Minerals Engineering 207 (2024) 108544

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Solving the mystery of natural pyrite flotation – A

mineralogy-based approach
E. Forbes a, b, *, M. Jefferson a, U. Yenial-Arslan a, b, C. Curtis-Morar c, R. O’Donnell d
a
Julius Kruttschnitt Mineral Research Centre, Sustainable Minerals Institute, University of Queensland, QLD 4068
b
ARC Centre of Excellence on Eco-Efficient Beneficiation of Minerals (CE200100009), Shortland NSW, 2307
c
Long-term Strategy, Planning and Projects, Glencore Zinc, Queensland, Australia
d
Global Capital Studies and Projects, Glencore Zinc, Queensland, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Many different types of pyrite have been observed in the mineral processing industry, varying in texture,
Mineral Flotation chemical composition, electrochemical properties, and flotation response. One particular type of pyrite has been
Mineralogy noted for its unusual behaviour is framboidal or “carbonaceous” pyrite. Specifically, a long history of carbo­
Pyrite
naceous pyrite presence has been recorded and studied at Glencore’s Mount Isa Mines.
Statistical Analysis
Carbonaceous pyrite
This work is a detailed examination of the mineralogical characteristics of ore samples containing different
Chalcopyrite/pyrite interactions pyrite types, to determine the underlying drivers of natural pyrite flotation. The work also revisits the original
hypothesis that pyrite floatability is facilitated by the presence of carbonaceous material within the mineral
matrix.
Overall, this work closely mirrored that of previous studies, with the results closely aligned. The main dif­
ference is that when multiple ore domains are subjected to the same analysis, the flotation behaviour of
carbonaceous content and pyrite appear to be independent. Instead, excessive natural flotation recoveries of
pyrite are likely to be driven by factors that include the effect of galvanic interactions between chalcopyrite and
pyrite.

1. Introduction pyrite presence has been recorded and studied at Glencore’s Mount Isa
Mines.
Pyrite is the world’s most abundant sulphide mineral and can take
many forms. It can be a highly problematic gangue species in base metal
1.1. Pyrite at Mount Isa
sulphide flotation. Pyrite is known to readily report to flotation con­
centrates, diluting their value and resulting in excess sulphur dioxide
The history of framboidal pyrites at Mount Isa Mines goes back
production during smelting. The presence of pyrite in tailings can also be
nearly 100 years. Early mineralogical studies on the Mount Isa deposit
a source of acid mine drainage, resulting in significant environmental
identified two prominent pyrite textures in the literature: Pyrite I and
damage.
Pyrite II. Pyrite I refer to finely disseminated close-packed pyrite with
On the other hand, pyrite is a common host for valuable metals, such
grain sizes ranging from 5 to 10 µm. Pyrite II refers to coarse euhedral
as gold. It is also increasingly being seen as a source of other “critical”
and subhedral pyrite with grain sizes ranging between 0.5 and 5 mm
metals such as cobalt. It is therefore a key mineral for achieving a suf­
(Davey and Slaughter, 1970; Grondijs and Schouten, 1937; Gulson et al.,
ficient supply of these metals for the renewable energy sector.
1985).
Many different types of pyrite have been observed in the mineral
Pyrite I has also been reported to present framboidal or atoll struc­
processing industry, varying in texture, chemical composition, electro­
tures and be closely associated with carbonaceous matter. Grondijs and
chemical properties, and flotation response. One particular type of py­
Schouten (1937) identified pyrite atoll structures with diverse sulphide
rite has been noted for its unusual behaviour is framboidal or
core replacements of pyrite and pyrite grains surrounded by graphite
“carbonaceous” pyrite. Specifically, a long history of carbonaceous
shells They determined by point counting, that about 50 % of pyrite in

* Corresponding author.
E-mail address: [email protected] (E. Forbes).

https://doi.org/10.1016/j.mineng.2023.108544
Received 17 October 2023; Received in revised form 13 December 2023; Accepted 14 December 2023
Available online 22 December 2023
0892-6875/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
E. Forbes et al. Minerals Engineering 207 (2024) 108544

