Los Angeles, London, New Delhi

and Singapore
http://www.sagepub.com © The Author(s), 2010. Reprints and permissions:
http://www.sagepub.co.uk/journalsPermissions.nav
ISSN 0734–242X
Waste Management & Research
2010: 28: 561–567
DOI: 10.1177/0734242X09343943

Characterization and recovery of copper values from

discarded slag
Bisweswar Das, Barada Kanta Mishra, Shivakumar Angadi, Siddharth Kumar Pradhan,
Sandur Prakash, Jayakrushna Mohanty
Institute of Minerals and Materials Technology, Council of Scientific and Industrial Research (CSIR), Bhubaneswar, India

In any copper smelter large quantities of copper slag are discarded as waste material causing space and environmental problems.
This discarded slag contains important amounts of metallic values such as copper and iron. The recovery of copper values from
an Indian smelter slag that contains 1.53% Cu, 39.8% Fe and 34.65% SiO2 was the focus of the present study. A complete inves-
tigation of the different phases present in the slag has been carried out by means of optical microscopy, Raman spectroscopy,
scanning electron microscopy (SEM), and X-ray diffraction (XRD) techniques. It is observed that iron and silica are mostly
associated with the fayalite phase whereas copper is present in both oxide and sulfide phases. These oxide and sulfide phases
of copper are mostly present within the slag phase and to some extent the slag is also embedded inside the oxide and sulfide
phases. The recovery of copper values from the discarded slag has been explored by applying a flotation technique using con-
ventional sodium isopropyl xanthate (SIX) as the collector. The effects of flotation parameters such as pH and collector con-
centration are investigated. Under optimum flotation conditions, it is possible to achieve 21% Cu with more than 80% recovery.

Keywords: copper slag, ore mineralogy, Raman spectroscopy, scanning electron microscopy, X-ray diffraction, slag flotation

Introduction
It has been estimated that for each ton of copper produced
about 2.2 tons of slag are generated. These slags contain
about 0.7–2.1% Cu (Gorai et al. 2003) and are discarded as a
waste product. Today the demand is such that even discarded
slag with about 1% Cu is considered to be a valuable
resource. Although a large quantity of copper lies in the dis-
carded slag, it is not possible to recover the metal values by
straightforward means. The difficulty in processing the slag is
directly linked to the association of the copper values with
other metals and compounds present in the slag. In this con-
text, the froth flotation technique has been identified as the
most economical option to recover copper values from slag
(Bota et al. 1995, Bruckard et al. 2004). In the Black Sea Cop-
per Works, Turkey, about 100 000 tons of slag are treated
annually by the flotation process, after being crushed and
ground, to recover additional copper values (Arslan & Ars-
lan 2002). However, it has been estimated that a copper
recovery higher than 80% is required to make such a process
economically feasible. The success of the process is depend-
ent on the presence of copper values either as sulfide or
oxide, the grindability of the slag and the liberation of cop-
per-bearing phases, among other factors.
The discarded slag coming from various sources has been
characterized by many researchers and reported in the litera-
ture. Two of the most recent studies relating to the character-
ization of discarded copper slag showed that copper was
present as Cu2O, Cu2S, CuO and free copper in different
amounts, depending on the slag chemistry (Sarrafi et al.
2004, Mihovsky et al. 2006). Another study carried out by
quantitative electron microprobe analysis of copper slag
indicated the presence of fine copper matte, fayalite and iron
particles in a silicate matrix (Imris et al. 2000). One interest-
ing X-ray diffraction (XRD) study of the slag showed the
presence of fayalite, magnetite and the complex matrix
phases among the other sulfide and oxide phases of copper
(Kiyak et al. 1999). Herreros et al. (1998) reported that most
of the copper was present in the silicate matrix of the slag.
Other phases such as magnetite, crystallite–fayalite, glassy
phases with high silica and low iron content and spherical
particles consisting of Cu2S and PbS are also present in cop-

Corresponding author: B. Das, Institute of Minerals and Materials Technology, Council of Scientific and Industrial Research (CSIR), Bhubaneswar-
751013, India.
E-mail: bdas@immt.res.in
Received 17 March 2009; accepted in revised form 3 July 2009

