Upload
german-reyes
View
22
Download
0
Embed Size (px)
Citation preview
atis p
. S
12, C
3 Oct
a sui
smelting trials. The smelting tests were conducted as part of a wider CSIROMinerals research project aimed at developing a single-
stage continuous copper making process.
conducting research aimed at developing a continuous
et al., 1984).
which copper concentrates are converted to low sulphur
copper value. The feasibility of the whole copper making
process will most likely depend on the eciency and cost
eectiveness of the copper recovery stage. The slags
produced so far in single-stage copper making labora-
tory studies have varied in copper content from about
3% to 20% Cu. The copper can occur in the slag as*
Minerals Engineering 17 (20Corresponding author. Tel.: +61-3-9545-8500; fax: +61-3-9562-
8919.copper converting process to replace the traditional two-
stage batch process that is currently used to produce
blister copper (Jahanshahi et al., 1994, 1995). Sirosmelt
type reactors and calcium-ferrite (lime-based) slags have
been used rather than the conventional iron silicate
slags. A number of advantages of using calcium-ferriteslags rather than conventional iron silicate slags have
been identied and are described elsewhere (Eerola
blister copper in a single step, that is the smelting and
converting steps occur in one continuous stage (Som-
erville et al., 1995). Preliminary kilogram-scale labora-
tory and pilot plant experiments have shown the
smelting process is technically feasible.
One feature of the direct copper converting process isthat copper losses to the slag phase can be high, meaning
the slag produced needs to be re-treated to recover theThe slags treated included two moderately reduced low copper slags, an oxidised high copper slag and a very reduced low copper
self-pulverising slag. The total copper content of the slags varied from 5.2% to 15.1% Cu. The copper was present in the form of
metallic copper, oxide copper and copper ferrite phases with the dominant phase in all slags being metallic copper. The slags were
crushed, ground, wet screened at 210 or 75 lm to remove coarse metallic particles of copper, and oated at natural pH (about pH11) using reagents and conditions appropriate for the selective recovery of the copper phases.
In rougherscavenger tests on the slags total copper recoveries between 80% and 87% were obtained for three of the slags tested.
The best result for the fourth slag (low copper self-pulverising slag) was 74% copper recovery. Coarse copper metal was present in
the otation tails for this test suggesting the pre-otation screen size was too coarse. Further optimisation of the grinding and
otation conditions for this slag should yield an improved copper recovery.
Minor dierences in copper/iron selectivity for the otation tests on the four slags were observed with the low copper slag giving
the most selective result and the high copper slag the least selective result. These trends were generally consistent with the kinetic
data obtained in the otation tests and the given copper mineralogy for the dierent slag types.
Crown Copyright 2004 Published by Elsevier Ltd. All rights reserved.
Keywords: Froth otation
1. Introduction
Over the past few years CSIRO Minerals has been
The research work has now been extended to focus on
the development of a single-stage continuous coppermaking process (so-called direct copper converting) inThe recovery of copper, by otslags made in continuou
W.J. Bruckard *, M
CSIRO Division of Minerals, Box 3
Received 2
Abstract
A series of laboratory batch otation tests was conducted onE-mail address: [email protected] (W.J. Bruckard).
0892-6875/$ - see front matter Crown Copyright 2004 Published by Elsevdoi:10.1016/j.mineng.2003.12.004on, from calcium-ferrite-basedilot plant smelting trials
omerville, F. Hao
layton South, Vic., 3169, Australia
ober 2003
te of calcium-ferrite-based slags made in continuous pilot plant
04) 495504This article is also available online at:
www.elsevier.com/locate/minengcopper metal (Cu0), copper oxide (Cu2O and CuO) and
ier Ltd. All rights reserved.
a calciumcopper ferrite phase ((Cu,Ca)O Fe2O3) withthe deportment of copper to each phase type being
dependent on the slag chemistry. The oxidation state
(Fe3/Fe2 ratio) and the CaO/Fe ratio in the slag arethe two critical smelting parameters. Both, for example,
inuence the stability of the ferrite phase, which is the
phase that presents the most diculty from the point ofview of recovering copper.
undertaken on these slags (referred to here as pilot
plant slags) to conrm or otherwise the previous o-
reduced low copper self-pulverising slag, and a fourth
moderately reduced low copper slag prepared from Es-
condida copper concentrate. A description of the slags
tested together with their chemical composition (ICP
analysis) is given in Table 1.
The analytical results in Table 1 show that the slags
tested varied in total copper content from 5.2% to 15.1%Cu with the Escondida slag having a mid-range value of
u0 (%
.22
.21
.95
.4
496 W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504tation results obtained and to help identify operating
conditions for any future continuous otation pilot
plant trials. The results of the batch otation work
undertaken on the pilot plant slags are presented in this
paper.
