Int. J. Miner. Process.

93 (2009) 256266

Contents lists available at ScienceDirect

International Journal of Mineral Processing

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o

A study of mechanisms affecting molybdenite recovery in a bulk copper/

molybdenum otation circuit
M. Zanin , I. Ametov, S. Grano, L. Zhou, W. Skinner
Ian Wark Research Institute, University of South Australia (The ARC Special Research Centre for Particle and Material Interfaces), Mawson Lakes Campus, Adelaide, South Australia 5095,
Australia

a r t i c l e i n f o a b s t r a c t

Article history: Molybdenite otation in the bulk copper/molybdenum otation circuit at Kennecott Utah Copper was
Received 23 April 2009 studied by means of a combination of plant metallurgical surveys, laboratory otation tests, mineralogical
Received in revised form 2 October 2009 analysis (QEM-Scan), surface analysis (ToF-SIMS) and contact angle measurements. It was demonstrated
Accepted 3 October 2009
that molybdenite recovery is inuenced by otation feed solids percent and by the mineralogy of the host
rock. Molybdenite recovery was consistently higher at reduced otation feed solids percent. Furthermore,
Keywords:
the recovery of molybdenite was signicantly lower from otation feeds with high limestone skarn ore
Froth otation
Molybdenite content. The major factors affecting the otation recovery of molybdenite from both porphyry and skarn
Sulphide ores copper ores are discussed. It is suggested that the lower otation recovery of molybdenite compared to the
Ore mineralogy copper sulphide is determined by several factors, including particle morphology, inherent hydrophobicity
and possible formation of slime coatings in the presence of gangue minerals typical of skarn ores.
Implications on plant performance are discussed, and solutions to restore molybdenite recovery presented.
2009 Elsevier B.V. All rights reserved.

1. Introduction 1961). The occurrence of large molybdenite crystals in vein controlled

Characteristically, in porphyry copper otation plants, molybde-
num exhibits lower recovery than copper, in spite of the apparent
natural hydrophobicity of molybdenite (Kelebek, 1988). Furthermore,
molybdenum recovery displays high variability. While copper
recovery is usually between 80% and 90%, molybdenum recovery
may range between 25% and 85% (Crozier, 1979). Among the copper
sulphide minerals, chalcopyrite usually has higher otation rate and
recovery, but also chalcocite, bornite, digenite and covellite can be
recovered to values higher than 80% if the relevant electrochemical
conditions are maintained in the slurry to minimise surface oxidation
(Orwe et al., 1998). Molybdenite recovery, on the contrary, may vary
signicantly from operation to operation, and also within different ore
bodies in the same operation. Copper and molybdenum recovery data
for a one year period of a typical porphyry copper otation plant are
reported in Fig. 1, showing high variability of molybdenum recovery
with time.
Extensive research has been carried out in an attempt to link the
otation response of molybdenite to the mineral crystal structure,
textural features and lithology. The degree of crystallisation is one of
the factors that have been reported to affect molybdenite otation
(Hernlund, 1961; Shirley, 1981; Podobnik and Shirley, 1982). Well-
crystallised molybdenite is considered fast oating, while the almost
amorphous variety is either slow oating or non-oating (Hernlund,
Corresponding author. Tel.: +61 8 8302 3263; fax: +61 8 8302 3683.
E-mail address: massimiliano.zanin@unisa.edu.au (M. Zanin).
mineralisation as opposed to nely disseminated molybdenite has also
been reported (Sutulov, 1975; Podobnik and Shirley, 1982), but no
clear connection with otation response has been established. Triffett
and Bradshaw (2008) demonstrated that molybdenite particles with
high aspect ratio (major axis over minor axis) have higher probability
of reporting to the concentrate, and that coarse particles with high
perimeter to area ratio tend to report to the tailings. It could be noted
that aspect ratio itself may not be the critical factor, but it may be a
proxy for hydrophobicity, as discussed further below.
Ametov et al. (2008) conducted a series of surveys at different
porphyry copper otation plants. In the operations investigated, the
recovery of molybdenite in the rougher/scavengers of the bulk Cu/Mo
otation circuit was consistently lower than the recovery of copper
sulphide, the difference ranging from about 2% to 12% (Ametov et al.,
2008). Furthermore, a reduction in the otation feed solids percent
(expressed as solids percent in the slurry by weight) resulted in an
increase in molybdenite recovery, while copper otation was almost
unaffected (Ametov et al., 2008). The different behaviour of copper
minerals and molybdenite with respect to feed solids percent was
explained in terms of otation hydrodynamics (Ametov et al., 2008).
1.1. Mechanisms affecting molybdenite recovery
The low and highly variable otation recovery of molybdenum
may be a result of several factors, all related to the properties of the
molybdenite (MoS2) mineral. Molybdenite crystal structure consists
of hexagonal layers of molybdenum atoms between two layers of