the Cu concentrator feed was framboidal. However, the samples used for 3. Materials and methods
the analysis represent the Cu concentrator feed on a specific day and
thus are not representative of the whole deposit. For a more detailed 3.1. Ore sample selection and characterisation
review of framboidal pyrites, their structures and orogenesis, please
refer to our earlier published work (Jefferson et al., 2023). The Mount Isa geology and geometallurgy team selected the ore
The mechanism behind the flotation properties of framboidal/ samples. The ore samples were received from two different orebodies,
carbonaceous pyrite was studied in detail by Grano et al. (1990). They OD1 and OD2. Ore domains containing mostly euhedral grains of pyrite
found that a significant portion of the pyrite material was consistently are referred to as coarse-grain pyrite (CGP), while ore domains with
naturally floatable across the entire Eh range. These findings conclude disseminated or very fine-sized pyrite grains are referred to as fine-
that the typical mechanism for pyrite flotation, i.e. sequential surface grained pyrite (FGP).
oxidation as a function of increasing surface potential, was not appli­ The OD1 domain contains coarse-grain pyrites (OD1-CGP), fine-
cable in this case. The mechanism of pyrite flotation was therefore grain pyrites (OD1-FGP) and leached zone (OD1-Leached). The OD2
entirely independent of electrochemistry. domain similarly contains both fine-grained and coarse-grained pyrite
To test this hypothesis, Grano et al. (1990) subjected the flotation textures. For a more detailed description of these samples’ core logging
concentrates to x-ray photoelectron spectroscopy (XPS). They found that and geometallurgical information, please consult our previous publica­
flotation concentrates contained significant quantities of graphitic car­ tions on the subject (Yenial-Arslan et al., 2023).
bon. This finding was in agreement with the original pyrite characteri­ Table 1 shows the modal mineralogy of the five ore samples as
sation work performed by Grondijs and Schouten (1937), who observed determined by mineral liberation analysis (MLA). A more detailed
that thin graphitic shells surrounded the fine grains of framboidal pyrite. analysis of selected mineralogical characteristics of the ore samples is
Therefore, the natural floatability of this pyrite type was ascribed to presented in Table A of the Appendix (Supplementary Materials).
the hydrophobic nature of graphite coatings of pyrite particles. How­ The samples selected for MLA analysis were also analysed for their
ever, it is important to stress that the presence of graphitic carbon within carbon content and speciation (Inorganic/organic and graphitic car­
the pyrite grains had only been visually observed using optical micro­ bon), with the results presented in Table 2.
scopy. The presence of carbon within the pyrite structure had not been
measured or quantified by other means. 3.2. Experimental procedures
The concept of the highly floatable, “carbonaceous” pyrite became
widely accepted within the processing philosophy at Mount Isa Mines. 3.2.1. Sample preparation
The copper processing circuit was built with a pre-flotation stage to aid The drill core samples were split into sub-samples of approximately
the early removal of floatable pyrites as well as other floatable gangue, 1.6 ± 0.07 kg mass using a rotary splitter. These sub-samples were
such as talc. New reagent schemes were developed and introduced sieved into seven size fractions: +1180 µm, − 1180 + 425 µm, − 425 +
specifically with the intent of depressing carbon and hence depressing 212 µm, − 212 + 106 µm, − 106 + 53 µm, − 53 + 38 µm and − 38 µm.
pyrite (O’Donnell and Muller, 2018). Each size fraction was characterised by Mineral Liberation Analysis
Since the original study in 1990, no further investigative study was (MLA) before grinding to preserve the ore textures for identification.
conducted to underpin the mechanism responsible for pyrite flotation For each test, 1.6 kg of sample was crushed in a Boyd crusher in a
further and “carbonaceous”, became a catch-all term for floatable pyrite closed circuit to 100 % finer than 3.35 mm. The crushing was done in
in general. multiple stages with different gaps (4 mm, 3 mm and 2 mm) to keep
consistency among the samples. Then, the freshly crushed ore was mixed
2. Aim and objectives with process water at a 60 % solids ratio by weight and wet ground in a
stainless-steel rod mill.
This work is part of a greater study that aims to build an under­ Each ore was ground to achieve a P80 of 106 µm. It should be noted
standing of the mechanisms of natural pyrite flotation for a wide array of that this grind size is significantly finer than what is currently used in the
different pyrite types. The work aims to use the enhanced understanding Mt Isa flotation circuit. This grind size was specifically chosen to achieve
of pyrite behaviour to help devise better methods for controlling pyrite a similar degree of liberation between ore samples, thus minimising the
flotation as well as creating better geo-metallurgical prediction capa­ effect of liberation as a variable in flotation outcomes. The mill was
bility for efficient mine planning. cleaned by grinding with sand for 10 min before each test. The mill was
Several aspects of the analysis (such as flotation behaviour and cleaned by grinding with sand for 10 min before each test.
certain ore characteristics) have been discussed in detail in our previous
publications (Jefferson et al., 2023; Yenial-Arslan et al., 2023). Only a 3.2.2. Reagents
brief overview of previously published data will be presented here, The analytical grade methyl isobutyl carbinol (MIBC) was used as a
where these data will be discussed in the context of new, previously
unpublished information.
Table 1
This work is a detailed examination of the natural flotation behav­
Mineralogical composition of the selected ore samples.
iour and mineralogical characteristics of ore samples containing
different pyrite types to determine the drivers of pyrite flotation. Five Mineral Wt % Composition
Grouping OD1- OD1- OD1- OD2- OD2-
different ore domains are subjected to flotation testing that closely CGP FGP Leached FGP CGP
mirrors the pre-flotation circuit of Mount Isa Mines copper concentrator.
Chalcopyrite 8.41 4.56 7.54 4.02 11.67
The outcomes of the flotation tests are then correlated to a wide range of
Silicates 28.06 34.82 70.40 30.03 33.64
mineralogical ore characteristics to determine the potential mechanism Ca Carbonates 55.45 33.32 1.72 27.39 39.57
that underpins flotation performance. Pyrite 4.30 22.23 11.69 34.60 11.56
The work also revisits the original hypothesis that pyrite floatability Pyrrhotite 0.13 0.14 0.12 0.49 1.32
is facilitated by the presence of carbonaceous material within the min­ Other Sulphides 0.01 0.03 0.24 0.01 0.01
Phyllosilicates 1.66 3.31 3.65 2.64 1.56
eral matrix. It should be noted that the work does not aim to optimise Cobaltite 0.00 0.07 0.02 0.01 0.00
flotation performance but rather to benchmark it under the same set of Galena 0.00 0.02 0.00 0.01 0.00
conditions to evaluate differences in floatability. Other 1.97 1.51 4.59 0.81 0.66
Deleterious 0.02 0.02 0.02 0.01 0.01
Py/Cp Ratio 0.51 4.88 1.55 8.61 0.99

2
E. Forbes et al. Minerals Engineering 207 (2024) 108544

Table 2 3.4. Data analysis

Carbon composition and speciation within the selected ore samples.
Carbon Type Wt % Composition 3.4.1. Entrainment analysis
OD1-CGP OD1-FGP OD1-Leached OD2-FGP OD2-CGP The recovery by entrainment of any given mineral can be modelled
Inorganic 7.51 4.29 0.47 3.9 5.47 by using the Entrainment Factor (ENTi), calculated for a known fully
Organic 0.12 0.44 0.45 0.74 0.67 liberated non-floatable gangue species (site), as shown in Eq. (1). In this
Graphitic <0.02 0.21 0.04 0.39 0.13 case, quartz was used as the non-floatable gangue species in question,
Total 7.63 4.73 0.92 4.64 6.14 resulting in Eq. (2). The degree of entrainment for any mineral (i) can
then be calculated by multiplying the entrainment factor by the re­
frother. No collector was used. The flotation gas was high-purity syn­ coveries of water and quartz (as a proxy for free gangue), using Eq. (3).
thetic air used at a flow rate of 8 L/min. The process water specifications Finally, the recovery by true flotation for a mineral (i) can be calculated
were used from the previous study (Forbes et al., 2022). The analytical using Eq.4.
grades of KCl, CaCl2, MgSO4, Na2SO4, and Na2CO3 were used in required (Massoffreegangueperunitmasswater)Con
amounts to make up process water. The analysis of this process water is ENT i = (1)
(Massoffreegangueperunitmasswater)Pulp
given in Table 3 below.
( W)
(M quartz ) M Pulp
3.2.3. Flotation test procedure ENT quartz = ( W ) Con × quartz (2)
M Con (M )Pulp
The flotation tests aimed to estimate the degree of natural floatability
of pyrite and provide a benchmark for comparison between the ore ( )
Ri = ENT quartz • Rwater (3)
domains. In each flotation test, 1.6 kg of freshly crushed ore was wet ent

ground in a stainless-steel rod mill at 60 % solid ratio, followed by ( ) ( ) ( )

flotation in a 3L bottom-driven Denver cell at 1200 rpm rotor speed. The Ri TF = Ri tot − Ri ent (4)
concentration of methyl isobutyl carbinol (MIBC) was kept at 20 ppm Where the symbols for Equations 1–––4 are defined as:
during the tests. The air was turned on one minute after the MIBC ENTi – Degree of Entrainment of
( )
Ri – Total recovery of component i
addition, and the concentrate was collected for 3 min by scraping every Component i
tot