2. Experimental procedures
2.1. Slag samples
Laboratory batch otation tests were conducted on
four slag samples containing dierent levels of copper.
The slags (referred to here as pilot plant slags) weremade in continuous pilot plant smelting tests, conducted
on-site at CSIRO Minerals using a Sirosmelt reactor
and copper sulphide concentrates from MIM and Es-
condida. Details of the smelting trials are described by
Jahanshahi et al. (1995).
The slags treated included three prepared from MIM
copper concentrate, namely a moderately reduced low
copper slag, a more oxidised high copper slag, a very
Table 1
Description and chemical analysis of slags
Slag type Cu (%) C
Very reduced low copper self-pulverising slag 5.24 5
Moderately reduced low copper Escondida slag 9.77 8
Moderately reduced low copper slag 7.87 6
Oxidised high copper slag 15.10 11
aMetallic copper.Flotation has been identied as the most economical
option to recover the copper from direct copper con-
verting slags (Somerville et al., 2000) where it has been
estimated that a copper recovery from the slag of above
90% is required to make the overall single-stage copper
making process economically feasible.
The results of batch otation tests undertaken on asuite of slags made in laboratory smelting tests (Bruc-
kard et al., 2003) showed that total copper recoveries
above 90% could be obtained from slags where the
predominant form of copper was metallic copper. To
prove the smelting process at pilot plant scale approx-
imately 850 kg of slag (three dierent slag types) was
produced in CSIRO Minerals pilot scale continuous
Sirosmelt reactor. Batch otation tests were then9.8% Cu. The CaO/Fe ratio was typically 0.63 except for
the Escondida slag where it was 0.55. In general, the
higher the copper content the more oxidised the slag. It
is also noted that for all the slags at least 75% of the
copper present was in the form of metallic copper. The
copper-bearing and other phases present in the slags are
discussed in more detail later.
2.2. Crushing and blending
All the pilot plant slags, except the self-pulverising
slag, were crushed before grinding. The slags were cru-
shed to pass 2 mm using a laboratory jaw crusher in
closed circuit with a screen. Coarse pieces of metallic
copper were removed by hand and discarded. They werenot included in metal recovery calculations. Crushed or
as received samples were then blended and ried into
200 g lots for otation testing. Sub-samples were cut out
for chemical and sizing analyses. The amount of tramp
copper removed during crushing, expressed as a per-
centage of the total slag weight, was variable and ranged
from 0% to 2%. It is expected that in any continuous
operation this tramp copper would be readily recovered.
2.3. Grinding and screening
For each test, a 200 g batch of slag (as received or
crushed) was mixed with 100 ml of distilled water and
ground at 67% solids by weight in a small stainless steel
rod mill (13 cm by 13 cm) with fteen stainless steel rods
(18 mm diameter). Grinding times were varied toachieve the desired P80 by weight for the ground prod-
ucts. The grind times for each slag type were established
by conducting standard grindability tests. For most tests
the target P80 was 40 lm. This grind size was found to besuitable in previous otation test-work on continuous
converting slags (Bruckard et al., 2003). Before each
grind the mill and rod charge were cleaned by grinding
with quartz and distilled water for 5 min.
)a Fe (%) Mg (%) Ca (%) Si (%) CaO/Fe
33.6 3.66 15.2 7.36 0.63
34.6 2.55 13.6 8.67 0.55
34.7 3.74 15.8 7.18 0.64
32.0 4.07 14.2 6.19 0.62
variable speed drive and was driven from below, to
allow the whole surface of the froth to be scraped with a
(Na2S 9H2O) was used as a sulphidising agent andAnalaR grade copper sulphate (CuSO4 5H2O) used
W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504 497was as an activator. Both of these reagents were addedas solids to the otation cell as required. The frother
used was a commercial polypropylene glycol frother
(Cyanamid Aerofroth 65) prepared as a 0.25% w/v
solution in distilled water. The otation gas was high
purity bottled air (a synthetic mixture of O2 and N2)
and distilled water was used in all tests to maintain the
cell volume.paddle at a constant depth and rate. The cell was tted
with a rubber diaphragm, sight tube and electronic
sensor for automatic detection and control of the pulp
level.
For both conditioning and otation the impeller
speed was 1200 rpm. Air was delivered to the cell at 8
dm3/min and pulp temperatures ranged from 18 to 22
C. Frother was added continuously during otation viaa variable speed dispenser at a rate of 1 ml/min. For a
slag charge of 200 g the addition of 1 ml of frother was
equivalent to 12.5 g/t.
