The critical importance of the grinding environment on ne particle recovery

in otation
Stephen Grano
Ian Wark Research Institute, The ARC Special Research Centre for Particle and Material Interfaces, University of South Australia, Mawson Lakes, Adelaide, SA 5095, Australia
a r t i c l e i n f o
Article history:
Received 10 April 2008
Accepted 28 October 2008
Available online 20 December 2008
Keywords:
Grinding environment
Grinding media
Fine particle otation
a b s t r a c t
This paper examines published and new experimental evidence on the effect of the grinding environment
on ne (10 lm) value mineral recovery in otation. Reasons for increases in ne value mineral recovery
from ores with fully electrochemically inert grinding media are discussed in relation to reduced surface
contamination by iron hydroxide emanating from the grinding media. The application of stirred milling
technology, which allows the use of fully inert grinding media, to primary grinding applications may lead
to increased ne value mineral recovery in otation rougher applications. It is suggested that the effect of
the grinding media, which is important for ne particles and progressively becomes more important as
the grind size becomes ner, is principally due to the abrasion mechanism of the minerals with the grind-
ing media in the production of ne particles. Opportunities for research and industry application are
discussed.
2008 Published by Elsevier Ltd.
1. Introduction
The importance of the grinding environment on the otation
behaviour of sulphide minerals and ores has been known for many
years (Rey and Formanek, 1960). However, it is not commonly
appreciated that the grinding environment may play a key role in
controlling ne value mineral oatability, as well as selectivity
against sulphide gangue mineral components. More recently, the
operating benets of grinding with fully inert ceramic media at
plant scale has been described as (i) increased recovery of ne va-
lue mineral particles, (ii) improved selectivity against sulphide
gangue mineral components, and (iii) reduced collector consump-
tion, amongst other benets (Pease et al., 2006a). Another author
has stated that the otation recovery of liberated valuable miner-
als in the 010 lm fraction, would be affected detrimentally to a
greater extent than the intermediate fraction by the presence of
the iron hydroxides derived from mild steel grinding media (John-
son, 2006). The development of stirred milling technology, utilising
fully inert grinding media, effectively avoids contamination of min-
eral surfaces from grinding media sources (Pease et al., 2006b). This
paper discusses the collective evidence of these benets, with a
specic focus on ne value mineral recovery, from both the pub-
lished literature and in a specic experimental study. Attention
will turn rstly to a brief review of the published literature in this
eld. During grinding a number of chemical mechanisms may
operate:
(a) Eh effect The reducing environment when grinding with
reactive steel media decreases the Eh and dissolved oxygen
of the slurry immediately after grinding (Grano et al.,
1990). Since thiol collector adsorption is strongly Eh depen-
dent this may reduce oatability if an aeration stage is not
provided before otation. In the publications and experi-
mental study discussed further below, the effect of Eh is con-
sidered, and efforts undertaken to minimise the effect of Eh
through its control after grinding and before collector
conditioning.
(b) Iron hydroxide coatings on surfaces Even with aeration and
restoration in the Eh after grinding there may be continuing
depression of value mineral oatability due to the presence
of iron hydroxide generated from oxidation of electrochem-
ically reactive steel media (Grano et al., 1990). As discussed
further below, this is the main contributing factor to reduced
ne particle oatability after grinding with electrochemi-
cally reactive steel media types.
(c) Oxygen reduction This can occur on some sulphide minerals
and promotes the formation of metal hydroxides.
(d) Precipitation from solution During the process of size reduc-
tion in milling, there is an increase in surface area and often
an increase in pulp temperature, factors which may promote
precipitation from saturated solutions. An example is the
surface precipitation of gypsum from calcium sulphate satu-
rated solution during secondary grinding in the lead/zinc
concentrator of Mount Isa Mines Limited (Cullinan, 1999).
This has implications for the water composition which is
used in the milling process.
0892-6875/$ - see front matter 2008 Published by Elsevier Ltd.
doi:10.1016/j.mineng.2008.10.008
E-mail address: stephen.grano@unisa.edu.au
Minerals Engineering 22 (2009) 386394
Contents lists available at ScienceDirect
Minerals Engineering
j our nal homepage: www. el sevi er . com/ l ocat e/ mi neng
(e) Galvanic coupling Galvanic coupling of the media with sul-
phide minerals can greatly accelerate oxidation of the media
and oxygen reduction on the sulphide minerals (Huang and
Grano, 2006a). Galvanic coupling and oxygen reduction may
also be inuenced by the conductivity of the water used in
milling, with electrochemically reactive steel media showing
greater levels of oxidation with high ionic strength water.
In the study by Grano et al. (1990), otation of a chalcopyrite
ore was conducted as a function of conditioning and otation Eh,
and after grinding in two different chemical environments
(Fig. 1). This work showed that there was increased chalcopyrite
recovery at the same conditioning and otation Eh values, over
the Eh range of 0400 mV, for the case of grinding in a ceramic mill
and with ceramic media (Grano et al., 1990). In this particular
study, as the conditioning and otation Eh was controlled to the
same values, this allowed the effect of the grinding environment
to be evaluated independently of the conditioning and otation
Eh (Grano et al., 1990).
Attention will turn now to the published evidence of how ne
particle oatability may be affected by the grinding environment.
In a study on a lead/zinc ore at laboratory scale, grinding with high
chromium alloy steel media increased ne (5 lm) galena maxi-
mum recovery compared with grinding with high carbon steel
(mild steel) media which is more electrochemically active (Cull-
inan et al., 1999) (Fig. 2a). The increased maximum recovery of ne
galena was associated with reduced surface contamination by iron
hydroxide, emanating from the high carbon steel media (Cullinan
et al., 1999). The otation rate constant dependency on particle size
followed the expected pattern with the otation rate constant
decreasing for particle sizes from 30 to 3 lm (Fig. 2b). There was
an increase in the otation rate constant for ultra-ne galena par-
ticles less than 3 lm, for both grinding environments, but this in-
crease was greater after grinding with high chromium alloy steel
media (Cullinan et al., 1999). Importantly, increasing collector
additions above 80 g/t did not increase ne galena maximum
recovery but, instead, increased sphalerite recovery in intermedi-
ate size fractions, reducing the separation efciency of galena
against sphalerite (Cullinan et al., 1999). This bears out the dif-
culty generally found of separating ne particles of value minerals
(10 lm) with low oatability from other sulphide minerals in
intermediate size fractions (2040 lm) (Johnson, 2006). Evidently,
the surfaces of the ne galena particles after grinding with high
carbon steel media were less conducive to collector adsorption
and hydrophobisation. In a plant context, this may lead to in-
creased collector additions and lower selectivity against sphalerite
with mild steel media (Pease et al., 2006b).
In a separate study on a copper ore at plant scale, grinding in a
fully autogenous plant mill gave higher ne (8 lm) chalcopyrite
recovery than grinding in a conventional plant ball mill which used
mild steel media (Grano et al., 1994) (Fig. 3). For each mill type,
samples of the mill products were screened to remove coarse par-
ticles, allowing laboratory otation testing (Grano et al., 1994). Dif-
ferences in the amount of iron hydroxide generated in the two
mills gave different pyrite and chalcopyrite otation responses
with respect to cyanide addition due to complexation of the iron
with cyanide (Grano et al., 1994). Most importantly, grinding in
Eh /mV
R
e
c
o
v
e
r
y