10 s. ( )
MW – Mass of water Ri ent – Recovery of component i by
entrainment
3.3. Analytical measurements Mi – Mass of component i
( i)
R TF – Recovery of component i by true
flotation
All flotation products were processed and subsequently sent for assay
analysis at ALS Global Analytical Laboratory in Brisbane. The flotation
3.4.2. Statistical analysis
products were assayed for copper and iron by XRF following a lithium
All the flotation tests were performed in triplicate. Throughout the
borate fusion with the addition of strong oxidising agents to decompose
manuscript, the error bars represent the 95 % confidence of the mean
sulphide-rich ores. The total sulphur was analysed with an induction
values. The obtained flotation results were correlated to selected
furnace. For graphitic carbon assays, the carbonates were first removed
mineralogical data, as well as the characteristics of carbonaceous ma­
with HCl (50 %), then roasted to remove organic carbon, and then in­
terial present in both the feed and concentrate samples for the five ore
duction furnace/IR.
domains. In each case, a correlation coefficient (R) and the coefficient of
MLA analysis was made using the Mineral Liberation Analyser at
determination (R2) were calculated. These values were used to deter­
Julius Kruttschnitt Mineral Research Centre (JKMRC). An FEI Quanta
mine the Student’s t distribution for each correlation, using Equation
650 MLA equipped with a Bruker Silicon Drift Detector was used to
(5), where n equals the number of tests per domain (Napier-Munn,
measure the samples using the XBSE measurement mod. The mineral
2014).
conversions from elemental assay were validated with MLA analysis, the
√̅̅̅̅̅̅̅̅̅̅̅
comparison between the measured and calculated mineral assays can be R• n− 2
found in our previous publication (Yenial-Arslan et al., 2023). t = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (5)
1 − R2
XPS analysis was conducted to determine the presence of graphitic
carbon on the pre-float concentrate and investigate the relationship The resulting values of t were used to determine the statistical
between carbon and pyrite. This analysis was carried out with a Kratos probability associated with the correlation. The criteria for significance
Axis Supra Plus XPS equipped with an Al Kα (1486.6 eV) X-ray source were decided as follows:
with a beam spot size of 200 µm. To qualitative visualise areas of higher
concentration of Fe and graphitic carbon in the sample, a heatmap • The probability of significance had to exceed the 95 % confidence
representation of the spectra intensity for each species was prepared for limit.
a 2.0 mm by 2.5 mm area, with a 0.5 mm distance between measurement • The correlation needed to be significant for both the total recovery
points. It is important to note that the surface irregularities within the and recovery by true flotation values.
sample would also influence the spectra’ intensities.
The strict significance conditions were chosen for the following
reasons:

• To avoid “rogue” or weak trends

Table 3 • To ensure that the trend was representative of true flotation and not
Composition of the synthetic process water used in the flotation tests. entrainment contributions
Elements of interest Unit, mg/L • To ensure that the correlation would still hold in the absence of
Sulfate as SO4 - Turbidimetric 2170 rigorous entrainment analysis, which is often not performed during
Calcium 404 industrial geo-metallurgical programs
Magnesium 354
Sodium 458
Potassium 130

3
E. Forbes et al. Minerals Engineering 207 (2024) 108544

4. Results

This section describes the flotation results of prefloat/total recovery

with the entrainment calculations outcomes and kinetic modelling of
flotation tests. (See Figs. 1-3).

4.1. Flotation results

4.1.1. Total flotation Recovery

The flotation tests were designed to mimic the flotation process
within the Mount Isa Mines pre-flotation circuit. This circuit aims to
discard naturally floating gangue material while minimising the flota­
tion of valuable mineral species (Runge et al., 2021). Any chalcopyrite
that reports to the flotation concentrate represents a loss of valuable
copper. For this reason, chalcopyrite flotation recovery is referred to as
“chalcopyrite loss” throughout the document. The results for the batch
flotation test for both pyrite and chalcopyrite are presented in Fig. 4.
The results show that chalcopyrite losses are relatively low in the
system, ranging from circa 0.5 ± 0.14 % to just 3.84 ± 0.45 %. These
values are in line with what is observed in the Mount Isa Mines pre- Fig. 2. Natural flotation recovery of pyrite as a function of Eh (Grano
flotation circuit, where similar chalcopyrite losses were recorded in a et al., 1990).
2017 site survey (Runge et al., 2021).
On the other hand, the pyrite recoveries ranged from 4.5 ± 0.89 % to
13.2 ± 1.64 %. The pyrite flotation results clearly demonstrate that
different flotation domains exhibit significantly different pyrite re­
coveries. Overall, these values are similar to those obtained by Grano
et al. (1990). While a general similarity between these results and those
obtained by Grano et al. (1990) can be highlighted, it is essential to note
that the studies are not directly comparable. The work of Grano et al.
(1990) was conducted on a different ore domain, utilised a ceramic
grinding medium, and performed flotation for over 8 min. This study
was performed on 5 different ore domains using a stainless steel grinding
medium. The flotation was carried out over 3 min with a different feed
particle size distribution.
The obtained results are, therefore, sufficient to demonstrate that
both studies described a level of natural pyrite floatability. However, the
methodology differences are too significant to compare the recoveries
more closely.
Another important factor to take into consideration is the particle
size distribution of the obtained flotation concentrates, presented in
Fig. 5. In all cases, the concentrates were characterised by very fine

Fig. 3. X-Ray Photoelectron Spectra showing presence of graphitic carbon in

pyrite-rich flotation concentrates (Grano et al., 1990).

particle size distributions, with P80 values ranging between 13.6 and
11.9 µm. Particle sizes this fine are highly likely to be susceptible to un-
selective recovery by entrainment (Wang et al., 2016). For this reason, it
is important to evaluate these flotation recoveries in terms of both the
entrainment and true recovery components.