561
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015
B. Das, B.K. Mishra, S. Angadi, S.K. Pradhan, S. Prakash, J.K. Mohanty
per slag. In general, the discarded copper slag is a complex
material mainly composed of sulfides, oxides, silicates, and
some free metallic copper and iron.
Most of the aforementioned characterization studies have
concentrated on identifying different phases to understand
their distribution, the phase morphology and copper slag
chemistry in relation to flotation. In those situations the slag
was typically a slowly cooled product of the smelter. In con-
trast, the slag on which the present study was focused was a
granulated slag, which was obtained by quenching the mol-
ten slag under a high pressure water jet, and in which the
copper-bearing phases are distributed at very fine size. The
basic objective of this investigation therefore was to deter-
mine the morphology and distribution of the various phases
in the slag for a further flotation process to obtain a copper
concentrate. To meet the study objective, the discarded cop-
per slag from a typical Indian copper plant was characterized
by different techniques with a view to the recovery of addi-
tional copper values.
Materials and methods
Copper slag origin and main composition
A batch of granulated copper slag obtained from a copper
plant in India was sampled in duplicate for the characterization
and beneficiation studies. The physical properties and chemi-
cal composition of the slag are given in Table 1. It contained
around 1.53% Cu, 38.53% Fe, 30.92% SiO2 and 1.24% S. The
physical properties such as the bulk density and specific grav-
ity of the sample were found to be similar to those of other
discarded copper slags found elsewhere (Shen & Forssberg
2003).
The Bond work index of the bulk slag sample was deter-
mined by using the standard grindability test. The feed mate-
rial for the test was prepared by passing the sample through a
3.35 mm sieve. The grinding was performed in stages and at the
end of each grinding cycle the entire product was discharged
from the mill and screened on a 0.15 mm test sieve. Fresh feed
Table 1: Physico-chemical characteristics of copper slag.
Parameter
Bulk density (g cm–3)
Specific gravity
Cu (%)
FeO (%)
Fe2O3 (%)
SiO2 (%) 30.91
Al2O3 (%)
CaO (%)
MgO (%)
Co (%)
Ni (%)
Mn (%)
S (%)
562
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015
material was added to the oversize to bring the total weight
back to that of the original weight of the feed. This material
was then put in the mill and further ground. After each cycle,
the same procedure of screening and grinding was followed
until the undersize per mill revolution became constant for the
last three grinding cycles. The average weight of the undersize
per revolution of the last three tests and the 80% passing size
of feed and product (0.150 mm) were determined and used
to calculate the work index (Wi) (see equation (1)).
44.5
W i = ---------------------------------------------------------------------------
- (1)
 0.23 0.82 10 10  
 ( p i ) × ( Gbp )  ------ – ------ 
  d p df  
where pi is the mesh size, Gbp is the amount of undersize in
grams per revolution, dp and df are the 80% passing size in
micrometres of the product and feed, respectively.
Characterization of copper slag
Optical microscopy was carried out to determine metal/slag
distribution and identify textural features. The liberation
characteristic of copper slag by the point counting method
was studied for different size fractions and XRD studies
were carried out for mineral phase identification. In addi-
tion, polished samples were analysed for identification of dif-
ferent copper, iron and silica phases in a Renishaw in Via
micro-Raman spectrometer equipped with a 514 nm green
laser, having 2 cm–1 spectral resolution of Raman shift with an
X–Y step resolution of 0.1 µm and confocal resolution of
2.5 µm. Finally, scanning electron microscopy–energy dis-
persive spectroscopy (SEM–EDS) studies were carried out
using a Hitachi 3400N equipment to analyse the elemental
distribution in different phases present in the copper slag
through point/area analysis to get an idea about the elemen-
tal composition of the different phases in a semi-quantitative
manner.
Copper slag flotation studies
A representative slag sample was taken and ground to below
Value 100 µm size at which all the flotation experiments were car-
ried out. The grinding of the material was carried out using a
2.4
ball charge according to the standard Bond method at 45%
3.49
ball charge and 250% circulating load using a check screen at
1.53
100 µm. A Denver D-12 sub-aeration flotation cell of 2 L
0.71 capacity was used for batch flotation studies. The sample was
55.1 conditioned at 60% pulp density with the desired amounts of
reagents for a predetermined time. The pulp density was
2.3 diluted to ~25% by addition of water and further conditioning
3.95 by addition of a frother. The agitation intensity (1400 rpm),
2.82 slurry level and pH were maintained during the experiments.
In the flotation experiments, freshly prepared sodium iso-
0.04
propyl xanthate was used as the collector, methyl isobutyl
0.02
carbinol (MIBC) as the frother and sodium silicate as the sil-
0.34
ica depressant. In all the experiments 75 g ton–1 of frother
1.24 and 200 g ton–1 of sodium silicate were used, and the concen-
 Characterization and recovery of copper values from discarded slag