/

%
v
Ceramic Media
Mild Steel Media
0
20
40
60
80
100
-500 -400 -300 -200 -100 0 200 300 400 100 500
Fig. 1. Chalcopyrite recovery from Mount Isa copper ore as a function of
conditioning and otation Eh after grinding in a ceramic media and mill at a grind
Eh of 325 mV, and in a mild steel media and mill at a grind Eh of 220 mV. The Eh
was controlled to different values after grinding by the addition of oxidants/
reductants. Sodium isobutyl xanthate collector 40 g/t addition; pH 8.18.3; grind
size 64% 74 lm, 2.6% Cu (adapted from Grano et al., 1990).
0
20
40
60
80
100
M
a
x
i
m
u
m

R
e
c
o
v
e
r
y
/

%
High Carbon Steel
High Chromium Steel
0
1
2
3
0.1 1 10 100
Particle Size / Microns
R
a
t
e

C
o
n
s
t
a
n
t

/
M
i
n
-
1
(b) Rate Constant
(a) Maximum Recovery
Fig. 2. (a) Maximum recovery and (b) otation rate constant of galena from Mount
Isa lead/zinc ore as a function of galena particle size after grinding with high carbon
steel and high chromium (26%) alloy steel. Sodium ethyl xanthate collector 80 g/t
addition; pH 7.88.2; grind size d
80
= 90 lm, 6% Pb and 9% Zn. The maximum
recovery and otation rate constant were calculated assuming a single oatable
species in the laboratory batch cell (adapted from Cullinan et al., 1999).
0
20
40
60
80
100
1 10 100 1000
Particle Size / Microns
R
e
c
o
v
e
r
y

/

%
Autogenous
Conventional
Fig. 3. Chalcopyrite recovery from Mount Isa copper ore as a function of particle
size after grinding in a fully autogenous mill and a conventional mill using mild
steel media. Sodium isobutyl xanthate collector 80 g/t addition; pH 8.28.3; grind
size d
80
= 130 lm, 3.2% Cu (adapted from Grano et al., 1994).
S. Grano / Minerals Engineering 22 (2009) 386394 387
the fully inert environment of the autogenous mill resulted in
higher ne chalcopyrite recovery due to less iron hydroxide pres-
ent in the system (Grano et al., 1994). Reasons why ne particles
of the value mineral are more adversely affected by iron hydroxide
emanating from grinding media sources are discussed further
below.
In another example, the otation recovery of single mineral
arsenopyrite was studied after grinding in the Magotteaux Mill