The pulp pH and potential were monitored continu-
ously during testing. The pulp potential was measured
using a high-impedance, dierential voltmeter with a
platinum ag electrode and a calomel reference elec-
trode. The performance of the electrode system waschecked using standard ferricferrous ion solution
(Light, 1972). Measured potential values were converted
to the standard hydrogen electrode (SHE) scale by the
addition of 245 mV. A Radiometer glass/saturated cal-
omel electrode was used to measure pH. Before each test
the pH of the system was calibrated using standard pH 7
and pH 11 buer solutions.
2.4.2. Flotation reagents
The otation collector used was a commercial
potassium ethyl xanthate (KEX). A fresh batch of
xanthate was prepared each day as a 0.1% w/v solution
in distilled water. Laboratory grade sodium sulphideCoarse metallic copper was removed from the ground
pulp prior to otation by wet screening at either 210 or
75 lm. The oversize material was removed, dried,weighed and analysed, while the undersize portion be-
came the otation feed. The coarse metallic fractions
(screen oversize) were included in the calculation of
overall metal recoveries.
2.4. Flotation
2.4.1. Flotation equipment
Samples were oated in a 3 dm3 modied stainless
steel Denver cell in which the impeller was tted with a2.4.3. Flotation procedures
Flotation tests were conducted on the wet screen
undersize portions of the ground slag pulps. Standard
operating procedures (Guy, 1992) and test conditions
established in previous otation work on copper slag
samples (Bruckard et al., 2003) were used. In general a
two-stage otation procedure was adopted. The rststage (rougher oat) involved the addition of
Na2S 9H2O (5 min conditioning then 5 min aeration),followed by the addition of KEX (5 min conditioning)
after which otation concentrates were collected over 8
min. Rougher concentrates were generally collected
after 1, 4 and 8 min otation. In the second stage
(scavenger oat), more KEX collector was added, the
pulp conditioned (2 min) and further concentrates col-lected over 8 min. Scavenger concentrates were generally
collected after 2, 4 and 8 min corresponding to cumu-
lative otation times of 10, 12, and 16 min. In some tests
a second scavenger concentrate was collected (4 min)
after the addition of CuSO4 5H2O activator and furtherNa2S 9H2O and KEX.
All tests were conducted at the natural pH of the
respective slag mixtures, which was generally betweenpH 10.5 and 11.5, and a scraping rate of once every 5 s
was maintained for all tests. Pulp potential was mea-
sured but not controlled and the natural Eh of the sys-
tem was between 200 and 300 mV SHE. The potential
decreased following the sulphide addition but returned
to the air set potential after the 5 min aeration stage.
The pulp volume was maintained at a preset level by
small additions of distilled water. All otation productswere weighed wet, to allow calculation of water recov-
eries, washed with alcohol, dried and prepared for
analysis in a standard manner. Full details of the test
conditions used are given in Table 2.
2.5. Chemical analysis of solid products
Slag head samples and otation products were as-sayed for total copper, iron, calcium, magnesium and
silica by inductively coupled plasma atomic emission
spectroscopy (ICP-AES) using a standard method. A
separate sodium peroxide fusion was required for the
silica analysis. Metallic copper was determined sepa-
rately using a brominemethanol leach procedure
incorporating an ICP-AES nish.
2.6. On-line solution analysis
On-line monitoring of xanthate present in the pulp
solution was conducted in a number of otation tests.
Filtered solution samples were collected continuously
from the pulp during otation and pumped to a diode
array detector, set up to measure xanthate absorbance at
301 nm. The solution samples were then recycled backto the otation cell.
size fractions (CS6 and CS7), below those obtained by
plant
)
498 W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504conventional cyclosizing (CS1-CS5), are collected usinga continuous centrifuge and a multiple decantation
procedure. All size fractions were analysed by ICP-AES.
In calculating elemental distributions and cut sizes in the
sub-sieve sizing range a specic gravity (SG) of 4.2 was
assumed for all the slags.
2.8. X-ray diraction analyses
X-ray diraction (XRD) patterns were recorded on a
Philips PW 1050 goniometer with a PW 1710 diraction
controller using CuKa radiation. Phases present wereidentied by comparison of the peak positions and
intensities with data published by the International
Centre for Diraction Data (ICDD).
3. Results and discussion
Test results are discussed in terms of slag mineralogy,
copper recovery, selectivity over gangue elements
(especially iron), reagent considerations and particle size
responses. The otation response for each slag type is
considered in sequence, which follows the chronological2.7. Sizing analyses
Sizing analyses was conducted on selected slags and
otation products using standard laboratory wet and
dry screening methods. Where sub-sieve sizing was re-
quired a modied CSIRO cyclosizing technique was
used (Kelsall et al., 1974). With this method two further
Table 2
Test conditions and total reagent additions for otation tests on pilot
Test no. Slag type Grind time
(min)
Pre-otation
screen size (lm
SL28 Self-pulverising 210
SL29 Self-pulverising 10 210
SL31 Escondida 32 75
SL32 Low copper 22 75
SL33 High copper 24 75
SL34 High copper 24 75
a P80 of otation tail.order in which the slags were produced and tested.Flotation tests have identication numbers SL28 to
SL34.