0301-7516/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.minpro.2009.10.001
 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266
Fig. 1. Historical recovery data for a typical porphyry copper plant. Bulk copper/
molybdenum otation circuit.
sulphur atoms (Fig. 2). Strong covalent bonds act within SMoS
layers, but only weak van der Waals forces between adjacent SS
sheets (Lince and Frantz, 2000). This strong anisotropy causes
preferential cleavage of the molybdenite crystal along the adjacent
SS sheets. As a result, during grinding, platelet shaped fragments,
exfoliating from larger particles, are generally produced. Furthermore,
Fig. 2. Crystal structure of molybdenite (from Lince and Frantz, 2000).
257
the particles formed are characterised by strongly hydrophobic and
inert faces and hydrophilic and reactive edges, generated by the
breakage of the covalent bonds. These peculiarities determine the
otation behaviour of molybdenite, which is a combination of (a)
particle morphology (shape and size) in relation to hydrodynamics
(Ametov et al., 2008), (b) particle inherent hydrophobicity, an
important factor controlled to a degree by the face/edge ratio
(Chander and Fuerstenau, 1972; Hoover, 1980), (c) particleparticle
interactions between molybdenite and gangue minerals (Raghavan
and Hsu, 1984), and (d) particle recovery across the froth phase
(Dippenaar, 1982; Zanin et al., 2008). The possible effect of each
contributing factor is discussed further below, and it is a purpose of
this paper to probe the relative importance of some of these
mechanisms.
1.2. Hydrodynamic effects (a)
In otation, the rate at which particles are removed from the slurry
by air bubbles can be represented by (Newell and Grano, 2007):
dNp
= kNp = Zpb Ecoll 1
dt
in which Np is the number of particles in the slurry at time t, k the
otation rate constant, Zpb the collision frequency, and Ecoll the
collection efciency. The collection efciency Ecoll can be described as
a product of the collision (Ec), attachment (Ea) and stability (Es)
efciencies:
Ecoll = Ec Ea Es 2
Hydrodynamic conditions have direct inuence on both collision
frequency and collection efciency. Important hydrodynamic para-
meters are bubble diameter, bubble velocity and turbulent energy
dissipation (Newell and Grano, 2007).
Ametov et al. (2008) argued that, molybdenite particles, due to
their peculiar shape factor, could be more sensitive to hydrodynamic
effects than copper mineral particles. In an agitated slurry, platelet
shaped molybdenite particles may align along streamlines of the
suspending liquid and, therefore, have lower probability of collision
with bubbles. In effect, this hypothesis purports that the hydrody-
namic diameter is the minimum dimension of the platelets. Increasing
turbulence would increase collision frequency and efciency, and
therefore increase the rate of particle collection. This could be
achieved either by increasing the impeller rotational speed or by
reducing the feed solids percent. Damping of local energy dissipation
occurs in suspensions containing higher volume percent of solids, as
higher volume percent of solids produces higher slurry viscosity,
particularly in the case of interacting particles (Schubert, 1999).
Reducing the feed solids percent reduces the slurry viscosity and
increases turbulence, having in turn a positive effect on particle-
bubble collision efciency (Ec).
1.3. Inherent hydrophobicity (b)
Due to the cleavage mechanisms of molybdenite, very thin
particles are potentially produced in grinding. The overall degree of
hydrophobicity of molybdenite particles depends on the relative
surface exposure of faces (hydrophobic) and edges (hydrophilic). In
otation, molybdenite particles with higher exposure of edges will
have lower probability of attachment to air bubbles. On the contrary,
particles with a high face/edge ratio will have higher probability of
being recovered, in agreement with the ndings of Triffett and
Bradshaw (2008). The overall degree of hydrophobicity of molybde-
nite particles can also be reduced by the adsorption of metal ions in
solution (Raghavan and Hsu, 1984). Molybdenite has negative zeta
258 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266

potential across a wide pH range (Wie and Fuerstenau, 1974). Table 1

However, adsorption of positively charged ions at the edges may Average head grade of quartzite and limestone skarn ore samples.

reduce the magnitude, or even reverse the sign of the zeta potential. Cu Fe Mo S SiO2 Al2O3 CaO K2O MgO
Calcium ions, in particular, have been shown to adsorb in the [%] [%] [ppm] [%] [%] [%] [%] [%] [%]
intermediate to high pH range (Healy, 1984). Adsorbed Ca2+ ions Quartzite 0.50 1.9 600 0.4 63 13.8 2.0 7.0 5.0
on the particle edges may reduce the contact angle and otability of Skarn 0.45 9.6 21 1.5 44 1.7 24 0.3 2.7
molybdenite particles.
Furthermore, signicant deformation of particles may occur in
tumbling mills (Hoover, 1980), and it is common to nd bent, otation feeds (synthetic mineral mixtures and blends of different ore
distorted, or striated molybdenite particles in the otation feed types) have been carried out at the Ian Wark Research Institute.
(Triffett and Bradshaw, 2008). Under these conditions, higher
exposure of edges may occur, due to the breakage of the covalent
bonds, and the spatial distribution and orientation of the hydrophilic 2. Materials and methods
edges may determine the probability of attachment of particles to air
bubbles on collision. Oily collector is typically added in molybdenite 2.1. Ores
otation to enhance the mineral's hydrophobicity. The presence of
exposed hydrophilic edges on the bent and distorted particles may The Kennecott Utah Copperton concentrator treats porphyry
also prevent spreading of oil droplets on hydrophobic surfaces, thus copper ore, in which molybdenite occurs as a minor phase. Typically,
reducing the effect of the collector. Mo concentrations range from 0.02% to 0.06%. Distinct zones of
mineralisation and alteration are mined at Kennecott, such that four
1.4. Inter-particle interactions with gangue minerals (c) main ore types have been described (Triffett and Bradshaw, 2008),
including a quartzite ore showing high molybdenum grade
The high reactivity of molybdenite edges may also determine and recovery and a more difcult to oat, low grade, limestone
particleparticle interactions with other minerals in the slurry, in the skarn ore.
form of slime coatings. Raghavan and Hsu (1984) showed that, in the In the current study, plant surveys have been undertaken with the
presence of Ca2+ ions, the addition of silica to a system of molybdenite plant processing blends of the four main ore types in different
particles causes a sharp decrease in molybdenite otation recovery. It proportions, and laboratory otation tests have been performed both
is possible that the calcium ions act as a bridge, favouring adhesion on plant slurries and controlled blends of quartzite and skarn ores,
between negatively charged molybdenite and silica particles. This prepared at laboratory scale. Average head grades of the quartzite and
mechanism may be relevant in the otation of an ore with specic skarn ore samples are reported in Table 1, and the mineralogical
gangue mineralisation and dissolution of ions. Inter-particle interac- composition (modal distribution by QEM-Scan) in Table 2. The two
tions between some of the gangue minerals and molybdenite may ores present distinct gangue minerals: large amounts of quartz and
cause slime coatings on the latter, reducing its otation recovery. This feldspar in the quartzite ore sample, and actinolite, andradite, talc and
effect could be signicant at high pH and high alkalinity, due to the calcite abundant in the limestone skarn ore sample (shown in Table 2
bridging effect of the adsorbed Ca2+ ions. This possibility is explored as the Skarn ore).
in this paper. Single mineral molybdenite samples from WILLYAMA-GEO Dis-
coveries were used to produce coarse molybdenite particles which
1.5. Froth phase recovery (d) were used for contact angle measurement by the sessile drop method.

