Minerals Engineering 23 (2010) 1198–1206

Contents lists available at ScienceDirect

Minerals Engineering

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

Flotation of copper sulphides assisted by high intensity conditioning (HIC)and concentrate recirculation

Erico Tabosa, Jorge Rubio *

Laboratório de Tecnologia Mineral e Ambiental (LTM), Departamento de Engenharia de Minas, PPGEM, Universidade Federal do Rio Grande do Sul (UFRGS),Avenida Bento Gonçalves. 9500, CEP 91501-970, Porto Alegre, RS, Brazil

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

Article history:Received 19 May 2010Accepted 6 August 2010

Keywords:FlotationHigh intensity conditioningConcentrate recyclingCopper sulphide fines

0892-6875/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.mineng.2010.08.004

* Corresponding author. Tel.: +55 51 3308 9474; faE-mail addresses: erico.tabosa@ufrgs.br (E. Tabosa

This paper describes the effect of the partial concentrate (rougher floated product) recirculation torougher flotation feed, here named concentrate recirculation flotation – CRF, at laboratory scale. The mainparameters used to evaluate this alternative approach were flotation rate and recovery of fine (‘‘F” 40–13 lm) and ultrafine (‘‘UF” <13 lm) copper sulphide particles. Also, the comparative effect of high inten-sity conditioning (HIC), as a pre-flotation stage for the rougher flotation, was studied alone or combinedwith CRF. Results were evaluated through separation parameters, grade-recovery and flotation rates,especially in the fine and ultrafine fractions, a very old problem of processing by flotation. Results showedthat the floated concentrate recirculation enhanced the metallurgical recovery, grade and rate flotation ofcopper sulphides. The best results were obtained with concentrate recirculation flotation combined withhigh intensity conditioning (CRF–HIC). The kinetics rate values doubled, the Cu recovery increased 17%,the Cu grade increased 3.6% and the flotation rates were 2.4 times faster. These were accompanied byimproving 32% the ‘‘true” flotation values equivalent to 2.4 times lower the amount of entrained copperparticles. These results were explained and proved to proceed by particle aggregation (among others)occurring after HIC, assisted by the recycled floatable particles. This ‘‘artificial” increase in valuable min-eral grade (by the CR) resulted in higher collision probability between hydrophobic particles acting as‘‘seeds” or ‘‘carrier”.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

The poor flotation recoveries of fine (‘‘F” 40–13 lm) and ultra-fines (‘‘UF” < 13 lm) mineral particles is very well known andmany studies, projects and processes have been proposed, but thismajor technical and economic problem still continues (Collins andRead, 1971; Fuerstenau et al., 1979; Trahar, 1981; Sivamohan,1990; Song and Lu, 2000; Rubio et al., 2003, 2006). The mineralfines formation or generation is diverse, from many sources: someparticles are necessarily produced in the grinding process foradequate liberation of the particles; another fraction comes fromre-grinding and a small quantity from particles shearing andsmearing (in conditioners for example). In some cases, certain min-eral ores must be ground to very fine particles because of the de-crease in the grade and the increase of complex ores. Underactual conditions, a great deal of fine minerals is discarded asslimes prior to flotation for lack of effective recovery methods.

Many alternatives for solving this problem have been tried, butunfortunately most of them have not found practical applications(Rubio, 2002, 2006; Song and Lu, 2000). A good example is the case

ll rights reserved.

x: +55 51 3308 9477.), jrubio@ufrgs.br (J. Rubio).

of selective flocculation with polymers that, despite more than30 years of extensive studies and a few (mainly iron) plants usingthis technique, still remain a ‘‘promising” process (Fuerstenauet al., 1979; Subrahmanyam and Forssberg, 1990; Sivamohan,1990; Mathur et al., 2000).