4.1.2. Recovery by true flotation

The overall flotation recoveries for pyrite and chalcopyrite were
evaluated to decouple both the entrainment and true flotation compo­
nents using Equations (1) through (4). For more details on this analysis
please consult previous publications (Yenial-Arslan et al., 2023). The
Fig. 1. Example of carbonaceous or framboidal pyrite at Mount Isa, exhibiting results of chalcopryite losses by both true flotation and entrainment are
atoll structures (O’Donnell and Muller, 2018).

4
E. Forbes et al. Minerals Engineering 207 (2024) 108544

Fig. 4. Pyrite recoveries and chalcopyrite losses after three minutes of

collector-less flotation.
Fig. 6. Chalcopyrite losses after three minutes of flotation expressed as both
true flotation and entrainment components.

Fig. 5. Particle size distributions and P80 values of flotation concentrates for
the five ore samples.

Fig. 7. Pyrite recoveries after three minutes of flotation expressed as both true
presented in Fig. 6. flotation and entrainment components.
The results confirm that a very significant proportion of chalcopyrite
loss can be attributed to loss by entrainment, whereby between 37 and mineralogical analysis was conducted on the combined size fractions of
92 % of all material is recovered via entrainment. This confirms the low the feed material as well as the flotation concentrates (please note that
natural floatability of chalcopyrite under the conditions representing the the MLA analysis was conducted on seven different size fractions, with
pre-flotation circuit at Mount Isa. the results for all + 38 mm material re-combined into a single dataset).
The pyrite recoveries obtained within the same tests are presented in Because the particle size distribution of the flotation concentrates was
Fig. 7. The results clearly show that unlike chalcopyrite most of the very fine (see Fig. 5), mineralogical characteristics of the − 38 µm frac­
pyrite was recovered by true flotation, with the entrainment contribu­ tion of the feed were also included in the analysis. These characteristics
tion ranging between 7 and 11 % of the overall recovery values. The included several categories:
recoveries by entrainment are very similar between the five different ore
domains and similar to those observed for chalcopyrite. This once again • Modal mineralogy of key mineral constituents.
confirms that the pyrite minerals present within these samples are • Pyrite liberation
characterised by a high degree of natural floatability under the condi­ • Pyrite mineral associations.
tions representing the pre-flotation circuit at Mount Isa. • The presence of both organic and graphitic carbon

5. Flotation drivers The correlation coefficients between flotation performance and

mineralogical characteristics were calculated, and their significance
Once the natural floatability of pyrite within five different ore do­ determined using Equation (5). The full list of the selected characteris­
mains was established, the results could be used to investigate the tics of the five ore domains, the matrix of the correlation coefficients
mineralogical drivers behind the flotation performance. The

5
E. Forbes et al. Minerals Engineering 207 (2024) 108544

associated with pyrite recoveries, as well as the probabilities of signifi­ mineral texture are well known to have a strong effect on flotation
cance associated with the correlation coefficients are listed in Tables A performance (Mishra et al., 2013; Seaman et al., 2012; Sutherland,
to C in the Appendix (supplementary material). Selected results are 1989; Tungpalan et al., 2019). This is because conventional MLA anal­
presented and discussed in the section that follows. ysis could not accurately estimate grain sizes of pyrites made up of fine-
grained clusters. An example image of MLA misclassification of small
pyrite clusters as large pyrite grains is shown in Fig. 8. For a more
5.1. The usual suspects detailed discussion of this phenomenon, please refer to our previous
publications on the subject (Jefferson et al., 2023).
The analysis of mineralogical characteristics typically associated
with flotation performance, such as overall feed mineralogy, particle 5.2. Effect of carbon
size distribution and pyrite liberation, is presented in Table 4. The use of
modal gangue mineralogy as a possible proxy for the identification of Grano et al. (1990) demonstrated that flotation concentrates con­
“carbonaceous” or highly floatable pyrite has been identified in previous taining naturally floating pyrite also contained graphitic carbon mate­
studies (Evans and O’Donnel, 2021). The strong link between the degree rial. It is therefore, possible that the differences in observed flotation
of liberation and flotation performance has been well established in behaviour could be driven by the varying presence of carbon within the
flotation literature (Farrokhpay and Fornasiero, 2017; Jameson, 2012; pyrite matrix.
Vianna, 2004), and is seen as one of the primary influencing factors in The flotation behaviour of carbonaceous material for the five
determining flotation performance. Finally, the effect of particle size selected ore domains was tested for both organic and graphitic carbon
distributions on flotation is one of the most well-established concepts in content. The carbon content of the flotation feed is summarised in
mineral processing (Trahar and Warren, 1976). However, the results Table 2. It is important to note that the presence of graphitic carbon in
clearly show that none of the tested factors were significantly correlated the flotation feed of the OD1 CGP sample was below the detection limit
with flotation performance for either the total recovery nor the true of 0.02 %. This made it impossible to determine the flotation recovery of
flotation recoveries of pyrite. graphitic carbon for this ore domain, even though significant quantities
In the case of modal mineralogy, factors such as pyrrhotite grade of carbon were detected in the flotation concentrate. For this reason, the
begin to approach significance but come short of the decided signifi­ flotation recoveries of graphitic carbon were excluded from the overall
cance criteria. Therefore, simple measurements such as basic or miner­ analysis.
alogical composition are insufficient to predict or explain pyrite To demonstrate the flotation behaviour of carbonaceous material,
flotation behaviour. both feed and concentrate grades of organic and graphitic carbon are
The particle size distribution in the flotation feed was kept consistent presented in Fig. 9 A and B. Fig. 9 A shows that organic carbon flotation
throughout the tests at the P80 of 106 µm. The particle size distribution concentrate varies very significantly between the different ores. For
of the five flotation concentrates were also very similar to one another some domains (OD1 FGP, OD1 Leach and OD2 FGP), organic carbon
(see Fig. 5), and it is therefore not surprising that no significant corre­ content appears similar to the feed material with little to no upgrade. In
lation was found between the size parameters and flotation outcomes. other cases (OD1 CGP and OD2 CGP), the concentrates significantly had
Similarly, the proportion of fully liberated pyrite particles in neither increased organic carbon content, suggesting a high degree of
the flotation feed nor the flotation concentrate resulted in significant floatability.
correlations between flotation recoveries. This is not surprising, as py­ On the other hand, graphitic carbon appears to readily enter the
rite liberation did not vary significantly between samples (43.37 % – flotation concentrate across all five tested domains, demonstrating sig­
52.11 % fully liberated pyrite across the sample set, see Table A in the nificant levels of upgrade. This clearly indicates that graphitic carbon
Appendix for details). This similarity was part of the experimental displays strong levels of natural floatability. This is expected as graphite
design, where the feed particle size distribution (P80 of 106 µm) was is known to be characterised by a high degree of natural hydrophobicity
specifically chosen so as to minimise the differences in pyrite liberation (Wakamatsu and Numata, 1991).
between samples. These results are in line with the findings of Grano et al. (1990),
This means that the differences between pyrite floatability are likely confirming that carbonaceous material contained within Mount Isa ore
caused by factors inherent to the orebody, rather than dependent on how readily enters flotation concentrates. However, further analysis is
the ore was processed. In other words, applying a different comminution required to see if carbon flotation is also a driver behind pyrite flotation
strategy to the ore domains that contain the finer pyrite grains (e.g. OD1 recovery.
FGP and OD2 FGP) to achieve better liberation or a more optimum Statistical correlations between the presence of carbon in both the
particle size distribution is unlikely to result in a relative difference in flotation feed and concentrates and pyrite flotation performance were
flotation performance between the different domains. calculated and tested for significance. The results of this analysis are
What is notably absent from this analysis is a correlation between presented in Table 5.
flotation performance and pyrite grain size, even though grain size and The carbon characteristics included in this analysis were the pres­
ence of carbon in the combined flotation feed, carbon concentrate grade,
Table 4 and the recovery of carbon in flotation concentrate. This analysis was
Probability of significance of correlation coefficients (R) between pyrite re­ performed for two carbon types, organic and graphitic.
coveries and mineralogical characteristics. The results presented in Table 5 demonstrate that pyrite flotation
Significance Probability Matrix (P) Pyrite Recovery behaviour did not correlate with any of the flotation characteristics of
Total True Flotation neither organic or graphitic carbon. There appears to be no significant
Modal Mineralogy in Combined Feed relationship between the floatability of pyrite and the floatability of
% Silicates 26 % 23 % organic and graphitic carbon. While both components are present in
% Calcium Carbonates 4% 10 % flotation concentrates, they appear to be floating independently of one
% Pyrite 64 % 61 %
another.
% Pyrrhotite 90 % 90 %
% Chalcopyrite 73 % 73 % The flotation concentrate obtained using the OD2 FGP orebody was
100 % Liberated Pyrite in Combined Feed 85 % 85 % further subjected to x-ray photoelectron spectroscopy (XPS) analysis to
100 % Liberated Pyrite in the − 38 µm Feed Fraction 36 % 40 % confirm the above results further. Fig. 10 presents a spatial representa­
100 % Liberated Pyrite in the Concentrate 30 % 32 % tion of the spectra intensity of iron and graphitic carbon in a 2.0 x 2.5
Particle Size Distribution in the Concentrate 12 % 9%
mm area.