Inherent hydrophobicity and shape may also play a role in the

2.2. Reagents
transportation of molybdenite particles across the froth phase.
Dippenaar (1982) and Hemmings (1981) found a correlation between
The reagents used in this study, both in the plant and laboratory
size and hydrophobicity of the suspended particles and destabilisation
tests, were: oily collector for molybdenite (generic diesel oil was used
of the froth phase by lm rupture. For a given particle size, the greater
in the laboratory tests), at a typical addition rate of 22 g/t, dicresyl-
the contact angle, the greater is the destructive compressive stress
dithiophosphate (S-8989) collector for copper minerals, at a typical
induced on the thin lm (Hemmings, 1981). This implies that
addition rate of 20 g/t, and methylisobutyl-carbinol (MIBC) as frother
intermediate to coarse particles having contact angle greater than
(15 to 35 g/t). The pH was adjusted by adding lime, with the exception
90 are effective lm breakers in froth otation (Dippenaar, 1982).
of some diagnostic tests in which KOH was used, as outlined in 2.3.4.
Flat and elongated molybdenite particles may fall into this category,
In the laboratory otation tests carried out at the Ian Wark Research
and have, for this reason, lower recovery across the froth phase (Zanin
Institute, synthetic process water was used, prepared by dissolving
et al., 2009), also called froth recovery (Savassi et al., 1997). In a
separate investigation (Zanin et al., 2008), it was shown that froth
recovery in the roughers of a porphyry copper otation plant
Table 2
decreases signicantly down the bank (from 60% in the rst cells to Mineralogical analysis of quartzite and limestone skarn ore samples.
20% in the last cells), and that froth recovery of molybdenite is
Mineral Formula Modal [%]
generally lower than froth recovery of the copper sulphide. Stability of
the froth phase is therefore a factor which should be taken into Quartzite Skarn
consideration at plant scale, particularly in the rougher/scavengers, Quartz SiO2 24.3 10.9
where major losses of coarse molybdenite occur. In this paper, K_Feldspar KAlSi3O8 38. 7 1.0
however, the focus was on identifying factors which affect molybde- Plagioclase NaxCa(1 x) (Al,Si)AlSi2O8 8.1 0.02
Montmorillonite (Na,Ca)(Al,Mg)6(Si4O10)3(OH)6 nH2O 4. 7 0.04
nite recovery in the collection zone, while froth phase issues have
Biotite K(Fe,Mg)3AlSi3O10(F,OH)2 15.0 0.06
been addressed elsewhere (Zanin et al., 2008). Actinolite Ca2(Mg, Fe)5Si8O22(OH)2 0.1 8.2
In the current study, investigation was carried out at Kennecott Andradite (Al) Ca3Al2(SiO4)3 0.14 58.7
Utah Copperton Concentrator. Plant surveys have been undertaken on Talc Mg3Si4O10(OH)2 0.01 0.08
otation feeds having different mineralogy, and ancillary laboratory Calcite CaCO3 1.0 4.9