Main problems associated to the fines flotation are caused bytheir intrinsic properties, among others, small mass (low momen-tum), high interfacial free energy. Thus, main flotation drawbacksare related to the low probability of bubble–particle collision andadhesion, mechanical entrainment and entrapment, ‘‘slime coat-ing”, higher flotation reagent adsorption, formation of dense frothsand low process kinetics. It is well known that ultrafine and coarseparticles float more slowly than particles in a mid-size range. Thesize limits are not precise but for many mineral systems, the ultra-fines are those below 5–10 lm, and the coarse sizes are above100 lm approximately. The flotation rate tends to decrease withdecreasing particle size, because of the increasing importance ofviscous and electrostatic forces, which are less important with lar-ger particles. The rate of bubble–particle collision is, for the coarserparticles higher than for the ultrafines, but the hydrodynamicforces associated with turbulence and shear in the pulp, lead toan increase in the rate of detachment. Thus as the particle size in-creases, it becomes increasingly difficult to maintain the capillary

E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206 1199

bonds between the particles and the bubbles and to withstandshearing, while being transported into the froth.

Most selective aggregation using chemicals has not proven to bea unique and general solution, the conventional or column type flo-tation machines do not generate the bubbles size distribution re-quired for an efficient capture of the very fines particles, neitherthe actual ‘‘conditioners” (merely pulp distributors) allow aggrega-tion of the fines particles, F–UF, themselves (Rubio et al., 2003,2006).

The concept of shear flocculation has been extending to the con-ditioning stage ahead of flotation (Bulatovic and Salter, 1989; Stas-sen, 1991). Many authors claim that energy transferred in the

Table 1Multifeed circuits reported on the literature.

Author Country/Mineral system

Comments

Firth et al. (1979) Australia/coal Investigation of feed s(staged flotation) werflotation. The reflotatiit also was the least a

Firth et al. (1978);

Tsai (1988) USA/coal Staged froth flotationrich particles are firstcareful control of frothsulfur contents has beother reported multismineral size and its dhave a volume mean

Fuerstenau (1988) USA/molybdenite

A brief discussion onpresented. Multifeed cImprovements in bothconservation of flotatiores of varying compoResults of an experimpresented to illustrateflotation roughers signper cent MoS2 while magent, that is used to

Hosten and Muratoglu (1996) Turkey/bituminouscoal

The possibility of impflotation scheme was icollected from the <0.batch tests simulatingoctanol) additions wealternative scheme fro

Fuerstenau et al. (2000) USA/coal andmolybdenite

Studies of coal and mflotation methods. Invtwo fractions with thestage. These circuits wwas produced with thLaboratory experimenresults in the flotationFor coal, studies showcontents comparableyields than conventioFor molybdenite flotathe single-stage roughconsumption. This aprougher concentrate

Rodrigues et al. (2005) Brazil/copper Investigation of a copcircuit changes were pand without staged reBest results were obtacombined with staged

Proyecto Flotación de finos y ultrafinos-Codelco-Chuquicamata – IM2 (2002),Tabosa (2007), Tabosa et al. (2009).

Brazil/chile/copper

Concentrate recirculatincludes one conventi(higher flotation kinetartificial increase of hformation, serving asultrafine particles. Bescombined with this teFlotation of sulphide fiwith a ‘‘standardized”18 per cent higher anCRF–HIC, compared w

conditioning stage, often expressed as conditioning time, has a pro-nounced effect on the concentrate recovery, grade and flotationrate. The high intensity conditioning process (HIC) enhanced theflotation recovery of the very fine particles of copper sulphides(Bulatovic and Salter, 1989), gold, uranium oxide and pyrite fines(Stassen 1991; Valderrama and Rubio, 1998), oxidized copper oresand copper and molybdenum sulphides (Rubio et al., 2003) andnickel ore (Chen et al., 1999). It is appear that this process is nowbeing applied industrial using special designed conditioners.