(Greet et al., 2004) and in a laboratory stirred bead mill using cera-
mic media (Huang and Grano, 2006b). The highest ne (10 lm)
arsenopyrite recovery was obtained for the case of grinding with
ceramic media (Fig. 4). Arsenopyrite oatability decreased with
an increasing amount of iron hydroxide in the mill discharge and
the degree of galvanic interaction of the arsenopyrite with grinding
media, evaluated for a range of different media types displaying
different electrochemical activity (Huang and Grano, 2006a). In
the case of ceramic media, and grinding in a stirred mill, the in-
creased arsenopyrite recovery noted was principally accounted
by the decreased surface contamination fromiron hydroxide rather
than any intrinsic benets of the stirred milling action (Huang and
Grano, 2006b). This point needs further examination.
Recently, advances in stirred milling technology have lead to
the possibility of grinding main stream feeds using fully inert
grinding media (Pease et al., 2006a). The important feature here
is to use large diameter and sufciently competent media to allow
size reduction of relatively coarse particles at plant scale (Gao et
al., 2000). In this current paper, large diameter ceramic media is
used in a tumbling mill to demonstrate the importance of fully
electrochemically inert grinding media on the otation recovery
of ne chalcopyrite and its selectivity against pyrite from a chalco-
pyrite containing ore. This will serve to further the evidence show-
ing howne particles may be affected by the grinding environment
and, more specically, the grinding media. Surface analysis using
X-ray photoelectron spectroscopy (XPS) is employed to quantify
iron hydroxide surface atomic concentration.
2. Experimental
The copper ore was obtained from Mount Isa Mines, Mount Isa.
The ore contained copper (3.6%) as chalcopyrite, pyrite (12.5%), and
non-sulphide gangue principally as quartz (61% SiO
2
, 2% Al
2
O
3
, 4%
CaO and 5% MgO). At a grind d
80
of 110 lm, the chalcopyrite is 90%
liberated (Johnson, 1987).
One kilogram of the crushed ore (1.7 mm) was ground in
either a closed steel lined tumbling ball mill (15 20 cm) with
mild steel media (12 cm), or in a closed fully ceramic tumbling
ball mill (21 19 cm) with ceramic media (12 cm) at 60%
solids by weight. Grinding in the mild steel ball mill with mild steel
media is referred to as mild steel media in the text and diagrams
below. Grinding in the ceramic ball mill with ceramic media is re-
ferred to as ceramic media in the text and diagrams below. No re-
agents were added to the grinding mills. The grinding time was
adjusted to achieve d
80
= 110 lm for both grinding environments
which was 15 min in the case of mild steel media and 140 min in
the case of ceramic media. The revolution speed was 40 rpm for
each grinding mill.
The pH was not intentionally controlled during grinding or in
otation. Due to the dolomite present in the ore, the pH was buf-
fered to approximately 8.2 (0.1), and was in this range under
the conditions studied. After grinding and transfer to the otation
cell an aeration stage was applied for 5 min to increase the Eh in
both cases to 230250 mV (SHE), avoiding effects related to Eh. So-
dium isobutyl xanthate (SIBX) collector was then added and condi-
tioned for a further 3 min with continuing aeration. Methyl
isobutyl carbinol as frother was added just prior to the commence-
ment of otation at an addition of 10 g/t for all tests.
Batch otation was carried out in a 3 dm
3
cell, operating at
800 rpm, and concentrates were collected at 0.5, 2.0 and 8.0 min
cumulative otation times. The otation data on an unsized basis
is presented as cumulative grade and recovery of copper and pyrite
at these otation times, assuming that all copper was present in
chalcopyrite and iron not resident in chalcopyrite was contained
in pyrite. The rst concentrate, the combined second and third con-
centrates, and tailings were sized using a combination of wet
screening at 38 lm, precycloning to obtain nominally 4 lm on
chalcopyrite, cyclosizing with recovery of the 48 lm fraction,
and dry screening to 150 lm (Johnson, 2006). The recovery of
the individual minerals is displayed as a function of the geometric
mean particle size in the gures. The supercial gas rate in ota-
tion was 0.2 cm s
1
. The estimated mean energy dissipation was
less than 1 W kg
1
of pulp. The sauter mean bubble diameter in
the collection zone of the otation cell was 0.8 mm.
XPS measurements on the aerated mill discharge samples (prior
to collector addition) were carried out with a Perkin Elmer PHI 560
spectrometer using Al Ka X-ray source. The initial surfaces were
rst examined in survey mode to identify all elements present
and then the various elemental regions scanned for chemical bond-
ing information. All surfaces were then etched to approximately
25 nm using an accelerated Ar
+
beam at 4 kV. The powdered nature
of the samples dictates that some proportion of the surface will not
be etched due to shadowing. Only atomic concentrations of the
major chemical species are reported. After charge correction, a
peak occurred at approximately 284.6 eV due to the presence of
graphitic carbon and unavoidable hydrocarbon contamination. A
C1s component also occurred at approximately 288290 eV corre-
sponding to carbonate from the dolomite contained in the ore.
Determining the ratios of these peaks allowed portioning of these
carbon types.
In general, a broad high intensity O1s emission near 532 eV
indicated the presence of several different oxygen containing spe-
cies (i.e., carbonate, silica, hydroxide and sulphate). The presence
of sulphur as either sulphide (161163 eV) or as sulphate
(168 eV) was indicated by S2p peaks and allowed partitioning
of these sulphur types. The presence of iron(III) as ferric hydrox-
ide/oxide or in chalcopyrite was indicated by a charge corrected
Fe 2p
3/2
peak at approximately 710 eV. Iron in pyrite occurs as a
shoulder to the above peak at approximately 707 eV. The presence
of copper(I) in chalcopyrite was indicated by a Cu 2p3/2 peak near
932 eV. Surface species were calculated assuming stoichiometric
0 10 20 30 40
M
a
x
i
m
u
m

R
e
c
o
v
e
r
y
/

%
0
25
50
75
Mild Steel
15% Cr
21% Cr
30% Cr
Ceramic
100
50
Particle Size / Microns
Fig. 4. Arsenopyrite recovery as a function of particle size at pH 7 after grinding in
the Magotteaux Mill

with mild steel, 15%, 21% and 30% chromium alloy steel
media, and after grinding in a stirred laboratory mill with ceramic media. Single
mineral arsenopyrite. Sodium isopropyl xanthate collector 800 g/t addition; pH 7,
grind size d
50
= 17 lm (adapted from Huang and Grano, 2006b).
388 S. Grano / Minerals Engineering 22 (2009) 386394
species of C; CO
2
3
(with for example Ca, Mg); Al
2
O
3
; SiO
2
; Cu
2
S;
SO
2
4
(with for example Ca, Fe), FeS
2
and Fe(OH)
3
as the major sur-
face species. There was close agreement in the total oxygen occur-
ring in these calculated species and the total atomic concentration
of oxygen.
3. Results
3.1. Particle and mineral size distribution in the mill products
While the same grind product d
80
(110 lm) on an all mineral
basis (solids) was achieved for both grinding environments, there
were small differences in the distribution of chalcopyrite between
the ceramic and mild steel media (Fig. 5). For both grinding envi-
ronments, chalcopyrite presented as a ner distribution than pyr-
ite, which in turn was ner than non-sulphide gangue (not
shown here) reecting differences in the hardness of the predom-
inant minerals. However, the overall distribution of chalcopyrite
was signicantly ner in the case of grinding with ceramic media
(d
80
45 lm) than with mild steel grinding (d
80
70 lm). Differ-
ences in grind product d
80
were less marked for pyrite (ceramic
media d
80
75 lm; mild steel media d
80
80 lm) and for non-
sulphide gangue (d
80
105 lm for both grinding environments).
There was less chalcopyrite contained in the 38105 lm size frac-
tion, with commensurately more chalcopyrite distributed uni-
formly throughout the size fractions below 38 lm in the case of
ceramic media (Fig. 5).
It is likely that differences in the size distribution of chalcopy-
rite reect differences in impact and stress intensity between the
two grinding mills, and hence differences in breakage rates be-
tween the minerals. Due to the low specic gravity of the ceramic
media (SG 3) compared with the mild steel media (SG 9), the
stress intensity and contribution from impact breakage would be
much lower for the ceramic media. Parry et al. (2006) has shown
that at low stirrer speeds, and hence low stress intensities, the
breakage rate of soft minerals were higher than those for hard min-
erals. In the current situation, while prolonged grinding times were
required to achieve the same particle d
80
in the case of the ceramic
media due to low absolute breakage rates, a greater difference in
breakage rate was apparent for the softer chalcopyrite relative to
the other minerals, particularly quartz. This may represent an
opportunity to improve the energy efciency of liberation of softer
value minerals by reducing the contribution of impact breakage
and exploiting differences in breakage rates between minerals at
low stress intensity. These subtle differences in the particle size
distribution produced with the different grind environments
0
-
5
5
-
7
7
-
9
9
-
1
7
1
7
-
2
2
2
2
-
3
8
3
8
-
4
5
4
5
-
5
5
5
5
-
7
5
7
5
-
1
0
0
+
1
0
5
Ceramic Media
0
5
10
15
20
Particle Size Range / Microns
%