3.1. XRD analyses of slags
XRD analyses were conducted on all the pilot plant
slags. The results show that the major copper phase
present in all four slags was metallic copper (Cu0). In thelow copper and self-pulverising slags there were minor
amounts of cuprite (Cu2O) while in the more oxidised
high copper pilot plant slag the levels of cuprite weresubstantial and there was also a trace of delafossite(Cu2O Fe2O3). These results are generally consistentwith the total copper and metallic copper analyses of the
slags (see Table 1).
In all the slags the major iron-bearing component was
magnetite, with minor amounts of metallic iron. The
magnesium occurred as a variety of magnesium and
calcium/magnesium silicates, and the calcium was also
present as calcium silicate phases.
3.2. Metallurgical results
Flotation test conditions and reagent additions are
listed in Table 2, while a summary of the metallurgical
results is given in Table 3. The results refer to the
combined two-stage rougherscavenger concentrate and
recoveries quoted include contributions from metalliccopper recovered by wet screening prior to otation.
3.2.1. Self-pulverising slag
Under certain circumstances calcium-ferrite slag will
self-pulverise, or disintegrate, during cooling (Inoue and
Suito, 1991; Juckes, 2002). It is believed that conditions
that favour self-pulverisation are a reduced slag and a
CaO/SiO2 ratio close to 2. In this work the CaO/SiO2ratio for the slags tested varied between 1.02 and 1.50
with the ratio for the self-pulverising slag being 1.35.
This apparent inconsistency was not investigated fur-
ther. The suitability of otation to recover copper from
this type of slag was examined using the very reduced
low copper self-pulverising pilot plant slag. The P80 of
the self-pulverising slag as received was 74 lm. The slag
slags
P80 (lm)a Total reagent additions (g/t)
KEX Na2S 9H2O CuSO4 5H2O
74 200 800 100
37 200 800 100
41 150 500
41 150 500
39 150 500
39 250 500 was oated as received and also after grinding (in thiscase the P80 of the otation tail was 37 lm). The overallresults for both tests (SL28 and SL29 respectively) are
given in Table 3 and a recovery-time plot for test SL28 is
shown in Fig. 1.
The results in Fig. 1 indicate that the selectivity of
copper over the gangue elements is reasonably good and
the copper recovery appears not to have reached a pla-
teau after 20 min otation. The step in the copperrecovery line at 8 min is due to the addition of extra
collector. Copper sulphate was added at the 16-min
mark of the oat and the shape of the recovery/time plot
ent (%
W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504 499Table 3
Metallurgical results for otation tests on pilot plant slags
Test no. Slag type Time (min) Compon
Cu
SL28 Self-pulverising 20 Ab 16.0
Rb 73.5
Fb 5.29
SL29 Self-pulverising 20 A 20.5
R 64.5
F 5.19
SL31 Escondida 16 A 29.3
R 80.2
F 9.71would seemingly indicate that this reagent has had little
eect in activating further copper for otation at this
stage of the oat.
In the test on as received slag, without grinding
(SL28), the total copper recovery was 73.5% at 16.0%
Cu (total recovery is recovery by otation and also from
metallic copper recovered by wet screening prior to
otation). This represents an enrichment ratio for cop-per (ratio of concentrate copper grade to feed copper
grade) of 3.02. By contrast, in the test on ground slag
(SL29), the total copper recovery was 64.5% at 20.5%
Cu, representing an enrichment ratio of 3.95. At rst
glance this result seems unusual in that grinding was
expected to liberate more copper into oatable size
SL32 Low copper 16 A 31.1
R 86.5
F 8.49
SL33 High copper 16 A 33.2
R 81.2
F 14.7
SL34 High copper 20 A 32.6
R 81.5
F 14.8
aCu enrichment ratio concentrate grade (%Cu)/otation feed grade (%CbA, assay; R, recovery (includes contributions from otation and wet scr
0102030405060708090
100
0 2 4 6 8 10 12 14 16 18 20Flotation time (min)
Cum
ulat
ive
reco
very
(%)
CuFeMgCa
Fig. 1. Metal recovery as a function of otation time for test SL28
(self-pulverising pilot plant slag, as received).ranges and yield a higher copper recovery. The nal
concentrate grade was higher in the test on the ground
slag and in fact at comparable copper recoveries
throughout the oats the copper grade of the concen-
) Cu enrichment ratioa
Fe Mg Ca
35.4 3.44 10.1 3.02
25.7 22.9 16.5
33.4 3.64 14.9
26.3 2.79 13.2 3.95
12.7 12.4 13.9
33.7 3.68 15.6
25.5 1.94 10.8 3.02
19.6 19.2 21.9
31.9 2.48 12.1
24.6 2.68 12.5 3.66
16.6 16.6 19.2
32.9 3.60 14.5
22.4 2.49 11.3 2.26
26.1 22.3 30.1
29.7 3.85 13.0
23.3 2.58 12.2 2.20
28.2 24.4 33.2
30.6 3.92 13.6
u).