However, there is a situation, not much explored, viz. staged ormultifeed flotation, whereby differ from conventional flotation inthat two or more feed inputs are used and improvements in both

ize distribution on the staged flotation of poorly floating coals. These circuitse two-stage reagent addition, reflotation of classified tailings and split feedon of classified tailings circuits not only gave the best metallurgical performance,ffected by pulp density variations and imperfect size classificationfor deep cleaning of fine coal. In this staged flotation, the slowly flotable, mineral-rejected, and the resulting overflow is recleaned at lower flotation rates througher concentration. As a result, a clean coal product of lower ash and lower pyriticen produced at a higher yield as compared with the single-stage flotation and

tage flotation circuits. The selectivity of this staged flotation is dictated by theistribution in the feed coal. The mineral particles that remain in the clean coaldiameter of 6 lm, with 90per cent being smaller than 13 lmthe potential of different kind of circuit, namely the multifeed circuits, wasircuits differ from conventional circuits in that two or more feed inputs are used.mineral recovery and grade (through autogenous carrier effects) and

on reagents. Multifeed circuits offer potential for large plants processing severalsition

ental application of this technique to the flotation of a molybdenite ore wasthe potential benefit from multifeed circuit. Molybdenite multifeed circuitificantly improved the grade of optimized single-stage roughers: from 14 to 34aintaining comparable molybdenite recovery and reducing Syntex (emulsifying

enhance molybdenite flotation) and frother consumption by one-halfroving the flotation efficiency of Turkish bituminous coal fines by a split feednvestigated in comparison with conventional single-stage flotation. Coal samples6 mm fines treatment circuit of an operating washery were used in laboratorysingle-stage and split feed flotation circuits. Collector (kerosene) and frother (iso-re used as major test variables. Split feed flotation appeared to be a betterm the metallurgical point of view

olybdenite flotation circuits were carried out experimentally, using semi-batchestigation of a new multifeed flotation circuits, in which flotation feed is split intoconcentrate from one stage being fed with the other half of the feed into a secondould permit reduction in reagent additions and higher grade rougher concentratee flotation of a molybdenite oretation showed that multifeed rougher circuits can provides can yield enhancedof naturally hydrofobic minerals, such as coal and molybdenite

ed that open multifeed rougher circuits can provide wield and product ashto single-stage rougher. Closed multifeed rougher circuits provided much highernal single-stage rougher circuits at lower reagent consumptiontion, an open two-stage multifeed rougher circuits appears to be far superior toer for producing a much higher grade rougher concentrate at lower reagent

pear to be due to the effects of conditioning fresh feed with reagent-coated

per flotation circuit aiming to identify copper losses on fine fraction. Flotationroposed, namely rougher concentrate recycling on scavenger–cleaner stage (withagent addition)ined when rougher concentrate was recycled on scavenger–cleaner stage,reagent addition (enhancement of 15 per cent copper recovery)

ion flotation (CRF) was proposed aiming to modify flotation circuits design. CRFonal flotation stage and a second stage in which its feed is compound of ‘‘first”ics particles) concentrates and fresh feed. On this stage (second stage) occurs anigher flotation kinetics particles. Those particles are suitable for aggregates‘‘seeds” for these aggregates, transporting lower flotation kinetics fine andides, aggregates formation can increase using high intensity conditioning (HIC)chniquenes results, using CRF technique and HIC combined (CRF–HIC), were compared

mill laboratory procedure. Concentrate copper recoveries were of the order of 7–d their process kinetics values were higher, 1.5 min�1 for CRF and 2.9 min�1 forith 1.2 min�1 for standardized mill laboratory procedure

1200 E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206

mineral recovery and grade (through autogenous carrier effects)and conservation of flotation reagents were reported. There aresome industrial cases whereby concentrate (a fraction) is, occa-sionally, being recycled but is not a common practice worldwide.