i
n
S
i
z
e
R
a
n
g
e
Mild Steel Media
Fig. 5. Chalcopyrite distribution as a function of particle size range in the mill product for both grinding environments.
Chalcopyrite Recovery/%
(a) Copper Grade / Recovery
0
5
10
15
20
25
30
C
u

G
r
a
d
e

/

%
(b) Pyrite Selectivity
0
10
20
30
40
50
60
60 65 70 75 80 85 90 95 100
P
y
r
i
t
e

R
e
c
o
v
e
r
y

/

%
Ceramic Media
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
Mild Steel Media
SIBX=80 g/t
SIBX=50 g/t
SIBX=120 g/t
SIBX=80 g/t
Mild Steel Media
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
Mild Steel Media
SIBX=80 g/t
SIBX=50 g/t
SIBX=120 g/t
SIBX=80 g/t
Fig. 6. Copper grade as a function of copper recovery (a) and pyrite recovery as a function of copper recovery (b) for various collector additions and for the two grinding
environments.
S. Grano / Minerals Engineering 22 (2009) 386394 389
necessitated that recovery be assessed on a particle size basis as
discussed further below.
3.2. Flotation separation on an unsized basis
It is clear that grinding with ceramic media produced a vastly
superior copper grade-recovery relationship (Fig. 6a), and selectiv-
ity against pyrite (Fig. 6b), than grinding with mild steel media.
The main contributing factor was the higher otation rate and ulti-
mate recovery of chalcopyrite in the case of ceramic media. A min-
or contributing factor was a higher otation rate of pyrite recovery,
as indicated by the higher recovery at short otation times (rst
data point at 0.5 min), for the mild steel media when compared
with ceramic media at the same collector addition. This partially
contributed to the lower selectivity of copper against pyrite for
the mild steel media at the same collector addition. A possible
explanation for the increased pyrite recovery with the mild steel
medium may be related to copper activation of pyrite which is en-
hanced at reducing Eh values during grinding (Peng et al., 2003b).
Of greater signicance, is the fact that adding very high collec-
tor additions (120 g/t) did not restore copper recovery appreciably
for the case of mild steel media. Indeed, adding high collector addi-
tions mainly serves to lower the copper grade-recovery relation-
ship due to increased pyrite recovery. In a plant context,
attempts to increase copper recovery by adding more collector
would increase pyrite recovery without necessarily increasing cop-
per recovery appreciably (Pease et al., 2006b).
3.3. Pulp and surface chemistry
When compared to ceramic media, grinding with mild steel
media reduces the Eh by 450 mV (Table 1). The reduction in Eh is
a result of the redox reactions such as
FeOH
3
e