eening); F, feed (calculated).trate is always higher in the test after grinding. This
suggests that in spite of the lower overall copper
recovery the liberation of copper-bearing particles hasimproved after grinding for this slag.
The copper recovery from screening was signicantly
lower in the test after grinding; only 2.5% compared
with 8.8% for the as received slag, reecting the ner size
distribution. In addition, the total copper recovery in the
otation stage decreased after grinding from 71% to
64%. These results suggest that using a ner screen in the
test after grinding might have yielded a higher totalcopper recovery. To verify this the otation tail from
test SL29 was sized and each fraction assayed. The
distribution data for copper is presented graphically in
Fig. 2 and this shows that two-thirds of the copper is
contained in particles coarser than 32 lm and nearlyone-third in particles coarser than 53 lm.
The grade of the +53 lm fraction (containing 31.8%of the copper in the tail) was 16.6% Cu while the gradeof the +75 lm fraction (containing 6.13% of the copper)was 29.5% Cu. Recall the grade of the combined ota-
tion (rougher/scavenger) concentrate for this test was
20.5% Cu. Combining the coarse material (+53 lmfraction) of the tail with the rougher/scavenger concen-
trate (in essence the equivalent to reducing the pre-
otation screen size from 210 to 53 lm) increases the
Another possibility relates to otation retention time.
The shape of the recovery-time plots for both tests
indicates copper recovery has not reached a plateau
after 20 min. One interpretation could be that the rela-
tively coarse metallic copper is slow oating.
Insucient collector levels may be another possible
reason for poor oatability of metallic copper. Thispossibility was investigated further by considering the
collector (xanthate) level in the otation pulp as a
function of otation time for both otation tests. The
collector concentration/time plot for test SL28 (as re-
ceived slag) is presented as Fig. 3. Reagent addition
levels and times are noted on the graph.
The three major peaks shown clearly in Fig. 3 after 0,
15 and 30 min relate to the three sequential collectoradditions of 100, 50 and 50 g/t KEX respectively. The
rate of disappearance of xanthate from solution is sim-
500 W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504overall copper recovery to 75.8% at a grade of 19.8%
Cu. This result is better than the overall result from
SL28 (73.5% recovery at 16.0% Cu), the comparative
oat on unground slag, in respect of both copperrecovery and grade.
While the above calculations help explain some of the
dierences between the two tests on the self-pulverising
slag and indicate how it might be feasible to lift total
copper recovery above 75% by optimising the screen size
after grinding, the more important question is why the
copper liberated by grinding into a oatable size range
did not oat as well as expected. Copper recoveries ex-ceeded 90% in previous otation tests on self-pulverising
slags made in laboratory smelting trials from the same
feedstock (Bruckard et al., 2003). Given that 99% of the
copper in the self-pulverising slag used in this work was
metallic copper it follows that the copper losses in the
0
10
20
30
40
50
0 10 20 30 40 50 60 70 80 90 100Particle size (microns)
Cop
per d
istri
butio
n (%
)
Fig. 2. Distribution of Cu as a function of particle size in otation tail
from test SL29 (self-pulverising pilot plant slag).otation tails will be predominantly metallic copper.
Possible reasons for the lower recoveries obtained could
include: the shape of the metallic copper particles,insucient otation time, or insucient collector or
sulphidising agent. Some of these points are discussed
below.
Spherical prills of metallic copper would be expected
to be harder or slower to oat than akes or attened
copper metal particles given there is greater scope for air
bubbles to make contact with akes or attened parti-
cles. Generally grinding has the eect of atteningspherical copper metallics and so this explanation would
seem to be at odds with the present metallurgical results.
However, the grind time for the self-pulverising slag was
short (10 min compared with 2232 min for the other
slag types) due to the already ne size of the as received
self-pulverising slag. The limited grinding may have had
an impact on the extent to which any metallic prills
present were attened. Some optical or SEM studieswould be needed to verify shape eects in this system.ilar after each addition and arises from a combination of
collector adsorption onto mineral surfaces, decomposi-
tion or oxidation of the xanthate in solution, and dilu-
tion by makeup water added during the otation stages.
