
samples for chemical analysis at different times to produce
different particle size distributions. The particle size distri-
bution of the six ground products was determined using a
Malvern Panalytical Mastersizer 3000. Chemical analyses
were performed using X-ray fluorescence on fused beads for
major elements and Fe2+ (or FeO) titration using potassium
dichromate. The amount of magnetite content was also
determined using a Satmagan S135 Magnetic Analyzer.
These samples were then submitted to magnetic concen-
tration using a Davis Tube Tester (model EDT) adjusted
with the current of 0.48 A, resulting in a 1 kG magnetic
field and shaking level of 50 cycle/min for 18 min. The
data generated and summarized in Figure 2. The decrease in
Table 1. XRD Rietveld refinement results for Kentobe feed and cobbing concentrate samples
Group Mineral Formula Feed Cobb. Conc.
Fe oxide Magnetite Fe3O4 43 66,1
Garnets Andradite-grossular Ca
3 Fe
2 (SiO
4 )
3 – Ca
3 Al
2 (SiO
4 )
3 17 9,9
Ca amphibole Actinolite Ca2(Fe,Mg)5Si8O22(OH)2 9,7 1,2
Carbonate Calcite CaCO
3 7 2,8
Ca clinopyroxenes Diopside-hedenbergite CaMgSi
2 O
6 – CaMgSi
2 O
6 6,5 3,3
Chlorite Chamosite (Fe,Mg)6(Si,Al)4O10(OH,O)8 4,4 2,9
Si oxide Quartz SiO2 4,1 1
Epidote Epidote Ca
2 (Fe,Al)
3 (SiO
4 )
3 (OH) 3,5 1,9
Feldspars Albite NaAlSi
3 O
8 2,2 3,2
Orthoclase KAlSi3O8 0,7 0,3
Sulphides Pyrrhotite Fe1-xS (x=0 to 0.17) 1,2 6,1
Pyrite FeS
2 0,6 1,4
magnetitic concentrate. Connelly and Yan (2009) pre-
sented a few flowsheets for treating different magnetite
ores in Australia and mentioned some important trends
in magnetite processing, such as, the use of mineral lib-
eration analysis to understand the mineral behaviour, con-
sider coarse cobbing as a critical upgrading and application
of high-pressure grinding rolls (HPGR) to save grinding
energy.
The current evaluation aims to increase even further the
iron content in the concentrate, aligned with the sustain-
able goals to produce green steel. To accomplish the goal,
the work was developed in three stages: i) perform mineral
characterization of the ore ii) test in bench lab scale a pos-
sible flowsheet iii) investigate opportunities for optimizing
the beneficiation route.
DEVELOPMENT
Two sample were collected in the industrial plant before
(i.e., feed sample) and after the dry magnetic separation
(i.e., cobbing concentrate). The material presented a top
size of 40 mm and it was crushed in a lab jaw crusher mill
at a top size of 8 mm.
Powder X-ray diffraction (XRD) acquisitions were per-
formed using a Bruker D2 Phaser diffractometer, which
features a goniometer in Bragg-Brentano geometry, a Co
Kα source (1.789 Å), and a Lynxeye 1D silicon-drift detec-
tor. Prior to the acquisition, the samples were ground to
a top size of 63µm using a ring mill. Diffractograms were
acquired as coupled 2θ/θ scans on the 7–80°2θ range, with
a step size of 0.02° 2θ, a dwell time of 1.0 to 1.5 sec. per
step. The results are presented in Table 1. These results led
to some important observations: i) iron is mostly associ-
ated to magnetite, but also occurs in some gangue min-
erals (e.g., garnets, amphiboles and pyroxenes), indicating
that iron recovery will be affected ii) pyrrhotite increases
significantly after the pre-concentration step due to its high
magnetic susceptibility, indicating that it is from the mag-
netic type as explained by Rezvani et al. (2024).
Mineral liberation analysis were conducted using a
Zeiss Mineralogic system, which combines a Zeiss Sigma
300 SEM-FEG with the Zeiss Mining Mineralogic soft-
ware to control and setup the analysis strategy. For that, the
feed sample was divided into 6 size fractions and mounted
into polished epoxy blocks: 45–53 µm, 75–106 µm, 150–
212 µm, 300–425 µm, 600–850 µm and 1.18–1.65 mm.
The results in Figure 1 show that magnetite is very well
liberated, especially in the 300–425μm size range and finer
ones, with more than 90% of liberated magnetite particles.
This confirms the potential for a concentration at a smaller
size fraction than the one used in the current pre concen-
tration process. However, gangue minerals (i.e., quartz,
garnet, and calcite) show poorer liberation as compared to
magnetite, most likely because these gangue minerals are
commonly associated in fine-grained textures.
To study the effect of ore fineness/grinding on the
performance of magnetic separation, the feed sample was
ground using ring lab mills commonly used to pulverize
samples for chemical analysis at different times to produce
different particle size distributions. The particle size distri-
bution of the six ground products was determined using a
Malvern Panalytical Mastersizer 3000. Chemical analyses
were performed using X-ray fluorescence on fused beads for
major elements and Fe2+ (or FeO) titration using potassium
dichromate. The amount of magnetite content was also
determined using a Satmagan S135 Magnetic Analyzer.
These samples were then submitted to magnetic concen-
tration using a Davis Tube Tester (model EDT) adjusted
with the current of 0.48 A, resulting in a 1 kG magnetic
field and shaking level of 50 cycle/min for 18 min. The
data generated and summarized in Figure 2. The decrease in
Table 1. XRD Rietveld refinement results for Kentobe feed and cobbing concentrate samples
Group Mineral Formula Feed Cobb. Conc.
Fe oxide Magnetite Fe3O4 43 66,1
Garnets Andradite-grossular Ca
3 Fe
2 (SiO
4 )
3 – Ca
3 Al
2 (SiO
4 )
3 17 9,9
Ca amphibole Actinolite Ca2(Fe,Mg)5Si8O22(OH)2 9,7 1,2
Carbonate Calcite CaCO
3 7 2,8
Ca clinopyroxenes Diopside-hedenbergite CaMgSi
2 O
6 – CaMgSi
2 O
6 6,5 3,3
Chlorite Chamosite (Fe,Mg)6(Si,Al)4O10(OH,O)8 4,4 2,9
Si oxide Quartz SiO2 4,1 1
Epidote Epidote Ca
2 (Fe,Al)
3 (SiO
4 )
3 (OH) 3,5 1,9
Feldspars Albite NaAlSi
3 O
8 2,2 3,2
Orthoclase KAlSi3O8 0,7 0,3
Sulphides Pyrrhotite Fe1-xS (x=0 to 0.17) 1,2 6,1
Pyrite FeS
2 0,6 1,4