6
E. Forbes et al. Minerals Engineering 207 (2024) 108544

Fig. 8. Comparison of pyrite grains in the backscatter electron (BSE) and mineral map images obtained using MLA analysis (Jefferson et al., 2023).

The results show that XPS analysis clearly detected the presence of set out in Section 3.4.2. The results show that out of the 38 tested ore
carbon in the flotation concentrate. Again this is entirely in line with the characteristics, only six have returned a significant correlation with total
results obtained by Grano et al. (1990). However, the sample appears to and true flotation pyrite recoveries.
have no spatial correlation between iron and carbon. Instead, carbon- While it would be a mistake to put too much faith in a single cor­
rich areas appear to mirror iron-poor areas, and vice versa. This relation/trend, the fact that all six of the significant characteristics share
further indicates that pyrite and carbon report to the flotation concen­ a relationship with chalcopyrite is a very strong indication that pyrite
trate independently. flotation behaviour is strongly linked to some form of chalcopyrite/py­
It is worth noting that the image area used in the XPS analysis was rite association.
2.0 x 2.5 mm, which was significantly smaller than the 8 x 8 mm image The impact of pyrite/chalcopyrite interactions in flotation is well-
Grano et al. (1990) used in their original carbonaceous pyrite study. known and established (Ekmekçi and Demirel, 1997; Forbes et al.,
However, even a higher image resolution available with the use of more 2018; Rao and Finch, 1988). This effect is a primary reason why these
modern XPS techniques is still insufficient to obtain an image that can minerals behave differently in single minerals versus binary or complex
distinguish between individual pyrite particles (P80 11 – 13 µm). This ore systems. The effect is primarily driven by the galvanic interactions
means that it is impossible to determine whether carbon is present taking place between two minerals with different rest potentials.
within the structure of individual pyrite particles, and a more sensitive These data do not provide sufficient information to definitively
high-resolution technique is required for this purpose. assign galvanic interactions as the primary mechanism for pyrite re­
coveries for these ore systems. Further study would be required to
5.3. Significant drivers for pyrite flotation performance definitively demonstrate the true drivers behind the variability in pyrite
recoveries. However, the solid link for chalcopyrite associations pro­
The results presented in previous sections have eliminated several vides an excellent potential proxy for assessment in geo-metallurgical
potential candidates for pyrite recovery drivers, including carbonaceous applications.
content, particle size distribution, and the degree of pyrite liberation. In addition, such a study would need to consider the full effect of
This leaves the question of which ore domain characteristics could be grinding media on the electrochemical environment of the flotation
responsible for differences in pyrite flotation performance. system and the galvanic activation/passivation of the minerals of
Table 6 lists all of the ore characteristics where the correlation co­ interest.
efficient with flotation performance has met all the significance criteria

7
E. Forbes et al. Minerals Engineering 207 (2024) 108544

Fig. 10. X-ray for electron spectra (carbon and iron) of a flotation concentrate
of the OD2 FGP or domain, containing 92% pyrite.