The data suggests that residual xanthate levels are suf-
ciently high to rule out the possibility that low collectorlevels may have adversely inuenced the copper otation
performance. The small rise and fall in xanthate level at
the 26-min mark is due to the addition of sodium sul-
phide. When the potential of the otation pulp is re-
duced by the addition of a reductant like sodium
sulphide some xanthate from the mineral surface is re-
leased back into solution and then re-adsorbed as the
potential increases when air is turned back on in the cellfor otation. The xanthate concentration/time data for
the otation test on ground self-pulverising slag (SL29)
is not shown but is similar to that for test SL28.
Only KEX collector was used in this study. Longer
chain xanthates or other stronger collectors could also
be considered in further work to help improve the o-
atability of coarse metallic copper.
0123456789
10
0 5 10 15 20 25 30 35 40Time (min)
KEX
conc
entra
tion
(ppm
)
500 g/t Na2S100 g/t KEX 50 g/t KEX
300 g/tNa2S
100 g/t CuSO4 50 g/t KEX
Fig. 3. KEX concentration as a function of otation time for otation
test SL28 (self-pulverising pilot plant slag). Note KEX additions of
100, 50, and 50 g/t made at 0, 15, and 30 min respectively. Flotation
periods are 614, 1725 and 3236 min.
3.2.2. Low copper, high copper and Escondida slags
Samples of crushed Escondida, low copper and high
copper pilot plant slags were each ground for dierent
times to determine the grind time required to achieve a
otation feed P80 of about 40 lm. Flotation tests werethen conducted on each slag type (Escondida, SL31; low
copper, SL32; and high copper, SL33, SL34). Table 2lists the test conditions and reagent additions while the
overall metallurgical results for each test are given in
Table 3. Recovery-time plots for each oat are shown in
Figs. 46 respectively.
The results in Table 3 and Figs. 46 show that after
16 min otation total copper recoveries of 80.2%, 86.5%
and 81.2% were obtained for the Escondida, low copper
and high copper pilot plant slags respectively, at gradesof 29.3%, 31.1% and 33.2% Cu. The enrichment ratios
obtained for copper were 3.02, 3.66 and 2.26 respec-
tively. The lowest enrichment values were obtained for
the tests on the high copper slag. In general the selec-
tivity of copper over the gangue elements for each slag is
reasonably good and as with the tests on the self-
puverising slag the copper recovery appears not to have
reached a plateau after 16 min otation in each test.
The data in Table 4 indicates that in all three tails the
distributions of calcium, iron and magnesium largely
follow the weight distribution. The copper distribution
for the high copper slag tail also follows the weight
distribution while for the low copper and Escondida slag
tails there is proportionally more copper at the coarseend. This is most likely due to the presence of coarser
metallic copper in the low copper and Escondida slag
tails, which has not oated. That there may be less
coarse metallics in the high copper oat tail may relate
to the higher levels of oxide copper (cuprite and dela-
fossite) in this slag. Certainly the unoated copper in all
three tails appears to lie in a oatable size range (
from
%)
C
1
1
100.0 100.0 100.0 100.0
eening stage at 75 lm.
0.1
1.0
0.1 1.0Fraction of iron unfloated
Frac
tion
of c
oppe
r unf
loat
edNo SelectivitySL33 High copperSL31 Escondida
SL32 Low copperSL29 Self-pulverising
502 W.J. Bruckard et al. / Minerals Engineering 17 (2004) 4955043.3. Selectivity comparison
The comparative copper otation behaviour with
respect to the (pilot plant) slag type was considered in
terms of the overall selectivity of copper over the main
gangue element, iron, in the otation tests conducted. A
selectivity plot for four of the otation tests undertaken
is presented in Fig. 7. In this plot, points on a line with a
slope of 1 indicate otation performance where copperand iron are recovered equally, that is, no selectivity.
Plots below this line are indicative of otation where
copper has been selectively recovered relative to iron.
Points closest to the bottom right-hand corner indicate
the most selective otation results.
Table 4
Elemental distributions as a function of particle size for otation tails
Test no. Slag type Size range (lm) Distribution (
Wt
SL31 Escondida +53a 7.18
)53+38 16.0)38 76.9
Total 100.0
SL32 Low copper +53a 6.20
)53+38 17.4)38 76.4
Total 100.0
SL33 High copper +53a 4.29
)53+38 17.2)38 78.5
Total 100.0
a This size fraction is actually )75+53 lm given the pre-otation scrThe selectivity results in Fig. 7 indicate no major
dierences in copper/iron selectivity for the otation
tests on the four slags. Overall the low copper slag givesthe most selective result and the least selective result
comes from the high copper slag. This is generally
consistent with the kinetic data obtained in the otation
tests and the given copper mineralogy for the dierent
slags types.