!FeOH
2
OH

E
0
0:56 V 1
FeOH
2
2e

!Fe OH

E
0
0:87 V 2
as a result of the preferential oxidation of the mild steel media to
form ferric and ferrous hydroxides. The increased presence of iron
hydroxides species on the particle surfaces after grinding with mild
steel media was conrmed by XPS analysis (Table 2). The dissolve
oxygen in the pulp after grinding with mild steel media was less
than 0.1 mg dm
3
. The pH of the pulp was also slightly decreased
after grinding with mild steel media, consistent with increased con-
sumption of hydroxyl ions to form precipitates with ferric and fer-
rous hydroxide (Peng et al., 2002). Aeration after grinding increased
the Eh values in the case of grinding with mild steel media to a sim-
ilar value to that of ceramic media, prior to collector conditioning.
Differences in otation recovery of ne chalcopyrite can not be ac-
counted by differences in Eh during the collector adsorption stage in
these experiments.
Grinding with ceramic media, decreased the surface concentra-
tion of iron hydroxide and commensurately increased the surface
atomic concentration of copper sulphide, silica and alumina (Table
2). The effect was greatest after etching to 25 nm suggesting that
the iron hydroxide occurred mainly as an over layer on other min-
erals. Hence, increases in the surface exposure of copper sulphide
after grinding with ceramic media may be interpreted to be a result
of lower surface coverage of the iron hydroxide. The over layer of
iron hydroxide evidently occurred to a degree on most of the min-
erals suggesting its unselective adsorption during grinding with
mild steel media. Measurements of EDTA extractable iron con-
rmed high concentrations of iron hydroxide present in the pulps
produced after grinding with mild steel media (395 mg/l) com-
pared with ceramic media (45 mg/l). Methods to discriminate iron
hydroxide precipitates from media and mineral sources are pres-
ently being pursued (Julianto, 2008).
3.4. Chalcopyrite recovery on a size-by-size basis
The key contribution to increased recovery of chalcopyrite for
the case of ceramic media is increased recovery of ne chalcopy-
rite, less than 10 lm (Fig. 7). As over 20% of the chalcopyrite in
the feed is in size fractions below 10 lm (Fig. 5), the low recovery
of this size fraction dominates losses in the case of mild steel med-
ia. High collector addition (120 g/t) does not increase ne copper
recovery at all in the case of grinding with mild steel media, but
does increase coarse copper recovery marginally at short
(0.5 min) and long (8.0 min) otation times. In contrast, grinding
with ceramic media shows much higher ne copper recovery
which accounts for the large difference in copper recovery between
the two grind environments, as previously noted on an unsized ba-
sis (Fig. 6). As in the case of the mild steel media, high collector
additions does not increase ne copper recovery at short and long
otation times. For the ceramic media, high collector additions
does not increase coarse (105 lm) copper recovery, a size fraction
which displayed lower recovery than for mild steel media. It is
speculated that most of the copper is contained in predominantly
liberated chalcopyrite particles for size fractions below 105 lm
and for both grind environments. However, the small amount of
chalcopyrite contained in the very coarsest size fraction (105 lm)
after grinding with ceramic media is probably contained in low
grade composites with silica which display very low hydrophobic-
ity. This would need conrmation using quantitative mineralogical
analysis. It is surmised that while the difference in breakage rate
between chalcopyrite and the non-sulphide gangue minerals was
greatest in the case of the ceramic media, a small amount of chal-
copyrite was contained within predominantly non-sulphide gan-
gue (silica) particles in the coarsest size fraction. In any case, the
large difference in copper recovery between the grinding environ-
ments on an unsized basis noted in Fig. 6 was due to the large dif-
ference in ne copper recovery, and that high collector additions
did not increase ne copper recovery at all.
Table 1
Eh and pH values in the mill discharge, before the aeration stage, for the two grind
environments.
Grind type pH Eh (mV)
Ceramic 8.3 230
Mild steel 8.1 220
Table 2
Atomic concentrations (%) of major species determined by XPS of mill discharge samples for the two grind environments, with and without 25 nm etching.
Etch Grind type Atomic concentration (%) of major species
Saturated hydrocarbon Carbonate Alumina Silica Copper sulphide Sulphate Iron sulphide Iron hydroxide
Without etch Mild steel 10 20.1 7.5 25.5 0.2 7.5 0 21.6
Ceramic 16 20 12.3 30.9 0.4 5.5 0.1 12.8
With 25 nm etch Mild steel 4.5 9.6 11.5 37.5 0.2 1.5 0.1 30
Ceramic 5 7.6 16 42 1.5 5.5 0.2 16
390 S. Grano / Minerals Engineering 22 (2009) 386394
3.5. Chalcopyrite otation rate constant as a function of size
Differences in chalcopyrite recovery as function of particle size
were reected in the distributed constant (assuming a single oat-
able and a non-oatable component) (Fig. 8). As the optimum sep-
aration was achieved at 50 g/t of SIBX, only otation rate constant
data at this collector addition is shown. The fraction of oatable
component, at innite otation time, was effectively the same as
the ultimate recovery at 8 min of otation time, the data for which
is shown in Fig. 7b. It is impossible to recover the non-oatable
fraction of ne chalcopyrite in the case of grinding with mild steel
media by simply increasing otation time, except through recovery
by the entrainment mechanism. Increasing the otation time will
have the effect of increasing pyrite recovery in intermediate size
fractions, but will not increase ne copper recovery, as discussed
further below.
It is also worthwhile pointing out that the very high otation
rate constants achieved for size fractions over the range of 0
105 lm in the case of ceramic media are characteristic of a hydro-
phobic surface with an advancing contact angle in the range of 70
80 across these size fractions (Muganda, 2008). The maximum
attainable contact angle for isobutyl xanthate is approximately
80 (Sutherland and Wark, 1955). At these high contact angles,
the attachment efciency approaches unity for ne particles less
than 10 lm (Dai et al., 1999), suggesting that the otation rate,
which is very high, is controlled through bubble-particle collision
frequency and efciency. In a plant context, the otation rate of
the ne hydrophobic chalcopyrite would be high, and recovery to
high values (>95%) would be possible given normal cell residence
times and hydrodynamics (Grano et al., 2007). In the case of the
mild steel media, a pronounced non-oatable fraction of chalcopy-
rite is observed (Fig. 7b), which can not be recovered at very long
otation times, and with very high collector additions. Fine parti-
cles, less than 5 lm, require a critical advancing contact angle
(63) be exceeded to allow their attachment to bubbles (Gontijio
et al., 2007). It is clear that in the case of mild steel media, approx-
0
20
40
60
80
100
R
e
c
o
v
e
r
y

/

%
0
20
40
60
80
1 10 100
1000
R
e
c
o
v
e
r
y

/

%
(a) Tf=0.5 min
(b) Tf=8.0 min
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
Mild Steel Media
SIBX=80 g/t
SIBX=50 g/t
SIBX=120 g/t
SIBX=80 g/t
Chalcopyrite
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
Mild Steel Media
SIBX=80 g/t
SIBX=50 g/t
SIBX=120 g/t
SIBX=80 g/t
Particle Size / Microns
Fig. 7. Chalcopyrite recovery as a function of particle size for otation times (Tf) of 0.5 min (a) and 8.0 min (b) for various collector additions and for the two grinding
environments.
0
1
2
3
4
5
6
1 10 100 1000
Ceramic Media
Mild Steel Media
R
a
t
e