Fig. 9. Feed and concentrate grades of both (A) organic and (B) graphitic Table 6
carbon for the tested ore domains. Probability of significance of the correlation coefficients (R), listing all corre­
lations that meet the significance criteria.
Significance Probability Matrix (P) Pyrite Recovery
Table 5 Total True Flotation
Probability of significance of correlations coefficients (R) between pyrite re­ Combined Feed
coveries and carbon content. Cp/Py Ratio 99 % 99 %
Significance Probability Matrix (P) Pyrite Recovery 100 % Liberated Chalcopyrite 99 % 99 %
Total True Flotation Pyrite in Binaries with Chalcopyrite 99 % 99 %
Feed ¡38 µm Fraction
Carbon in the Combined Feed % Chalcopyrite 96 % 96 %
% Organic Carbon 32 % 30 % 100 % Liberated Chalcopyrite 99 % 99 %
% Graphitic Carbon 71 % 70 % % Chalcopyrite in Binaries with Pyrite 97 % 97 %
Carbon Concentrate Grade
% Organic Carbon 79 % 79 %
% Graphitic Carbon 69 % 70 % flotation was found to be consistent with previous results reported
Carbon Recovery
for a similar system. In each case, collectorless flotation resulted in
% Organic Carbon 15 % 11 %
minor levels of chalcopyrite loss, which were consistent with those
found in the Mount Isa Mines pre-flotation circuit.
6. Summary and conclusion • A rigorous statistical analysis was performed to determine the cor­
relations between pyrite flotation performance (for both total re­
This work examined a series of mineralogical characteristics of five coveries and recoveries by true flotation) and a set of selected ore
different ore domains to determine the ore’s specific properties that characteristics. The statistical significance of the correlation co­
drive pyrite’s natural performance. A specific focus was placed on efficients was also calculated.
evaluating the effect of graphitic carbon and its role in inducing the • No significant correlations were found between pyrite floatability
floatability of “carbonaceous” pyrite. The findings can be summarised as and the typical ore characteristics associated with determining
follows: flotation performance, including modal ore mineral content, pyrite
liberation and particle size distribution of both feed and concentrate.
• Each of the tested ore domains was found to contain naturally • In all cases, graphitic carbon was found to readily enter the flotation
floatable pyrite, with significant differences in the natural float­ concentrates, in line with previous work on a similar system. How­
ability of pyrite between the ore domains. The degree of pyrite ever, no significant correlations were found between pyrite

8
E. Forbes et al. Minerals Engineering 207 (2024) 108544

floatability and carbon content. This included the feed and concen­ Planning and Projects. Liza Forbes reports financial support was pro­
trate grade of both organic and graphitic carbon. vided by the Advance Queensland Industrial Research Fellowship
• Further XPS analysis of the flotation concentrate of one of the ore (AQIRF112). If there are other authors, they declare that they have no
domains (OD2 FGP) containing 92 % pyrite confirmed that no spatial known competing financial interests or personal relationships that could
correlation could be found between the iron and carbon content of have appeared to influence the work reported in this paper.
the sample.
• The only ore characteristics that were found to correlate significantly CRediT authorship contribution statement
with pyrite flotation performance were those linked to the presence
of chalcopyrite. The available data is insufficient to make a definitive E. Forbes: Conceptualisation, Methodology, Formal analysis,
conclusion regarding the exact mechanism involved. However, this Investigation, Writing – orignial draft, Writing – reviw & editing,
initial result points towards the influence of galvanic interactions Funding aquisition. M. Jefferson: Formal analysis, Investigation,
between chalcopyrite and pyrite as a strong driver behind pyrite Writing – review & editing, Data curation. U. Yenial-Arslan: Formal
recoveries. In addition, ore characteristics relating to chalcopyrite analysis, Investigation, Methodology, Data curation. C. Curtis-Morar:
(in particular, chalcopyrite/pyrite associations within the ores) Resources, Writing – review & editing. R. O’Donnell: Writing – review
could serve as proxies for identifying highly floatable pyrite in geo- & editing, Resources.
metallurgical analysis.
Declaration of competing interest
Overall, this work mirrored that performed by Grano et al. (1990),
with the results of the two studies very closely aligned. The main dif­ The authors declare that they have no known competing financial
ference is that when multiple ore domains are subjected to the same interests or personal relationships that could have appeared to influence
analysis, carbonaceous content and pyrite flotation behaviour appear the work reported in this paper.
independent. This puts significant doubt over the concept of “carbona­
ceous” pyrite as Mount Isa Mines ores’ major floatable sulphide gangue Data availability
component. Instead, excessive natural flotation recoveries of pyrite are
likely to be driven by different factors. These factors may include the The data that has been used is confidential.
effect of galvanic interactions between chalcopyrite and pyrite. How­
ever, further work is required to verify this hypothesis. Acknowledgement
An additional aspect of these naturally floating pyrites that requires
further attention is that of pyrite texture. It has been shown that stan­ The authors acknowledge the financial support of the Queensland
dard MLA analysis is insufficient to describe pyrite grain size accurately. Government and Glencore through the Advance Queensland Industrial
Therefore, a different way of analysing and quantifying pyrite texture is Research Fellowship (AQIRF112) in the completion of this work. The
necessary to investigate the impact of textural differences on pyrite authors would also like to further acknowledge the ARC Centre of
floatability thoroughly. Excellence on Eco-Efficient Beneficiation of Minerals (CE200100009),
Declaration of interests. for the ongoing support of the author’s research activities. Finally, the
Liza Forbes reports financial support, equipment, drugs, or supplies, authors would like to acknowledge Prof Tim Napier Munn for his advice
and travel were provided by Glencore Zinc Long-Term Strategy, on the statistical analysis methodology.

Appendix

Full data Tables

Table A. Selected Ore Characteristics.

Selected Characteristics Ore Domains

OD1 CGP OD1 FGP OD1 Leach OD2 FGP

COMBINED FEED
Modal Mineralogy
% Silicates 28.06 34.82 70.40 30.03
% Calcium Carbonates 55.45 33.32 1.72 27.39
% Pyrite 3.85 22.48 12.84 34.67
% Pyrrhotite 0.11 0.09 0.05 0.33
%Chalcopyrite 6.99 4.13 5.75 3.50
Cp/Py Ratio 1.82 0.18 0.45 0.10
Pyr/Cp Ratio 0.03 0.00 0.00 0.01
Pyr/Py Ratio 0.03 0.00 0.00 0.01
% Organic Carbon 0.12 0.44 0.45 0.74
% Inorganic Carbon 7.51 4.29 0.47 3.90
% Graphitic Carbon 0.02 0.21 0.04 0.39
% Total Carbon 7.63 4.73 0.92 4.64
Liberation Characteristics
100 % Liberated Pyrite 48.89 45.42 52.11 43.37
100 % Liberated Chalcopyrite 79.77 67.57 71.60 63.67
Sulphide Mineral Associations
% Pyrite in Binaries with Chalcopyrite 2.61 1.64 2.13 1.57
% Pyrite in Binaries with Pyrrhotite 2.29 0.68 3.55 1.68
(continued on next page)