3.4. Comparison of otation behaviour between labora-
tory and pilot plant slags
In a previous study (Bruckard et al., 2003), batch
otation tests were conducted on continuous converting
slags made in batch laboratory smelting tests using
copper sulphide concentrates from MIM. This copper
concentrate was the same material used as feedstock for
the continuous pilot plant smelting trials that produced
the pilot plant slags used in the tests discussed in thepresent work. Given the otation test conditions used in
each test program were similar, it is worthwhile com-tests on Escondida, low copper, and high copper pilot plant slags
u Fe Mg Ca
10.7 7.37 7.39 6.37
20.7 16.5 16.4 14.4
68.6 76.1 76.3 79.2
00.0 100.0 100.0 100.0
12.0 6.43 6.35 5.62
26.3 18.8 18.2 14.5
61.7 74.8 75.5 79.9
00.0 100.0 100.0 100.0
4.74 4.85 4.80 3.19
16.9 18.9 18.9 13.4
78.4 76.3 76.3 83.4paring the otation behaviour between the laboratory
slags and the pilot plant slags, for each slag type, pro-
duced from the MIM copper concentrate. The otationresponse of the laboratory and pilot plant slags by slag
type is compared in Fig. 8, a grade-recovery plot for
copper. Note this data represents the otation response
only for each slag and any additional copper recovery by
pre-otation screening is not shown on these curves.
It can be seen from Fig. 8 that the copper grade-
recovery curves of the laboratory slags, for the three slag
types considered (low copper, high copper, and self-pulverising), essentially lie on the same extrapolated
lines as those for the respective pilot plant slags. There
were minor dierences in the head grades of the relative
slags (8.20%, 13.8% and 2.87% Cu for the low copper,
high copper and self-pulverising laboratory slags
respectively, and 7.87%, 15.10% and 5.24% Cu for the
Fig. 7. Copper/iron selectivity plot for otation tests on pilot plant
slags.
have a steeper gradient than that for the other two slags
types.
obtained in the otation tests and the given copper
mineralogy for the dierent slags types. The more oxi-
analysis of otation products, and Ms. Nicki Agron-
Olshina for conducting XRD analysis. Financial sup-
W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504 503In spite of the similarities in the extrapolated curves,
that nal copper recoveries were dierent for each
comparable slag type for the laboratory and pilot plant
slags suggests that further work is still needed to identify
the reasons for dierences in the copper otation re-
sponse that arise when the smelting step is scaled upfrom batch laboratory to continuous pilot plant level.
On the basis of the evidence presented in this work it
seems that understanding more fully the otation
behaviour of metallic copper in these systems should be
the focus of any such studies.
4. Conclusions
A series of batch laboratory otation tests was con-respective pilot plant slags) and these dierences will
also inuence the grade-recovery response to some ex-
tent. It is interesting to note that the low copper curves
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100Cumulative copper recovery (%)
Cum
ulat
ive
copp
er g
rade
(%)
Low copper laboratory slag
High copper laboratory slagSelf-pulverising laboratory slagSelf-pulverising pilot plant slag
Low copper pilot plant slagHigh copper pilot plant slag
Fig. 8. Cumulative copper grade as a function of cumulative copper
recovery for otation tests on laboratory and pilot plant slags.ducted on a suite of dierent calcium-ferrite slags made
in continuous pilot plant smelting trials. The slags in-
cluded a moderately reduced low copper slag, a more
oxidised high copper slag, a very reduced low copper
self-pulverising slag, and a moderately reduced low
copper Escondida slag. The slags varied in copper con-tent from 5.2% to 15.1% Cu. The dominant form of
copper in all the slags was metallic copper. A simple
grinding/screening/otation procedure was used in all
tests with reagent additions typically 150200 g/t KEX
and 500800 g/t Na2S 9H2O.In rougher/scavenger otation tests on the pilot plant
slags total copper recoveries between 80% and 87% were
obtained for three of the slag types tested (low copper,high copper and Escondida). The best result for the
forth slag (self-pulverising) was 74% copper recovery.
Coarse copper metal was present in the otation tails for
this test suggesting the pre-otation screen size was notport for this work was provided by the Australian
Government Cooperative Research Centre program
through the former G.K. Williams Cooperative Re-
search Centre for Extractive Metallurgy, a joint venture
between the CSIRO Division of Minerals and The
University of Melbourne Department of ChemicalEngineering.