tration of the collector varied between 200 and 900 g ton–1. that the maximum amount of metal values was found at
Dilute sodium hydroxide and hydrochloric acid of analytical 100 µm particle size.
grade were used whenever necessary as pH regulators. The
concentrates and tailings of the flotation process were col- Raman spectroscopy
lected separately, dried, weighed, and analysed for copper to Micro-Raman analysis of the copper slag sample indicating
assess product quality and recovery. The kinetic studies were different phases is shown in Figure 2. The grey phases (A
carried out by collecting the concentrate at cumulative time and B) are mostly glass like, consisting of silicon and iron
intervals of 30, 60, 90, 120, 150, 180 and 300 s. The flotation oxides, whereas the bright phases (C and D) are copper-rich
data was used to study the kinetics by means of a first-order oxides and sulfides. The bright phases (C and D) contained
rate model (see equation (2)) (Lynch et al. 1981, Wills 1990). sulfide of copper and iron and small amounts of iron oxide.
Analysis of the Raman spectra indicated that the peak at
R = R∞[1 – exp(–kt)] (2) ~220 cm–1 corresponds to pyrrhotite/cuprite and the peak at
285 cm–1 corresponds to chalcopyrite, magnetite or hematite
where R is the mineral recovery (in %) at any time interval t, (Perez-Robles et al. 1999). The peaks around 400 and 500 cm–1
R∞ is the maximum recovery percentage of the mineral at can be assigned to covellite, chalcopyrite, iron sulfide, pyrite
infinite time t∞ and k is the rate constant expressed in s–1. or bornite, whereas the peaks at 1300 and 1580 cm–1 can be
clearly assigned to pyrrhotite and magnetite, respectively. In
Results and discussion addition to the usual darker glassy and brighter sulfide phases,
Optical microscopic study other phases such as graphite and pure copper particles were
The discarded copper slag contained various sizes of copper- also observed.
bearing phases, i.e., Cu2O, Cu2S, CuO and free copper. A slag particle which has a distinct core and shell with cor-
These phases were mostly rounded indicating solidification responding Raman spectra is shown in Figure 3. From the
from a molten state and were mostly present within the slag Raman data it can be inferred that the shell is copper-rich,
phase and sometimes as a rim completely encircling the slag. whereas the core is deficient in copper. In the Raman study
In some cases globules of slag were also seen embedded pure metal phase particles were also detected in the micro-
inside the metal (Figures 1(a) and (b)). The liberation char- structure but there was no discernable Raman peak from the
acteristic of copper slag is shown in Table 2. It is observed particles which clearly proved that those pure phases are
either copper or iron particles.
Table 2: Liberation study of the copper slag.
SEM and EDS analysis
Size (µm) Slag (%) Free metal (%) Metal + slag (%) The SEM of copper slag is shown in Figure 4. The bright
Bulk 92.17 1.45 6.38 grey phase (b) as observed in the microphotograph is the slag
+2000 90 1.50 8.50 phase whereas the white phase ((a), (c)) is the copper-bear-
–2000 + 1000 91.50 1.78 6.72 ing phase which contains iron and copper in association with
–1000 + 500 91.76 1.12 7.12
sulfur and oxygen. It was observed that the copper-bearing
phases were not uniform in composition and were often sur-
–500 + 100 91.95 4.46 3.59
rounded by the slag phase. The variation in composition
–100 +50 89.13 10.23 0.64
was due to the segregation during solidification and the
–50 90 3.76 6.24 presence of inclusions. So, from a separation standpoint,

Fig. 1: (a) Microphotograph showing copper metal (white) within slag (grey); (b) sharp boundary between slag phase and copper metal phase,
Parallel Nicols, X100.

563
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015
B. Das, B.K. Mishra, S. Angadi, S.K. Pradhan, S. Prakash, J.K. Mohanty

Fig. 2: Photograph showing different phases present in copper slag sample with corresponding Raman spectra (A and B, slag phases; C and D,
copper-rich oxides and sulfides).