9
E. Forbes et al. Minerals Engineering 207 (2024) 108544

(continued )
Selected Characteristics Ore Domains
OD1 CGP OD1 FGP OD1 Leach OD2 FGP

% Chalcopyrite in Binaries with Pyrite 1.71 76.42 5.70 9.03

% Cpt in Binaries with Pyrrhotite 0.06 67.57 0.05 0.51
FEED ¡38 µm FRACTION
Modal Mineralogy
%Calcium Carbonates 55.68 33.17 2.11 28.38
%Silicates 22.86 26.27 57.83 20.74
%Pyrrhotite 0.16 0.22 0.13 0.48
%Pyrite 4.95 27.32 15.36 40.22
%Chalcopyrite 10.92 6.27 10.45 5.17
Cp/Py Ratio 2.21 0.23 0.68 0.13
Py/Pyrh Ratio 0.03 0.01 0.01 0.01
Liberation Characteristics
100 % Liberated Pyrite 74.01 81.57 64.72 82.21
100 % Liberated Chalcopyrite 70.38 56.00 58.58 51.28
Mineral Associations
% Pyrite in Binaries with Chalcopyrite 0.89 1.20 1.20 1.12
% Pyrite in Binaries with Pyrrhotite 1.50 1.06 1.06 1.63
% Chalcopyrite in Binaries with Pyrite 1.63 6.98 6.98 9.24
% Chalcopyrite in Binaries with Pyrrhotite 0.01 1.31 0.05 0.70
CONCENTRATE
Particle Size Distribution 13.15 13.03 15.63 13.04
100 % Liberated Pyrite 74.01 81.57 64.72 82.21
Carbon Concentrate Grade
Organic Carbon 4.61 0.80 0.38 0.51
Inorganic Carbon 4.65 3.50 5.33 3.62
Graphitic Carbon 2.01 2.03 2.71 2.17
Carbon Recovery
Organic Carbon Pre-Float Recovery 0.50 0.57 2.05 0.54
Inorganic Carbon Pre-Float Recovery 31.64 24.40 30.10 20.28

Table B. Correlation Coefficient (R) Matrix with Pyrite Recoveries.

Correlation Matrix (R) Pyrite Recovery
TOTAL TRUE FLOTATION

COMBINED FEED PROPERTIES

Modal Mineralogy
% Silicates 0.21 0.18
% Calcium Carbonates − 0.53 − 0.50
% Pyrite 0.81 0.80
% Pyrrhotite − 0.62 − 0.61
%Chalcopyrite − 0.95 − 0.95
Cp/Py Ratio − 0.86 − 0.85
Pyr/Cp Ratio − 0.77 − 0.77
Pyr/Py Ratio 0.26 0.24
% Organic Carbon − 0.56 − 0.54
% Inorganic Carbon 0.59 0.59
% Graphitic Carbon − 0.56 − 0.53
% Total Carbon 0.67 0.64
Liberation Characteristics
100 % Liberated Pyrite − 0.74 − 0.75
100 % Liberated Chalcopyrite − 0.96 − 0.96
Sulphide Mineral Associations
% Pyrite in Binaries with Chalcopyrite − 0.96 − 0.96
% Pyrite in Binaries with Pyrrhotite − 0.21 − 0.24
% Chalcopyrite in Binaries with Pyrite 0.64 0.67
% Chalcopyrite in Binaries with Pyrrhotite 0.57 0.61
FEED ¡38 µm FRACTION PROPERTIES
Modal Mineralogy
%Calcium Carbonates − 0.54 − 0.51
%Silicates 0.20 0.17
%Pyrrhotite − 0.54 − 0.54
%Pyrite 0.76 0.75
%Chalcopyrite − 0.89 − 0.89
Cp/Py Ratio − 0.83 − 0.82
Py/Pyrh Ratio 0.72 0.71
Liberation Characteristics
100 % Liberated Pyrite − 0.29 − 0.32
100 % Liberated Chalcopyrite − 0.90 − 0.90
Mineral Associations
% Pyrite in Binaries with Chalcopyrite − 0.57 − 0.57
% Pyrite in Binaries with Pyrrhotite − 0.70 − 0.71
% Chalcopyrite in Binaries with Pyrite 0.92 0.91
% Chalcopyrite in Binaries with Pyrrhotite 0.80 0.82
SELECTED CONCENTRATE PROPERTIES
(continued on next page)

10
E. Forbes et al. Minerals Engineering 207 (2024) 108544

(continued )
Correlation Matrix (R) Pyrite Recovery
TOTAL TRUE FLOTATION

Particle Size Distribution 0.10 0.07

100 % Liberated Pyrite 0.23 0.26
Carbon Concentrate Grade
Organic Carbon − 0.67 − 0.67
Inorganic Carbon − 0.74 − 0.72
Graphitic Carbon − 0.57 − 0.59
Carbon Recovery
Organic Carbon Pre-Float Recovery 0.15 0.11
Inorganic Carbon Pre-Float Recovery 0.77 0.77

Table C. Matrix of Probabilities of Significance (P) of the Correlation Coefficients.

Significance Probability Matrix (P) Pyrite Recovery
TOTAL TRUE FLOTATION

COMBINED FEED PROPERTIES

Modal Mineralogy
% Silicates 26 % 23 %
% Calcium Carbonates 4% 10 %
% Pyrite 64 % 61 %
% Pyrrhotite 90 % 90 %
%Chalcopyrite 73 % 73 %
Cp/Py Ratio 99 % 99 %
Pyr/Cp Ratio 94 % 93 %
Pyr/Py Ratio 87 % 87 %
% Organic Carbon 32 % 30 %
% Inorganic Carbon 68 % 65 %
% Graphitic Carbon 71 % 70 %
% Total Carbon 67 % 64 %
Liberation Characteristics
100 % Liberated Pyrite 85 % 85 %
100 % Liberated Chalcopyrite 99 % 99 %
Sulphide Mineral Associations
% Pyrite in Binaries with Chalcopyrite 99 % 99 %
% Pyrite in Binaries with Pyrrhotite 26 % 31 %
% Chalcopyrite in Binaries with Pyrite 76 % 79 %
% Chalcopyrite in Binaries with Pyrrhotite 69 % 73 %
FEED ¡38 µm FRACTION PROPERTIES
Modal Mineralogy
%Calcium Carbonates 66 % 62 %
%Silicates 25 % 22 %
%Pyrrhotite 65 % 65 %
%Pyrite 87 % 86 %
%Chalcopyrite 96 % 96 %
Cp/Py Ratio 92 % 91 %
Py/Pyrh Ratio 93 % 93 %
Liberation Characteristics
100 % Liberated Pyrite 36 % 40 %
100 % Liberated Chalcopyrite 96 % 96 %
Mineral Associations
% Pyrite in Binaries with Chalcopyrite 68 % 68 %
% Pyrite in Binaries with Pyrrhotite 81 % 82 %
% Chalcopyrite in Binaries with Pyrite 97 % 97 %
% Chalcopyrite in Binaries with Pyrrhotite 90 % 91 %
SELECTED CONCENTRATE PROPERTIES
Particle Size Distribution 12 % 9%
100 % Liberated Pyrite 30 % 32 %
Carbon Concentrate Grade
Organic Carbon 79 % 79 %
Inorganic Carbon 84 % 83 %
Graphitic Carbon 69 % 70 %
Carbon Recovery
Organic Carbon Pre-Float Recovery 15 % 11 %
Inorganic Carbon Pre-Float Recovery 77 % 77 %