C
o
n
s
t
a
n
t

/

m
i
n
-
1
Chalcopyrite
0
1
2
3
4
5
6
Particle Size / Microns
Fig. 8. Distributed otation rate constant of chalcopyrite as a function of particle size at 50 g/t SIBX collector addition and for the two grinding environments.
S. Grano / Minerals Engineering 22 (2009) 386394 391
imately 40% of the ne chalcopyrite does not exceed this critical
contact angle and therefore appears in a non-oatable component
which reports to the otation tailing. In the case of grinding with
ceramic media, a high contact angle value on the ne particles is
possible giving rise to high maximum recovery and otation rate
constant.
3.6. Pyrite recovery on a size-by-size basis
The main contribution to the increased otation rate and ulti-
mate recovery of pyrite in the case of mild steel media was from
increased recovery of pyrite in intermediate size fractions (15
55 lm), when compared with ceramic media at the same collector
addition (Fig. 9). This concurs with previous studies on the chalco-
pyrite-pyrite mineral system that have shown that grinding under
reducing conditions promotes pyrite activation by copper (Peng et
al., 2003b). In any case, increasing collector addition to increase
ne chalcopyrite recovery only serves to increase pyrite recovery
in intermediate size fractions, for both grind environments. The
high oatability of pyrite in the intermediate size fractions at high
collector addition, which would be apparent in the rougher con-
centrate, would make its separation in subsequent cleaning stages
difcult without taking specic steps to actively depress pyrite by
some means. This observation concurs with those at plant scale
which show that adding more collector generally does not increase
ne value mineral recovery, but does increase recovery of gangue
sulphide minerals in intermediate size fractions, when regrinding
with reactive steel media (Pease et al., 2006a).
Importantly, the optimum separation of chalcopyrite from pyr-
ite on an unsized basis (Fig. 6b) occurs for grinding with ceramic
media at the lowest collector addition. Adding high collector addi-
tions does not increase ne chalcopyrite recovery, but increases
pyrite recovery in intermediate size fractions. In a plant context,
lower collector additions could be expected for the case of ceramic
grinding media as high copper recovery is achieved at low collector
additions (Pease et al., 2006b). It should be noted that the usual
practice for this ore is to add sodium cyanide, an effective pyrite
depressant in this system (Grano et al., 1994). Further improve-
ments to selectivity may be possible by controlling other aspects
of the grinding environment apart from the media type, such as
pH and the dissolved oxygen content during grinding as has been
shown for a complex copperleadzinc ore (Huang and Grano,
2006c). Particularly, pyrite otation is very sensitive to oxygen
content during grinding due to its inadvertent metal ion activation
(Huang and Grano, 2006c). Recently, a specialised laboratory
grinding mill has been developed which allows the pulp chemistry
during grinding to be more conveniently controlled to optimise
selectivity and to allow closer correlation of the laboratory to plant
grinding chemistry (Greet et al., 2004). Future advances in stirred
milling technology may allow pulp chemical control of these types
of mills also.
4. Discussion
It is apparent that the surfaces of the minerals become contam-
inated with iron hydroxide emanating from their contact with the
grinding media in the case of grinding with reactive mild steel
media. It could be surmised that the ne chalcopyrite is affected
to a greater extent than the other size fractions, due to the mech-
anism of generation of ne particles in abrasion with the grinding
media and other minerals. Evidence for the association of iron
hydroxide derived from reactive mild steel media with ne, and
not coarse, galena particles during grinding has been presented
elsewhere (Peng et al., 2003a). Time of ight secondary ion mass
spectrometry showed a higher concentration of iron hydroxide
on ne galena particles and much less on coarse galena particles
(Peng et al., 2003a). A possible explanation may also be related
to the higher levels of oxidation of the ne sulphide particles
enhancing adsorption of iron hydroxide through a range of mech-
anisms (Peng et al., 2003a). The exact mechanism/s are still un-
clear, and methods to characterise the surfaces of ne particles
exclusively of other size fractions in real ore slurries still need
development (Kinal, 2008).
Attention will now turn to the mechanism for the depression of
chalcopyrite by iron hydroxide. Other studies have shown that the
collectorless and collector-induced otation of chalcopyrite (size
range 3875 lm) is reduced by the presence of small concentra-
tions of ferric hydroxide over the pH range of 68 (Grano et al.,
(a) Tf=0.5 min
0
20
40
60
80
100
(b) Tf=8.0 min
0
20
40
60
80
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
SIBX=50 g/t
SIBX=120 g/t
Mild Steel Media
Pyrite
SIBX=50 g/t
SIBX=120 g/t
Ceramic Media
SIBX=50 g/t
SIBX=120 g/t
Mild Steel Media
1 10 100 1000
Particle Size / Microns
R
e
c
o
v
e
r
y

/

%
R
e
c
o
v
e
r
y

/

%
Fig. 9. Pyrite recovery as a function of particle size for otation times (Tf) of 0.5 min (a) and 8.0 min (b) for various collector additions and for the two grinding environments.
Only collector additions of 50120 g/t of SIBX are shown for clarity.
392 S. Grano / Minerals Engineering 22 (2009) 386394
1997a,b). Reduction in the adsorption of thiol collectors onto chal-
copyrite is the principal mechanism responsible, with high afnity
sites for thiol adsorption on chalcopyrite obscured by the presence
of ferric hydroxide (Grano et al., 1997b). Hence, the effective con-
tact angle on the chalcopyrite surfaces is reduced by the continuing
presence of hydrophilic iron hydroxide overlayers which do not
interact with thiol collectors at neutral to alkaline pH values. The
consumption of xanthate collector by grinding media oxidation
products has been suggested as a possible mechanism to account
for decreased sulphide mineral recovery with reactive steel grind-
ing media (Rey and Formanek, 1960). However, a more likely
explanation is related to the continuing presence of the iron
hydroxide layer and the reduced area on the particle surface which
can be rendered hydrophobic through collector adsorption. Only
relatively small surface coverages of hydrophilic iron hydroxide,
of the order of 20%, are required to reduce the contact angle of
the chalcopyrite to less than the critical contact angle required
for ne particle attachment to bubbles (Gontijio et al., 2007).
Experimental evidence has shown that the otation recovery of
intermediate size galena (2538 lm), with low surface area, is se-
verely depressed by iron hydroxide at only 8% of an equivalent
monolayer (Bandini et al., 2001).
In the case of ne chalcopyrite particles, less than 10 lm, these
require a higher critical contact angle to affect their attachment
to bubbles when compared with intermediate sized particles (Gon-
tijio et al., 2007). Thus while intermediate sized particles may have
some coverage of iron hydroxide, the oatability of ne particles of
chalcopyrite may be more greatly affected. At a critical coverage of
iron hydroxide, bubble attachment can no longer take place and
the ne particles report to the tailings as a non-oating compo-
nent. Even for the oating fraction, the otation rate of the ne
particles is reduced when grinding with mild steel media due to
the decreased contact angle on these particles. If ne particles of
chalcopyrite have indeed higher coverages of iron hydroxide,
then the reduced otation rate and maximum recovery of ne
chalcopyrite after grinding with reactive steel media can be easily
explained.
Attention will now turn to whether the benets of grinding
with fully inert media is only a result of decreased contamination
by iron hydroxide, or if there is an intrinsic benet also in the attri-
tion mechanism of size reduction which may be more important
with the ceramic media in the current study. With attritioning, sur-
face oxidation products may be removed from the particle sur-
faces. This has been addressed, to an extent, in a separate study
on arsenopyrite otation (Huang and Grano, 2006b). In the case
of ceramic media, and grinding in a stirred bead mill, the increased
arsenopyrite recovery noted was principally accounted by de-
creased surface contamination from iron hydroxide (Huang and
Grano, 2006b). However, further work is required to differentiate
these mechanisms, particularly for the case of hydrophilic surface
coatings which may be present on minerals prior to grinding.
Examples include iron hydroxide from grinding media contamina-
tion derived from previous grinding stages in plants, oxidation
products of the minerals at the surface, and gangue slime coatings
which seems to be particularly important in molybdenite otation.
It is also worthwhile pointing out that the magnitude of the
benets of grinding with electrochemically inert grinding media
will depend upon the prevailing pulp chemical environment (pH,
Eh), media type, ore mineralogy and target particle size. As an
example, in a laboratory study on Solobo copper ore, Goncalves
et al. (2003) found that grinding in a ceramic mill with ceramic
media did not yield the same otation results as that produced
by grinding in a mill with stainless steel rods. Goncalves et al.
(2003) speculated that the reason for this was related to oxidation
of the predominantly bornite mineral during prolonged grinding in
the oxidising environment of the ceramic mill and media
(Eh + 200 mV). It is well established that bornite may oxidise more
rapidly than chalcopyrite (Fullston et al., 1999).
Controlling the pulp chemical environment, specically the
pulp oxygen content, during grinding with ceramic media may
mitigate against undue mineral oxidation during grinding. In a re-
cent study on a complex copperleadzinc sulphide ore, it has
been found that enhanced selectivity of chalcopyrite against gan-
gue sulphide minerals was achieved by using both high chromium
grinding media and low oxygen content during grinding (Huang
and Grano, 2006c). The above discussion demonstrates the impor-
tance of considering also the chemical environment generally in
the evaluation of grinding media. The Magotteaux Mill