Fig. 3: Micro Raman analysis of the metallic phase: (a) core-copper deficient phase; (b) shell-copper rich phase (left); Raman spectra (right).

fine size grinding would be needed to liberate the copper- and O with minor amounts of S, whereas the copper bearing
bearing phases. phase was composed of Cu, Fe, S and O with very small
A few selective points/area were also traced for semi- amounts of Si. The comparison of both phases proved that a
quantitative analysis leading to chemical composition of the higher oxygen and iron concentration was present in the slag
slag and copper-bearing phases in the sample. The elemental phase.
composition of these phases and the corresponding spectra is
shown in Figure 4((a), (c)) and (b) and summarized in Table 3. Flotation studies
It was observed that copper slag mainly consisted of Fe, Si Copper slag flotation studies were carried out on the ground
sample (d80 ~ 75 µm) to determine the response with regard
to the recovery of copper values. During grinding it was
Table 3: Semi-quantitative analysis of metal and slag phases
(selected points) by SEM. observed that the granulated slag sample was difficult to grind.
It took almost 45 min to grind the sample to below 100 µm
Weight (%) size. This may be due to the presence of glassy phases typical
Reference points
O Si S Fe Cu of slag or due to the presence of metallic phases that plasti-
cally deformed during grinding. Whatever may be the case,
(a) Metal phase 44.70 1.00 22.91 17.49 13.90
the grindability index was found to be 26.8 kW h ton–1, which
(b) Slag phase 40.85 18.78 1.98 38.39 0.00
was quite high in comparison with standard ores and miner-
(c) Metal phase 38.59 0.72 18.76 14.52 27.41 als. For example, the grindability index of copper and lead

564
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015

Fig. 4: SEM of copper slag with corresponding spectral analysis.
Points a and c; metallic phase, b: slag phases.
sulfide ore were only 13.13 and 10.19 kW h ton–1, respec-
tively (Gupta & Yan 2006).
The ground material was floated and the recovery as a
function of pH is shown in Figure 5. It must be stated that
pH is an important parameter in the context of flotation as it
determines the surface characteristics of different minerals
present in the pulp. Sulfide minerals flotation is enhanced at
alkaline pH in the presence of sodium isopropyl xanthate
(SIX), as xanthate molecules decompose at highly acidic pH
and iron sulfide minerals are depressed in alkaline pH.
Therefore, in alkaline pH good selectivity is obtained for
copper sulfide minerals. Several other tests were carried out
by varying pH from 8.0 to 10.0 at a constant SIX concentra-
tion of 200 g ton–1. The results indicate that it is possible to
float the copper values in the entire pH range (i.e., between
8 and 10). Copper grade of 11 to 17% in the concentrates
with 66 to 82% recovery yield was achieved. However, the
best floatability of copper was found at pH 9.
The effect of xanthate concentration in the flotation of
copper slag at a pH value of 9 is shown in Figure 6. It was
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015
Characterization and recovery of copper values from discarded slag
Fig. 5: Flotation of copper slag as a function of pH.
Fig. 6: Flotation of copper slag as a function of collector concentration
at pH 9.
observed that the grade of copper was higher at lower xant-
hate concentrations. Froth weight and copper recovery yield
increased with increasing reagent concentration. Overall it
was possible to recover 71 to 94% of copper values with a
concentrate copper grade of 15.26 to 23.53%. The highest
concentrate copper grade of 23.5% with 93% recovery was
achieved at a xanthate concentration of 600 g ton–1. It must
be pointed out that due to the complex nature of the slag,
particularly in terms of the distribution of various copper-
bearing phases, it was difficult to attain consistency in the
feed grade prior to each experiment.
The flotation kinetics of the copper slag at pH 9.0 and
SIX concentration of 600 g ton–1 is shown in Figure 7. The
results clearly show that the flotation followed first-order
kinetics. From the experimental results, the recovery of cop-
per is plotted against flotation time. It was observed from
Figure 7(a) that as the flotation time increased the recovery
also increased, but the rate of increase decreased with pro-
longed flotation time and eventually remained constant. This
565
B. Das, B.K. Mishra, S. Angadi, S.K. Pradhan, S. Prakash, J.K. Mohanty

Fig. 7: (a) Kinetic behaviour of copper slag flotation (time-recovery plot); (b) flotation follows first-order kinetics.

Fig. 8: XRD pattern of flotation concentrate and feed showing the presence of bornite (B), chalcopyrite (CP) and fayalite (F).