References Evans, C., O’Donnel, A.R., 2021. Back to basics – a practical approach to troubleshooting
plant grade/recovery issues in the Mt Isa Copper Concentrator. In: Mill Operators
Conference 2021. Brisbane, Australia, Brisbane, Australia, pp. 456–467.
Davey, J., Slaughter, P., 1970. Changes in lead-zinc flotation practice at Mount Isa Mines
Farrokhpay, S., Fornasiero, D., 2017. Flotation of coarse composite particles: Effect of
Limited, In AIME World Symposium on Mining and Metallurgy of Lead and Zinc. AIME
mineral liberation and phase distribution. Advanced Powder Technology 28 (8),
401–427.
1849–1854.
Ekmekçi, Z., Demirel, H., 1997. Effects of galvanic interaction on collectorless flotation
Forbes, E., Britoe Abreu, S., Tungpalan, K., Sashigunan, R., Runge, K., O’Donnell, R.,
behaviour of chalcopyrite and pyrite. International Journal of Mineral Processing 52
2022. Effect of residual reagents on chalcopyrite losses at Mount Isa Mines copper
(1), 31–48.

11
E. Forbes et al. Minerals Engineering 207 (2024) 108544

operation: Part I – evaluation of mineral recoveries. Minerals Engineering 185, Rao, S.R., Finch, J.A., 1988. Galvanic Interaction Studies on Sulphide Minerals. Canadian
107706. Metallurgical Quarterly 27 (4), 253–259.
Forbes, E., Smith, L., Vepsalainen, M., 2018. Effect of pyrite type on the electrochemistry Runge, K., Brito e Abreu, S., Evans, C., O’Donnel, R., 2021. Analysis of the Mount Isa
of chalcopyrite/pyrite interactions. Physicochemical Problems of Mineral Processing Copper concentrate preflotation circuit using advanced diagnostic techniques In Mill
54 (4), 1116–1129. Operators Conference 2021. AusIMM, Brisbane, Australia, pp. 178 - 192.
Grano, S., Ralston, J., Smart, R.S.C., 1990. Influence of electrochemical environment on Seaman, D.R., Burns, F., Adamson, B., Seaman, B.A., Manton, P., 2012. Telfer processing
the flotation behaviour of Mt. Isa copper and lead-zinc ore. International Journal of plant upgrade–the implementation of additional cleaning capacity and the
Mineral Processing 30 (1), 69–97. regrinding of copper and pyrite concentrates. In: Proceedings 11th AusIMM Mill
Grondijs, H.F., Schouten, C., 1937. A study of the Mount Isa ores [Queensland, Operators’ Conference. AusIMM, Hobart, TAS, pp. 373–381.
Australia]. Economic Geology 32 (4), 407–450. Sutherland, D.N., 1989. Batch flotation behaviour of composite particles. Minerals
Gulson, B.L., Perkins, W.G., Mizon, K.J., 1985. Lead isotope studies bearing on the Engineering 2 (3), 351–367.
genesis of copper orebodies at Mount Isa, Queensland and Mount Isa silica dolomite Trahar, W.J., Warren, L.J., 1976. The floatability of very fine particles — A review. Int. J.
and copper orebodies the result of a syntectonic hydrothermal alteration system Miner. Process 3, 103–131.
reply. Economic Geology and the Bulletin of the Society of Economic Geologists 80 Tungpalan, K., Wightman, E., Evans, C., 2019. Key ore textures influencing separation
(1), 189–190. behaviour of ores, In Life with Ore Deposits On Earth, Proceedings of the 15Th SGA
Jameson, G.J., 2012. The effect of surface liberation and particle size on flotation rate Biennial Meeting. Society for Geology Applied to Mineral Deposits (SGA), pp. 1524-
constants. Minerals Engineering 36–38, 132–137. 1527.
Jefferson, M., Yenial-Arslan, U., Evans, C., Curtis-Morar, C., O’Donnell, R., Parbhakar- Vianna, S.M.S.M., 2004. The effect of particle size, collector coverage and liberation on
Fox, A., Forbes, E., 2023. Effect of pyrite textures and composition on flotation the floatability of galena particles in an ore, In Floatability of galena particles in an ore.
performance: A review. Minerals Engineering 201, 108234. Thesis (Ph.D.) - University of Queensland, 2004.
Mishra, G., Viljoen, K.S., Mouri, H., 2013. Influence of mineralogy and ore texture on Wakamatsu, T., Numata, Y., 1991. Flotation of graphite. Minerals Engineering 4 (7),
pentlandite flotation at the Nkomati nickel mine. South Africa. Minerals Engineering 975–982.
54, 63–78. Wang, L., Runge, K., Peng, Y., Vos, C., 2016. An empirical model for the degree of
Napier-Munn, T.J., 2014. Statistical Methods for Mineral Engineers: How to Design entrainment in froth flotation based on particle size and density. Minerals
Experiments and Analyse Data. Julius Kruttschnitt Minerals Research Institute. Engineering 98, 187–193.
O’Donnell, A.R., Muller, M.L., 2018. Selectively targeting hydrophobic gangue minerals Yenial-Arslan, U., Jefferson, M., Curtis-Morar, C., Forbes, E., 2023. Pathway to Prediction
at the Mount Isa Mines Copper Concentrator. In: 14th AUSIMM Mill Operators of Pyrite Floatability from Copper Ore Geological Domain Data. Minerals 13 (6),
Conference. AUSIMM, Brisbane, Australia, pp. 413–423. 801.

12