otation tests have been carried out plant slurries. Tests on controlled Modal mineral distribution for the major gangue minerals determined by QEM-Scan.
 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266
Table 3
Composition of synthetic process water compared with a sample of process water from
the KUCC copper circuit.
Ion Synthetic H2O Process H2O
[ppm] [ppm]
Na+ 1308 1310
Ca++ 798 798
K+ 87 87
Mg++ 129 129
Cl 1818 1940
SO
4 2653 2590
HCO 189 160
3
TDS 6980 7010
pH 7.6 7.2
Cond. [ S/cm] 6700 8000
salts in demineralised water to reproduce the composition of
Kennecott process water (Table 3).
2.3. Methods
2.3.1. Plant surveys
The Copperton concentrator consists of grinding circuits (SAG
milling followed by ball milling) and two otation circuits: a bulk
copper otation circuit, where copper/molybdenum concentrate is
produced, and a molybdenite otation plant, in which molybdenite in
the bulk concentrate is separated from the copper minerals. At the
time of investigation, the copper circuit consisted of 5 parallel rows of
rougher/scavenger cells. The rougher concentrate was treated in
rougher cleaners (2 parallel rows of cells) without regrinding. The
scavenger concentrate reported to regrinding (ball milling) along
with the rougher cleaner tailing. The regrind circuit product was
upgraded in three scavenger cleaner stages, the nal concentrate from
which, combined with the rougher cleaner concentrate, produced the
bulk copper/molybdenum concentrate. The target d80 of the otation
feed was 200 m, and the pH in the rougher/scavengers was
controlled between 9.5 and 10. A bulk concentrate assaying about
2530% Cu and 24% Mo was produced, depending on the ore blend
processed. Since most of the molybdenite losses occurred in the
rougher/scavengers of the copper circuit, this part of the circuit
became the focus of investigations.
Four plant surveys were undertaken, in which timed lip samples
and in-pulp samples were collected from each cell down one of the
rougher/scavenger rows. In each survey, four sampling rounds were
performed, over a period of 90 min, in which multiple samples were
collected from the rougher/scavenger tailings to reduce experimental
error. Prior to sampling, plant stability was ensured from the control
room, and plant data (throughput, feed solids percent, air ows, and
lip levels) were recorded. All metallurgical samples were initially
assayed unsized. Samples from selected surveys were sized by wet/
dry sieving and assayed on a size-by-size basis. Data reconciliation
and mass balance were performed using the software Bilmat 9.3.
Surveys 1 and 2 were performed with the plant processing a blend
with high quartzite ore content, while surveys 3 and 4 were carried
out with the plant processing a blend with high limestone skarn ore
content. Surveys 1 and 4 were performed at lower solids percent in
the otation feed, while surveys 2 and 3 were carried out at higher
solids percent (Table 4). The percent solids, which under standard
operating conditions ranged between 30% and 35%, was adjusted by
varying the water owrate to the rougher otation distribution box.
The inuence of feed solids percent on the recovery of molybdenite
from the two different ore blends was studied.
During survey 1, additional samples of slurry from the concentrate
of cells 1 and 11 were collected for surface analysis by ToF-SIMS.
Samples were placed in plastic vials, purged with nitrogen to remove
oxygen and frozen in liquid nitrogen. The samples were stored frozen
259
in a cryogenic container until ToF-SIMS analysis was carried out. Since
the focus of the study was recovery of coarse molybdenite, the ToF-
SIMS analysis was carried out on the +150 m particles, which were
isolated by wet sieving. The surface chemistry of the fast (concentrate
from cell 1) and slow (concentrate from cell 11) oating particles was
investigated, in order to correlate possible differences in surface
composition to otation response. It was not possible to analyse
molybdenite in the otation tailings, which ideally contain the non-
oating molybdenite, due to the extremely low molybdenum grade,
i.e. a statistically signicant number of molybdenite particles could
not be analysed. No samples of slurry for ToF-SIMS analysis were
collected during surveys 34. Therefore, it was not possible to
compare the surface composition of molybdenite in surveys with
different feed ore blends and gangue mineralogy. This important
aspect of the study was, however, conducted on samples generated at
laboratory scale, as described further below.
2.3.2. Laboratory otation tests on plant slurries
In parallel with the surveys, laboratory otation tests were
undertaken on conditioned rougher feed slurries collected in the
plant. The tests were conducted using a 5 l Agitair otation machine
with forced air supply. The samples were taken from the otation feed
box where all reagents except frother were added. An impeller speed
of 1000 rpm and air owrate of 5 l/min were used in the tests. The
slurry (conditioned rougher feed) was oated without further
collector addition, and at the same pH (9.510) as the plant.
Concentrates were collected after 1, 3, 6 and 10 min. All the otation
products were assayed on an unsized and size-by-size basis. Unsized
and size-by-size recoveries were calculated.
2.3.3. Laboratory otation tests on reconstructed feeds
In diagnostic tests carried out at the Ian Wark Research Institute,
samples of quartzite and limestone skarn ores provided by Kennecott
were used to reproduce plant slurries. Three different controlled feed
blends were prepared: 100% quartzite, 75:25 quartzite/skarn and
50:50 quartzite/ skarn ores. The ore blends were crushed using
laboratory jaw and cone crushers, homogenised and ground in a
laboratory Galigher tumbling mill, using stainless steel rods as
grinding media. Grinding was calibrated to a target d80 = 200 m, by
varying the grind time for each feed type. Synthetic process water
(Table 3) was used to simulate plant conditions. The same pH and
reagents scheme were used as in the tests carried out at Kennecott.
Flotation tests were carried out as described in 2.3.2.
Similarly to the plant surveys, after each laboratory otation test
the coarse particles (+150 m) from the rst concentrate and the last
concentrate were collected by wet screening and analysed by ToF-
SIMS. Surface analysis was conducted for the concentrates collected in
otation tests on both 100% quartzite ore and 50:50 blend of quartzite
and skarn ores, to draw a link between molybdenite surface
composition, gangue mineralogy and otation response .
2.3.4. Laboratory otation tests on coarse molybdenite particles
The effect of calcium and magnesium ions in solution and gangue
minerals on the recovery of molybdenite was studied separately.
Coarse molybdenite particles (+150 m) were obtained by dry
screening of a molybdenite concentrate from Kennecott (nal product
Table 4
Feed composition and operating conditions during the plant surveys.
Ore blend Cu Mo Survey Solids pH
type
[%] [%] # [%]
High quartzite 0.3 0.06 1 27 9.8
2 35 10.0
High skarn 0.5 0.06 3 35 10.1
4 27 10.1
260 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266
of the molybdenum plant). Particles were washed in Ensolv (n-propyl
bromide N 93%) and ethanol to clean the surface from any residual
collector. Samples of quartzite and skarn ore were oated to remove
copper minerals and molybdenite, and the otation tailings were
collected. The coarse molybdenite particles were then blended into