(Greet et
al., 2004) has been specially developed to allow the pulp chemical
environment to be manipulated and was used in the study by
Huang and Grano (2006c). The Magotteaux Mill

allows the pulp

oxygen content and pH during grinding to be continuously con-
trolled independently of the grinding media type (Greet et al.,
2004). Further development of stirred milling technology to allow
continuous control of the pulp chemical environment in grinding is
also warranted.
In terms of other pulp chemical parameters, the grind pH is also
critical. Generally, it has been observed that the Eh value of a ball
discharge increases as the grind pH increases with lime or soda ash
addition (e.g., Grano et al., 1997c) suggesting that media oxidation
reactions are inhibited at high pH. This may explain, for example,
the increase in ne chalcopyrite recovery with lime addition to
mills (Orwe et al., 1997). The pH of the grind needs also to be con-
sidered in the pulp chemical matching of the laboratory to the
plant (Greet et al., 2004). However, it seems possible that even at
high pH there is still surface contamination by grinding media
through other mechanisms which requires further examination.
In other cases, it is not possible to operate at elevated pH such as
in the otation of pyrite and arsenopyrite for gold recovery.
Another important aspect in the evaluation of the impact of the
grinding media is the particle size distribution after milling. In the
study of Goncalves et al. (2003) the amount of nes (10 lm) in
the laboratory mill product was relatively low at 10% compared
with the currently reported study which was 20%. Clearly, differ-
ences in the target grind d
80
and the use of rods in the study by
Goncalves et al. (2003), rather than balls used in this current study,
are contributing factors. This highlights the importance of examin-
ing otation behaviour as a function of particle size at least. It is ex-
pected that the relative benets of grinding with electrochemically
inert media will increase with decreasing target grind size due to
the increasing contribution of ne particles less than 10 lm (Ye,
2008). Grinding technology which minimises the production of
ne particles will also assist with improving recovery and selectiv-
ity (Pease et al., 2006a).
Recent developments in large scale stirred milling technology
using competent ceramic media will allow their application to rel-
atively coarser particle applications (Anyimadu et al., 2006; Pease
et al., 2006a,b). It is now possible to grind a rougher tailing from
150 lm to 55 lm in a 3.3 MW IsaMill using MT1 ceramic media
(Pease et al., 2006a). Quantifying contributions to increased rough-
er recovery from improved liberation, narrower particle size distri-
butions and improved surface properties of the mineral particles
remain a target for research.
5. Conclusions
Experimental evidence of the effect of the grinding environment
on ne (10 lm) value mineral recovery in otation was pre-
sented. Reasons for increases in ne value mineral recovery from
ores with fully electrochemically inert grinding media are dis-
cussed in relation to reduced surface contamination by iron
S. Grano / Minerals Engineering 22 (2009) 386394 393
hydroxide emanating from the grinding media. It is suggested that
the effect of the grinding media, which is important for ne parti-
cles and progressively becomes more important for ner grind
sizes, is principally due to the abrasion mechanism of the minerals
with the media in the production of the ne particles, though this
mechanism needs to be substantiated through further experimen-
tal studies.
Acknowledgments
The author gratefully acknowledges nancial support and per-
mission to publish from Xstrata Copper. The experimental work
described in this paper was conducted in the early 1990s in a pro-
ject with Mount Isa Mines Ltd. The author would like to express
gratitude to Bill Johnson for his guidance in this and other projects.
References
Anyimadu, A.K., Rule, C.M., Knopjes, L., 2006. The development of ultra-ne grinding
at Anglo Platinum. In: International Platinum Conference Platinum Surges
Ahead. The South African Institute of Mining and Metallurgy, pp. 143149.
Bandini, P., Prestidge, C., Ralston, J., 2001. Colloidal iron oxide slime coatings and
galena particle otation. Minerals Engineering 14 (5), 487497.
Cullinan, V., 1999. Improving the otation response of ne galena, Ph.D. Thesis,
University of South Australia.
Cullinan, V., Grano, S.R., Greet, C., Johnson, N.W., Ralston, J., 1999. Investigating ne
galena recovery problems in the lead circuit of Mount Isa Mines lead/zinc
concentrator, Part I: grinding media effects. Minerals Engineering 12, 655669.
Dai, Z., Fornasiero, D., Ralston, J., 1999. Particle-bubble attachment in mineral
otation. Journal of Colloid and Interface Science 217 (1), 7076.
Fullston, D., Fornasiero, D., Ralston, J., 1999. Zeta potential study of the oxidation of
copper sulde minerals. Colloids and Surfaces A: Physicochemical and
Engineering Aspects 146 (13), 113121.
Gao, M., Young, M.F., Cronin, B., Harbort, G., 2000. IsaMill media competency and its
effect on milling performance. Mineral and Metallurgical Processing 18 (2),
117120.
Goncalves, K.L.C., Andrade, V.L.L., Peres, A.E.C., 2003. The effect of grinding
conditions on the otation of a sulphide copper ore. Minerals Engineering 16
(11), 12131216.
Gontijio, C., Fornasiero, D., Ralston, J., 2007. The limits of ne and coarse particle
otation. Canadian Journal of Chemical Engineering 85, 739747.
Grano, S.R., Ralston, J., Smart, R.St.C., 1990. The inuence of electrochemical
environment on the otation behaviour of Mount Isa Mines copper and lead/
zinc ores. International Journal of Mineral Processing 30, 6997.
Grano S.R., Wong, P., Skinner, W., Johnson, N.W., Ralston, J., 1994. The effect of
autogenous and ball mill grinding on the chemical environment and otation of
the copper ore of Mount Isa Mines Limited. In: III Latin-American Congress on
Froth Flotation (University of Concepcion Publ.), Concepcion, Chile, pp. 351
388.
Grano, S.R., Cnossen, H., Prestidge, C.A., Skinner, W., Ralston, J., 1997a. Surface
modications in the chalcopyrite sulphite ion system, II. Dithiophosphate
collector adsorption study. International Journal of Mineral Processing 50 (12),
2745.
Grano, S.R., Sollaart, M., Prestidge, C.A., Skinner, W., Ralston, J., 1997b. Surface
modications in the chalcopyrite sulphite ion system, I. Collectorless otation,
XPS and dissolution study. International Journal of Mineral Processing 50 (12),
126.
Grano, S.R., Johnson, N.W., Ralston, J., 1997c. Control of the solution interaction of
metabisulphite and ethyl xanthate in the otation of the Hilton ore of Mount Isa
Mines Limited, Australia. Minerals Engineering 10 (1), 1739.
Grano, S.R., Akroyd, T., Mular, M., 2007. A model study of copper rougher recovery
optimisation at PT Freeport Indonesia. In: Proc. Ninth Mill Operators
Conference. Aust. Inst. Min. Metall Publ., pp. 2538.
Greet, C.J., Small, G.L., Steinier, P., Grano, S.R., 2004. The Magotteaux Mill