kind of behaviour is well established and follows the first- Conclusions

order kinetic model (equation (1)) described earlier. Fur-
thermore, the order of the reaction could be confirmed by
plotting ln(R∞/R∞ – R) against the flotation time, t, which
shows a linear relationship (Figure 7(b)). The flotation rate
constant was found to be 0.0354 s–1, which is close to the typ-
ical rate constants associated with flotation of copper sulfide
ores (Kracht et al. 2005).
The XRD pattern of flotation concentrate and feed slag
are shown in Figure 8. The XRD pattern indicates that the
feed sample was mainly composed of fayalite (FeSiO3),
which can be identified by the presence of 2.4984, 1.7764 and
2.8194 Å peaks. The flotation concentrate contains bornite
and chalcopyrite (3.1548, 3.0244, 1.9279 and 1.8518 Å peaks)
besides some fayalite.
Various characterization studies undertaken for the copper
slag indicate that it mainly consisted of silica and iron oxides
with traces of sulfides. The results of the beneficiation stud-
ies undertaken for copper slag indicate that the sample was
amenable to beneficiation by flotation with sodium isopropyl
xanthate as the collector. The slag sample needed grinding to
a particle size of less than 100 µm where the liberation of
copper occurred as found in the mineralogical studies. It was
possible to obtain a copper grade of 21% with 80% recovery
at a xanthate concentration of 600 g ton–1 at slightly alkaline
pH. The copper flotation followed a first-order kinetic equa-
tion. Thus, copper values from discarded slags that have
been lying for years in different plants in India could be
recovered to bring some revenue to the facilities.

566
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015

References
Arslan, C. & Arslan, F. (2002) Recovery of copper, cobalt and zinc from
copper smelter and converter slags. Hydrometallurgy, 67, 1–7.
Bota, G., Gican, I. & Poputa, G. (1995) The beneficiation of the non-fer-
rous metals from secondary resources at regia autonoma a Cupru-
lui Deva-Romaina. In: Krstev, B. & Golomeov, B. (eds.): Proceed-
ings of 6th Balkan Mineral Processing Symposium, The Ministry of
Science in Government of Republic of Macedonia, 18–21 Sept.,
pp. 426–431. Ohrid, Macedonia.
Bruckard, W.J., Somerville, M. & Hao, F. (2004) The recovery of copper,
by flotation, from calcium-ferrite-based slags made in continuous
pilot smelting trials. Minerals Engineering, 17, 495–504.
Gorai, B., Jana, R.K. & Premchand (2003) Characterization and utiliza-
tion of copper slag- a review. Resources, Conservation and Recycling,
39, 299–313.
Gupta, A. & Yan, D.S. (2006) Mineral Processing Design and Operations.
Elsevier Publication, Oxford,UK.
Herreros, O., Quiroz, R., Manzano, E., Bou, C. & Vinals, J. (1998) Cop-
per extraction from reverberatory and flash furnace slags by chlo-
rine leaching. Hydrometallurgy, 49, 87–101.
Imris, I., Rebolledo, S., Sanchez, M., Castro, G., Achurra., G. & Hernan-
dez, F. (2000) The copper losses in the slags from the EL Teniente
process. Canadian Metallurgical Quarterly, 39, 281–289.
Kiyak, B., Ozer, A., Altundogan, S.H., Erdem, M. & Tumen, F. (1999) Cr
(VI) reduction in aqueous solution by using copper smelter slag.
Waste Management, 19, 333–338.
Downloaded from wmr.sagepub.com at NORTH CAROLINA STATE UNIV on April 18, 2015
Characterization and recovery of copper values from discarded slag
Kracht, W., Vallebuona, G. & Casali, A., (2005) Rate constant modeling for
batch flotation, as a function of gas dispersion properties. www.cap-
tura.uchile.cl/dspace/bitstream/2250/2345/1/Kracht_W.pdf (accessed
14 March 2009).
Lynch, A.J., Johnson N.W., Manlapig, E.V.& Thorne, C.G. (1981) Min-
eral and Coal Flotation Circuits and their Simulation. Elsevier Scien-
tific Publishing Co., Amsterdam.
Mihovsky, M., Hadzhiyski, V., Todorov, L. & Dimitrova, D. (2006)
Plasma-induction technology for reduction processing of copper
concentrate pellets. Journal of the University of Chemical Technology
and Metallurgy, 41, 55–60.
Perez-Robles, F., Garcıa-Rodrıguez, F.J., Jimenez-Sandoval, S. &
Gonzalez-Hernandez, J. (1999) Raman study of copper and iron
oxide particles embedded in a SiO2 matrix. Journal of Raman Spec-
troscopy, 30, 1099–1104.
Sarrafi, A. Rahmati, B., Hassani, H.R. & Shirazi, H.H.A. (2004) Recov-
ery of copper from reverberatory furnace slag by flotation. Minerals
Engineering, 17, 457–459.
Shen, H.T. & Forssberg, E. (2003) An overview of recovery of metals
from slags. Waste Management, 23, 933–949.
Wills, B.A. (1990) Mineral Processing Technology. Pergamon Press,
Oxford, UK.
567