the tailings of quartzite and skarn ores, to produce synthetic ore
containing 0.15% MoS2 by weight. Samples were oated, with the
collectors and frother added according to the conditioning scheme
described in 2.2. Low conductivity water, produced by reverse
osmosis, two stages of ion exchange and two stages of activated
carbon prior to nal ltration, was used in these experiments. The
water pH was adjusted to 10 by adding KOH, and the concentration of
ions was varied by adding Ca(NO3)2 and Mg(NO3)2.
2.4. Characterisation techniques
2.4.1. Contact angle of molybdenite particles
Large molybdenite crystals were obtained from WILLYAMA-GEO
Discoveries. Molybdenite crystals were cut and cleaved using a
scalpel, to expose fresh crystal edges and faces. The contact angle was
measured at different pH values and Ca2+ concentrations in the
ranges typically observed in the plant (811 for pH and 010 2 M for
Ca2+ ions). After conditioning, the water advancing and receding
contact angle of face and edge of individual molybdenite particles
were measured by the sessile drop method.
2.4.2. ToF-SIMS
TOF-SIMS spectra were obtained using a PHI TRIFT II System
equipped with a gallium liquid metal ion gun (LMIG) in pulsed mode.
For insulating samples the surface charge is compensated using a
pulsed electron ood gun. The mineral samples were mounted on to
indium foil. The primary beam current employed in the present study
was 600 pA (DC measurement). In static mode, the analysis is
conned to the top two monolayers. An excitation voltage of 25 kV
was used in un-bunched mode to give a spatial resolution of better
than 0.5 m and pulse length adjusted to give a mass resolution (m/
m) of ~ 4000. Imaging of the sample involved mapping for positive
and negative ions for surface regions of mineral particles of interest,
i.e. MoS2. Several tens of particles in each processing stream are
analysed and the data statistically presented in terms of normalised
(to total ion yield) intensity of signals and 95% condence intervals. In
this mode, statistical differences in surface chemistry between the
same minerals in different streams or under different pulp conditions
may be discerned. In the present case, we are investigating surface
chemistry differences between oating and non(slow)-oating MoS2
particles.
3. Results
3.1. Plant Studies and Laboratory Tests on Plant Slurries
The grade/recovery relationship for copper and molybdenum for
the four plant surveys and the laboratory otation tests carried out in
parallel with the plant surveys are reported in Fig. 3. The ultimate
recovery (combined rougher/scavenger otation) for copper and
molybdenum is also reported in Table 5. In the plant surveys,
molybdenum recovery ranged from 75% to 92%, while copper
recovery from 85% to 92%. The recovery of molybdenum was
consistently lower than copper recovery, the difference increasing at
higher feed solids percent and in the presence of limestone skarn ore.
For both copper and molybdenum, the lowest recovery was achieved
at high solids percent and in the presence of limestone skarn ore
(survey 3). Compared to copper, molybdenum recovery was more
sensitive to the feed solids percent, being signicantly enhanced
when solids percent was reduced from 35% to 27%. Both copper and
molybdenum recoveries were lower in the presence of limestone
skarn ore in the feed blend, but molybdenum recovery decreased
more than copper recovery. With a reduction in the feed solids
percent, however, it was possible to partially restore molybdenum
recovery to values characteristic of quartzite ore (from 75% to 85%).
The laboratory tests (Fig. 3c and d) showed generally higher
recoveries for both copper and molybdenum compared to the plant
surveys (Fig. 3a and b), but with a similar trend with respect to feed
composition and solids percent. Copper recovery was marginally
affected by changes in feed solids percent in the absence of limestone
skarn ore, showing some difference (90% recovery at 35% solids versus
93% recovery at 27% solids) when limestone skarn ore was blended to
otation feed. Molybdenum recovery, on the contrary, was always
higher at low solids percent. Differences in recovery became
signicant (93% versus 85%) in the presence of limestone skarn ore.
Size-by-size analysis of the otation products was undertaken for
surveys 3 and 4 (feed blend with high skarn ore), in which a
signicant increase in molybdenum recovery was obtained by
reducing the feed solids percent. Results (Fig. 4a) revealed increased
molybdenum recovery across all size ranges, but particularly for the
coarse size fractions (+150 m). The effect on copper was much less
pronounced. In the laboratory scale tests, the recovery of molybde-
num was higher than in the plant across all size ranges. Furthermore, a
reduction in the feed solids percent produced a signicant effect on
coarse (+150 m) molybdenum bearing particles, for which recovery
increased from 60% to 80% when the feed solids percent was reduced
from 35% to 27%.
Two additional otation tests were undertaken at laboratory scale,
at 30% and 45% solids in the rougher otation feed. These concentra-
tions were achieved by manipulating samples of plant feed collected
during survey 3 (35% solids), the former by diluting with process
water, the latter by ltering and re-suspending the solids in process
water. The trend (Fig. 4d) conrmed what was observed in the
previous tests, showing high sensitivity of the coarse molybdenite
particles to feed solids percent. The feed solids percent is an important
driver to molybdenite recovery overall, and particularly in coarse
particle size fractions.
In Fig. 4, bar charts reporting the distribution of copper and
molybdenum in the feed are superimposed on the recovery curves.
The area of each bar in the charts is proportional to the relative mass
of mineral in the specied particle size range. Considering that, at the
standard plant grind, the +150 m size fraction contains about 20% of
the total molybdenite in the feed, the recovery of this size fraction is
critical to the overall molybdenite recovery.
Surface analysis results on the slurry samples collected during
survey 1 (in which the otation feed consisted mainly of quartzite
type ore, and no limestone skarn was present) are reported in Fig. 5.
No signicant difference in surface composition between fast
(concentrate from Cell 1) and slow (concentrate from Cell 11)
oating molybdenites was found in this particular case of the
quartzite ore type. It could therefore be concluded that the smaller
differences in molybdenite recovery noted between plant and
laboratory scale (Fig. 3) may be ascribed to changes in hydrodynam-
ics. However, this conclusion will depend on the feed type, as
discussed further below.
3.2. Laboratory otation tests on reconstructed feeds
The otation recoveries of copper and molybdenum in laboratory
scale tests on blends of quartzite ore and skarn ore are reported in
Fig. 6. A sharp decrease in molybdenite recovery was observed at
increasing concentration of limestone skarn ore in the feed. The
ultimate molybdenite recovery decreased from 92% in the absence of
skarn ore to 85% with 25% skarn ore and 56% with 50% skarn ore in the
otation feed. Copper otation was affected to a much lesser extent.
This is in agreement with the plant surveys and tests carried out on
plant slurries. It should be noted that, since the limestone skarn ore
 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266 261