:
investigating the effect of grinding media on pulp chemistry and otation
performance. Minerals Engineering 17 (78), 891896.
Huang, G., Grano, S., 2006a. Galvanic interaction between grinding media and
arsenopyrite and its effect on otation, Part I. Quantifying galvanic interaction
during grinding. International Journal of Mineral Processing 78, 182197.
Huang, G., Grano, S., 2006b. Galvanic interaction between grinding media and
arsenopyrite and its effect on otation, Part II: effect of grinding on otation.
International Journal of Mineral Processing 78, 178213.
Huang, G., Grano, S., 2006c. Improving the otation behavior of a sulde ore by
controlling electrochemical interactions during grinding, electrochemical
society transactions. Electrochemistry in Mineral and Metal Processing VII 2,
920.
Johnson, N.W., 1987. An appreciation of chalcopyrite metallurgy in the copper
concentrator, Mount Isa Mines Technical Report.
Johnson, N.W., 2006. Liberated 010 lm particles from sulphide ores, their
production and separation. Recent developments and future needs. Minerals
Engineering 19 (68), 666674.
Julianto, J., 2008. Distinguishing and quantifying the relative contributions of
oxidized iron from grinding media and mineral sources, M.Sc. Thesis, University
of South Australia.
Kinal, J., 2008. Characterising the surface phases on ne particles less than 10 lm in
heterogenous mixtures, Ph.D. Thesis, University of South Australia.
Muganda, S., 2008. Correlating particle contact angle to otation rate in a well
dened turbulent environment, Ph.D. Thesis, University of South Australia.
Orwe, D., Grano S.R., Lauder, D.W., 1997. Chalcocite oxidation and its inuence on
ne copper recovery at the Ok Tedi concentrator, Papua New Guinea. In: Fifth
Mill Operators Conference, Madang, PNG, (Aust. Inst. Min. Metall. Publ.), pp. 85
96.
Parry, J., Klein, B., Lin, D., 2006. Comparison of ultra-ne grinding technologies. XXIII
International Mineral Processing Congress, Istanbul, Turkey (pp. 177184).
Pease, J.D., Curry, D.C., Barnes, K.E., Young, M.F., Rule, C., 2006a. Transforming ow
sheet design with inert grinding media the IsaMill, Paper 16, 38th Annual
Meeting of the Canadian Mineral Processors, pp. 231249.
Pease, J.D., Curry, D.C., Young, M.F., 2006b. Designing otation circuits for high nes
recovery. Minerals Engineering 19, 831840.
Peng, Y., Grano, S.R., Fornasiero, D., Ralston, J., 2002. Towards prediction of oxidation
during grinding 1. Galena otation. Minerals Engineering 15, 493498.
Peng, Y., Grano, S.R., Fornasiero, D., Ralston, J., 2003a. In: Lorenzen, L., Bradshaw, D.
(Eds.), The effect of grinding conditions on ne galena otation and its
separation from pyrite. XXII International Mineral Processing Congress., Sth.
Afr. Inst. Min. Metall. Publ., pp. 696705.
Peng, Y., Grano, S.R., Fornasiero, D., Ralston, J., 2003b. Control of grinding conditions
in the otation of chalcopyrite and its separation from pyrite. International
Journal of Mineral Processing 69, 87101.
Rey, M., Formanek, V., 1960. Some factors affecting the selectivity in the differential
otation of lead zinc ores, particularly in the presence of oxidised lead
minerals. In: Proc. 5th Int. Mineral. Process Congr, London, pp. 343353.
Sutherland, K.L., Wark, I.W., 1955. Principles of Flotation. Australasian Institute of
Mining and Metallurgy Inc., Melbourne (pp. 98).
Ye, X., 2008. Optimising grinding and regrinding chemistry, Ph.D. Thesis, University
of South Australia.
394 S. Grano / Minerals Engineering 22 (2009) 386394