References
Bruckard, W.J., Somerville, M., Heyes, G.W., 2003. Flotation
recovery of copper from calcium-ferrite-based slags produced in
laboratory smelting trials. In: Lorenzen, L., Bradshaw, D.J. (Eds.),
Proc. XXII Int. Min. Process. Cong., Cape Town, South Africa,
South African Institute of Mining and Metallurgy, vol. 2, pp. 985
995.
Eerola, H., Jylha, K., Taskinen, P., 1984. Thermodynamics of
impurities in calcium ferrite slags in copper re-rening conditions.
Trans. Inst. Min. Metall., (Sect. C: Miner. Process. Extr. Metall.)
93, C193C199.
Guy, P.J., 1992. The development of laboratory batch otation
equipment and practice at CSIRO Australia. In: The AusIMM
Annual Conference, Broken Hill, The Australasian Institute of
Mining and Metallurgy, Melbourne, pp. 7579.dised the slag, which is normally associated with a
higher total copper content, the greater the proportion
of copper present as copper oxide or copper spinel/fer-
rite phases, rather than as metallic copper. The spinel/
ferrite phases would seem to be slower oating and
given they contain both copper and iron, selective cop-
per recovery is compromised by their presence.
Acknowledgements
The authors wish to acknowledge Mr. Peter Guy
and Mr. Graeme Heyes for assistance in conducting
preliminary otation testing, Mr. Mick Manuele for
preparing samples for analysis, the sta of the Ana-
lytical Section of CSIRO Minerals for completing theoptimal. It is expected that an improved copper recovery
for this slag might be possible with an extended grind
time, a ner pre-otation screen size and possibly a
longer oat. The scope for improving the copper
recovery for the other slag types to above the target 90%
mark would seem to lie in optimising the otation
conditions to improve the oatability of relatively coarsemetallic copper present.
Minor dierences in copper/iron selectivity for the
otation tests on the four slags were observed with
the low copper slag giving the most selective result and
the high copper slag the least selective result. These
trends were generally consistent with the kinetic data
Inoue, R., Suito, H., 1991. Siliconoxygen equilibrium and nitrogen
distribution between CaOSiO2 slags and liquid iron. Metall.
Trans. B 23B, 613621.
Jahanshahi, S., Wright, S., Somerville, M., 1994. Continuous convert-
ing of copper matte using calcium ferrite slags. In: Proceedings of
the 6th AusIMM Extractive Metallurgy Conference, The Austral-
asian Institute of Mining and Metallurgy, Melbourne, pp. 8994.
Jahanshahi, S., Somerville, M., Hollis, R.G., 1995. Direct converting
of copper concentrate in Sirosmelt reactors. In: Proceedings of
Copper 95Cobre 95 International Conference, The Metallurgical
Society of the Canadian Institute of Mining and Metallurgy, vol.
IV, pp. 367381.
Juckes, L.M., 2002. Dicalcium silicate in blast-furnace slag: a critical
review of the implications for aggregate assembly. Trans. Inst. Min.
Metall., (Sect. C: Mineral Process. Extr. Metall.) 111, C120C128.
Kelsall, D.F., Restarick, C.J., Stewart, P.S.B., 1974. Technical note on
an improved cyclosizing technique. In: Proc. Australas. Inst. Min.
Metall., No. 251, pp. 910.
Light, T.S., 1972. Standard solution for redox potential measurements.
Anal. Chem. 48, 10381039.
Somerville, M., Norgate, T., Jeeries, P., Vecchio-Sadus, A., Jahan-
shahi, S., 1995. Single stage copper makingowsheet develop-
ment. In: Proceedings of Copper 95Cobre 95 International
Conference, The Metallurgical Society of the Canadian Institute
of Mining and Metallurgy, vol. VI, pp. 1533.
Somerville, M., Norgate, T., Jahanshahi, S., 2000. Single stage copper
makingassessment of slag treatment options. In: Proceedings
Minprex 2000, International Conference on Mineral Processing
and Extractive Metallurgy, The Australasian Institute of Mining
and Metallurgy, Melbourne, pp. 453459.
504 W.J. Bruckard et al. / Minerals Engineering 17 (2004) 495504
The recovery of copper, by flotation, from calcium-ferrite-based slags made in continuous pilot plant smelting trialsIntroductionExperimental proceduresSlag samplesCrushing and blendingGrinding and screeningFlotationFlotation equipmentFlotation reagentsFlotation procedures
Chemical analysis of solid productsOn-line solution analysisSizing analysesX-ray diffraction analyses
Results and discussionXRD analyses of slagsMetallurgical resultsSelf-pulverising slagLow copper, high copper and Escondida slags
Selectivity comparisonComparison of flotation behaviour between laboratory and pilot plant slags
ConclusionsAcknowledgementsReferences