Fig. 3. Grade/recovery relationship for copper and molybdenum in the plant surveys (a and b) and in the laboratory tests on plant feed (c and d). The ultimate recovery at the
completion of otation is also indicated (dashed lines).

contains very little molybdenum (Table 1), it is the molybdenite observations at plant scale, which also showed no surface chemical
particles in the quartzite ore which are depressed in the presence of differences on molybdenite. In the presence of 50% skarn ore, on the
skarn ore. contrary, samples of fast and slow oating molybdenites showed
The ToF-SIMS spectra for the fast oating molybdenite (Con 1) and differences in surface chemistry (Fig. 7). In the fast oating sample
slow oating molybdenite (Con 4) collected in the tests with 0% and (Con 1) there are much more exposed molybdenum and sulphur
50% skarn ore in the otation feed are reported in Fig. 7. compared to the slow oating sample (Con 4). Furthermore, Ca, K, Fe,
Samples of fast (Con 1) and slow (Con 4) oating molybdenite as well as O and OH groups, were more abundant on the latter sample.
in the presence of quartzite type ore only, showed very small Signicantly higher Mg on the surface of molybdenite in the presence
differences in surface chemical composition. The different species of skarn ore was also noted in both fast and slow oating fractions
appear in the same proportion on the surface of the two samples, (Fig. 7a).
within statistical error (Fig. 7). This is in agreement with previous
3.3. Laboratory otation tests on coarse molybdenite particles
Table 5
Ultimate recovery of copper and molybdenum in plant (after rougher/scavenger The effect of calcium and magnesium ions in solution on coarse
otation) and laboratory. molybdenite (+150 m) recovery from quartzite and limestone skarn
ores is presented in Fig. 8. In the case of quartzite ore, the cumulative
Ore blend Survey Solids Plant recovery Lab recovery
type
recovery of molybdenite was high, approaching 90% after 8 min of
# [%] Cu [%] Mo [%] Cu [%] Mo [%] otation. The addition of calcium and magnesium ions to the pulp did
High quartzite 1 27 92 92 95 95 not have an effect on molybdenite otation. In contrast, molybdenite
2 35 92 88 94 92 recovery from limestone skarn ore was considerably lower (60%) even
High skarn 3 35 85 75 90 85
in the absence of calcium and magnesium ions, and decreased further
4 27 88 85 93 93
to 30% when Ca2+ and Mg2+ ions were added. Apparently, there is a
262 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266

Fig. 4. Size-by-size recovery of copper and molybdenum in plant surveys 3 and 4 (after rougher/scavenger otation) (a and b) and in the laboratory otation tests on plant feed
(c and d). At laboratory scale, additional feed solids percent values (30% and 45%) were obtained by manipulating plant feed samples. The distribution of copper and molybdenum in
the feed is also reported. Feed blend contained high limestone skarn ore.

synergistic effect between the presence in solution of Ca2+ and Mg2+
ions and gangue minerals of the skarn ore.
The effect of pH and calcium ions on the contact angle of
molybdenite was also investigated (Fig. 9). The tests showed that
faces and edges of molybdenite particles have considerably different
contact angles. Face contact angle was high (about 100) and
independent of solution pH and calcium concentration. In contrast,
contact angle on the edges was low (maximum 45), and decreased with
increased pH and calcium ion concentration ([Ca2+] N 4 10 3 M).
4. Discussion
From an industrial perspective, the most important nding from
the plant studies was the consistent increase in molybdenite recovery
in rougher/scavenger otation at reduced feed solids percent. This is
in agreement with previous observations for different otation plants
(Ametov et al., 2008). The coarse (+ 150 m) size fractions were
affected the most by changes in feed solids percent, and therefore
uctuations in the overall recovery of molybdenite in the plant are
determined to a large extent by the otation response of the coarse
particles. Particle size and shape, morphology and surface composi-
tion are all contributing factors. Insufcient molybdenite liberation in
the coarse size fractions was also considered in the rst instance as a
possible cause for the low otation recovery. However, QEM-Scan
analysis of the scavenger tailings (Triffett, private communication)
showed that molybdenite in the + 150 m size fraction mainly
consists of liberated particles. This conrms that other mechanisms,
as outlined in Section 1.1, play signicant roles.
The hydrodynamic behaviour of the at molybdenite particles in a
turbulent environment may explain the increase in molybdenum
recovery observed at reduced solids percent (Fig. 3). Higher local
turbulent energy dissipation may increase the collision efciency for
the at and elongated particles, as hypothesised in Section 1.2. The
fact that molybdenite recovery in laboratory otation tests was higher
than in the plant (Fig. 4) may also be due hydrodynamic effects.
Compared to the plant cells, laboratory batch otation cells have
higher local turbulent energy dissipation (Newell and Grano, 2007),
and therefore higher collision efciency, Ec, and, according to Eq. (1),
higher otation rate. An interesting feature is that it is the coarse and
liberated molybdenite that is mainly affected, the recovery of which is
 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266 263

Fig. 5. Positive (a) and negative (b) ToF-SIMS normalised intensities for + 150 m molybdenite particles. Average with 95% condence, N N 23. Fast oating (Cell 1) and slow oating
(Cell 11) molybdenite from plant survey 1 (no skarn ore in the feed).

Fig. 6. Copper and molybdenum recovery in laboratory otation tests on reconstructed feeds (blends of quartzite and skarn ores in different ratios). Tests at 35% solids (synthetic
process water used); Agitair 5 l cell; impeller speed 1000 rpm; air 7 l/min. 0% skarn ore = 100% quartzite ore.
264 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266

Fig. 7. Positive (a) and negative (b) ToF-SIMS normalised intensities for on + 150 m molybdenite particles. Average with 95% condence, N N 23. Fast oating (Con 1) and slow
oating (Con 4) coarse particles (N 150 m) in tests with 100% quartzite ore and 50:50 blend of quartzite and skarn ores.

presumably limited by stability efciency, Es (Pyke et al., 2003). The
high surface/bulk ratio of the coarse molybdenite particles may
however be benecial for the stability of particlebubble aggregates,
because of a higher surface force/inertial force ratio. Surface forces
favour attachment of particles to bubbles, while inertial forces
promote detachment. A high surface/bulk ratio results in high
stability efciency, even in high turbulent conditions such as in
laboratory otation and at reduced solids percent.
Another outcome of the plant studies was that the effect of a
reduction in feed solids percent on molybdenite recovery was high for
some feed blends (high limestone skarn) while it was much lower for
others (high quartzite). This was observed at plant scale (Fig. 3b), in
laboratory otation tests on plant slurries (Fig. 3d), and in tests on
reconstructed feed blends (Fig. 6). Therefore, it is likely that factors
other than hydrodynamics and collision efciency, such as surface
modication and particleparticle interactions also play a role. ToF-
SIMS analysis indicated the presence of elements associated with
hydrophilic components on the slow oating coarse molybdenite
particles in the presence of skarn ore (Fig. 7). This could be either from
the deposition of ne slimes or adsorption of dissolved ions. In any
case, the effect is strongly ore type dependent. In the absence of skarn
ore, no difference in surface composition between fast and slow
molybdenite was apparent at plant (Fig. 5) and laboratory (Fig. 7)
scales, suggesting that the different otation recoveries of molybde-
nite at high and low solids percent may be ascribed to hydrodynamic
effects, possibly arising from the intrinsic shape of the particles, rather
than from the surface cleaning of slime coatings. Therefore, the
hypothesis of slime coatings may hold for some feed types and be less
important for others. It is surmised that at low pulp density there is
less likelihood of inter-particle interactions causing slime coatings. A
reduction in slime coatings on the surface of molybdenite particles
results in increased contact angle. Even a small increase in contact
angle could cause previously non-oatable coarse particles to become
oatable and the ultimate recovery is therefore increased.
The high concentration of Ca2+ and SO2 4 in Kennecott process
water (Table 3) may also suggest deposition of calcite and gypsum
from solution due to oversaturation. This could also be a contributing
factor, but it is not sufcient to justify the different degrees of
molybdenite depression observed in the presence of quartzite and
skarn ores. Diagnostic otation tests on coarse molybdenite particles
in the presence of limestone skarn ore and different levels of calcium
and magnesium ions in solution (Fig. 8) corroborate the hypothesis
that interactions between molybdenite and some gangue minerals in
the skarn ore lead to molybdenite depression. The effect could be
induced by metal ions adsorbed at the molybdenite edges, as also
suggested by contact angle measurements carried out separately on
 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266
Fig. 8. Recovery of coarse molybdenite particles (+ 150 m) in otation tests on
synthetic feeds. Molybdenite particles from the nal concentrate, after surface cleaning,
were blended with ground quartzite and skarn ore (0.15% MoS2 by weight). Tests were
in demineralised water, at pH 10, in the presence and absence of Ca2+ and Mg2+ ions.
faces and edges (Fig. 9). The edges of molybdenite particles are
particularly reactive, and adsorption of Ca2+ and/or Mg2+ may have
the combined effect of reducing the overall hydrophobicity of
molybdenite particles and bridging for particleparticle interactions
with the negatively charged gangue minerals. This is also in
agreement with the ndings of Raghavan and Hsu (1984) on
molybdenite/silica mineral systems. Slime coatings are possibly
causing molybdenite depression in the presence of skarn ore. Minerals
which are much more abundant in the limestone skarn ore are:
actinolite, andradite, talc and calcite (Table 2). These minerals are
potentially responsible for the high Mg, Ca and Fe observed on the
surface of molybdenite when oated in the presence of limestone
skarn ore (Fig. 7). This conclusion would be consistent with the fact
that the presence of the above mentioned minerals in the otation
feed has been correlated to periods of low molybdenite recovery
(Triffett et al., 2008).
Fig. 9. Contact angle (advancing) of coarse molybdenite particles (+150 m) measured
in proximity of face and edges, after conditioning at different pH values and Ca2+
concentrations. Contact angle measured by means of the sessile drop method.
265
5. Conclusions
The experimental results conrm that the otation of molybdenite
from porphyry copper ores is more sensitive to the operating
environment than that of the copper sulphide minerals. The recovery
of molybdenum in the rougher/scavengers of the bulk Cu/Mo
otation circuit at Kennecott Utah Copper is highly variable, and
always lower than the recovery of copper. Particularly low molybde-
num recovery was observed in the presence of limestone skarn ore.
As a general rule, it was found that operating the rougher/scavenger
otation rows at a lower feed solids percent (27% against 35%)
ensures higher and more stable molybdenite recovery.
Several factors have been identied which could affect molybde-
nite otation, all of them related to the peculiar properties of
molybdenite. Preferential cleavage along the weakly bound SS layers
may impart to the molybdenite particles singular hydrodynamic
behaviour (low collision efciency due to the at and elongated
particle shape), and anisotropy in hydrophobicity (hydrophobic faces
and hydrophilic, highly reactive, edges). It has been shown that edges
have much lower contact angle compared to faces, in particular at
high pH (pH N 10) and high alkaline conditions ([Ca2+] N 10 2 M), as
in typical otation environment.
When limestone skarn ore was blended to the feed in laboratory
otation tests, molybdenite recovery decreased signicantly, similarly
to what was observed in the plant. Higher concentrations of Ca, Fe, Mg
and K were measured on the slow oating molybdenite particles
compared to the fast oating, which have been correlated to the
presence of some gangue minerals typical of the skarn ore. A bridging
effect of calcium and magnesium adsorbed on the surface of
molybdenite and some ne gangue particles in the slurry is possible
and may lead to the formation of slime coatings, thus reducing the
otation recovery of molybdenite.
In terms of strategies to increase molybdenite recovery at plant
scale, operating the rougher/scavengers at low solid percent was a
solution which gave immediate benets. This benet may be a result
of a combination of hydrodynamic effects (higher local turbulent
energy dissipation and collision efciency), rheology and reduced
formation of slime coatings. It also appears that preventing slimes
from adsorbing on molybdenite particles during otation may benet
molybdenite recovery. This is the subject for future investigation.
Acknowledgements
The authors would like to acknowledge the nancial support of the
sponsors of the AMIRA P260E project and the Australian Research
Council (ARC). Support from the Management at Kennecott Utah
Copper and site personnel is also gratefully acknowledged.
References
Ametov, I., Grano, S.R., Zanin, M., Gredelj, S., Magnuson, R., Bolles, T., Triffett, B., 2008. In:
Zuo, W.D., Yao, S.C., Liang, W.F., Cheng, Z.L., Long, H. (Eds.), Copper and
molybdenite recovery in plant and batch laboratory cells in porphyry copper
rougher otation: Proceedings of XXIV International Mineral Processing Congress,
Beijing, China 2428 September 2008, volume 1, pp. 11291137.
Chander, S., Fuerstenau, D.W., 1972. On the natural otability of molybdenite. Trans.
Am. Inst. Min. Metall. Engrs. 252, 6269.
Crozier, R.D., 1979. Flotation reagent practice in primary and by-product molybdenite
recovery. Min. Mag. 140, 174178.
Dippenaar, A., 1982. The destabilisation of froth by solids. I. The mechanism of lm
rupture. Int. J. Miner. Process. 9, 114.
Healy, T.W., 1984. Pulp chemistry, surface chemistry, and otation. Symp. Ser.
Australas. Inst. Min. Metall. 40, 4356.
Hemmings, C.E., 1981. On the signicance of otation froth liquid lamella thickness.
Trans. Inst. Min. Metall. Sect. C 90, C96C102.
Hernlund, R.W., 1961. Extraction of molybdenite from copper otation products. Q. J.
Colo. Sch. Min. 56, 177196.
Hoover, M.R., 1980. Water chemistry effects in the otation of sulphide ores a review
and discussion for molybdenite. In: Jones, M.J. (Ed.), Complex Sulphide Ores. IMM,
London, pp. 100112.
266 M. Zanin et al. / Int. J. Miner. Process. 93 (2009) 256266
Kelebek, S., 1988. Critical surface tension of wetting and of oatability of molybdenite
and sulphur. J. Colloid Interface Sci. 124, 504514.
Lince, J.R., Frantz, P., 2000. Anisotropic oxidation of MoS2 crystallites studied by angle-
resolved X-ray photoelectron spectroscopy. Tribol. Lett. 9, 211218.
Newell, R., Grano, S., 2007. Hydrodynamics and scale-up in rushton turbine otation
cells: Part 1 cell hydrodynamics. Int. J. Miner. Process. 81, 224236.
Orwe, D., Grano, S.R., Lauder, D.W., 1998. Increasing ne copper recovery at the Ok Tedi
concentrator, Papua New Guinea. Miner. Eng. 11, 171187.
Podobnik, D.M., Shirley, J.F., 1982. Molybdenite recovery at Cuajone. Min. Eng.
14731477 October.
Pyke, B., Fornasiero, D., Ralston, J., 2003. Bubble particle heterocoagulation under
turbulent conditions. J. Colloid Interface Sci. 265, 141151.
Raghavan, S., Hsu, L.L., 1984. Factors affecting the otation recovery of molybdenite
from porphyry copper ore. Int. J. Miner. Process. 12, 145162.
Savassi, O.N., Alexander, D.J., Johnson, N.W., Manlapig, E.V., Franzidis, J.-P., 1997.
Measurement of froth recovery of attached particles in industrial cells. In: Lauder,
D. (Ed.), Proceedings of the Sixth Mill Operators Conference, Aust. Inst. Min. Metall.
Publ., Melbourne, pp. 149155.
Schubert, H., 1999. On the turbulence-controlled microprocesses in otation machines.
Int. J. Miner. Process. 56, 257276.
Shirley, J.F., 1981. Byproduct molybdenite plant design. Can. Min. J. 1981, 2728 March.
Sutulov, A., 1975. Copper Porphyries. Miller Freeman Publications, San Francisco.
Triffett, B., Bradshaw, D., 2008. The role of morphology and host rock lithology on the
otation behaviour of molybdenite at Kennecott Utah Copper. AusIMM Publication
Series, 9th International Congress for Applied Mineralogy, ICAM 2008
Proceedings, 2008, pp. 465473.
Triffett, B., Veloo, C., Adair, B.J.I., Bradshaw, D., 2008. An investigation on the factors
affecting the recovery of molybdenite in the Kennecott Utah Copper bulk otation
circuit. Miner. Eng. 21, 832840.
Wie, J.M., Fuerstenau, D.W., 1974. Effect of dextrin on surface properties and the
otation of molybdenite. Int. J. Miner. Process. 1, 1732.
Zanin, M., Gredelj, S., Grano, S.R., 2008. Factors affecting froth stability in mineral otation
and implications on minerals recovery: a case study. In: Kuyvenhoven, R., Gomez, C.,
Casali, A. (Eds.), Proceedings of Procemin 2008, Santiago (Chile), pp. 197206.
Zanin, M., Wightman, E., Grano, S.R., Franzidis, J.-P., 2009. Quantifying contributions to
froth stability in porphyry copper plants. Int. J. Miner. Process. 91, 1927.