Several authors (Hu and Huang, 1977; Firth et al., 1978, 1979;Tsai, 1988; Rubio, 1988; Fuerstenau, 1988 Fuerstenau et al.,2000; Hosten and Muratoglu, 1996; Rodrigues et al., 2005; Proyec-to Flotación de finos y ultrafinos – Codelco – Chuquicamata – IM2,

Fig. 1. Schematic representation of the configuration of various conventi

100

95

Cummulative weight, %

90

85

80

75

t, %

igh

we

tive

ula

mm

Cu

70

65

60>53 35 25

Particle size,

Fig. 2. Copper and mass size

2002; Tabosa, 2007; Tabosa et al., 2009) – see Table 1 – on theanalysis of some coal, copper and molybdenum sulphide flotationcircuits reported the multifeed circuit as a potential alternative cir-cuit. According to them, several factors may be responsible to theseimprovements. Split feed flotation on multifeed circuits, wherebythe concentrate of part of a rougher circuit is mixed with fresh feedonto another rougher circuit, reducing reagents consumption andimproving fine mineral particles recovery. Also, flotation rate in-

onal and multifeed flotation circuits (after Fuerstenau et al. (2000)).

1.20

1.15

Copper, %

1.10

1.05

1.00

r, %

ppe

Co

0.95

0.90

0.85

0.8015 <15

µm

distribution in sample.

E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206 1201

creases due to an ‘‘artificial” increase of valuable mineral grade(faster flotation rate), where hydrophobic particles may act as‘‘seeds” (carrier) to fine particles. Besides this, lower reagents con-sumption, due to the effects of conditioning fresh feed with re-agent-coated rougher concentrate.

Fuerstenau et al. (2000) show that multifeed rougher circuitscan provide enhanced results in the flotation of naturally hydro-phobic minerals, such as coal and molybdenite. For coal, it wasshown that open multifeed rougher circuits (Fig. 1) can provideyield and product ash contents comparable to single-stage rougher.Closed multifeed rougher circuits provided much higher yieldsthan conventional single-stage rougher circuits at lower reagentconsumption. For molybdenite flotation, an open two-stage multi-feed rougher circuit appears to be far superior to the single-stage

Fig. 3. Schematic representation of the configuration of a sta

Fig. 4. Schematic representation of the configuration of

rougher for producing a much higher grade rougher concentrateat lower reagent consumption.

Firth et al. (1978) suggest a staged (multifeed) flotation circuitsfor an Australian poorly floating coal. These circuits were two-stage reagent addition, reflotation of classified tailings and splitfeed flotation. According to them, the key to successful flotationof poorly floating coals is the equitable distribution of collectorsto all size fractions.

The aim of this work is to compare, at laboratory scale, stan-dardized rougher flotation tests with partial concentrate (rougherfloated product) recirculation to rougher flotation feed (concen-trate recirculation flotation – CRF). Also, the comparative effect ofhigh intensity conditioning (HIC), as a pre-flotation stage for therougher flotation, is studied alone or combined with CRF. Compar-

ndard mill test and details of the flotation cell and froth.

a concentrate recirculation flotation (CRF) mill test.

1202 E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206

isons were measured through the separation parameters: grade,recovery (global and true flotation) and flotation rate constants in-cluded size-by-size results to assess the flotation behaviour of thefine fractions of copper sulphides.

2. Experimental

2.1. Materials

2.1.1. OreCopper porphyry samples, containing 0.98–1.10% Cu were sup-

plied by Codelco – Chile. About 1/4 of the mass and about 40% ofthe total copper content is in the fines and ultrafines fractions(<35 lm). Fig. 2 shows feed size-by-size mass and copperdistribution.

2.1.2. ReagentsLime, as pH regulator to pH 10.5 was added at the conditioning

stage together with SF-113 (sodium isopropyl xanthate) and SF-508 (Promoter) at 30 and 8 g t�1 concentration, respectively. Thefrother was a propylene glycol, DF-250, at 25 g t�1.

Fig. 5. Details of the: (a) 6-blade Rushton impeller and (b) baffles ad

Fig. 6. Schematic representation of the configuration

2.2. Methods

2.2.1. Standard mill testThe standard mill tests were carried out in a laboratory D12

Denver flotation machine and in a square flotation cell with 3 Lof capacity, endowed with a manual scrapper to remove the froth.The pulp with 38% in weight and with pH in 10.5 (monitored withan AnalionTM pH meter model PM 608), was conditioned with thecollectors SF-113 (25 g t�1), SF 506 (8 g t�1) and the frother(25 g t�1) for one minute to allow collector and frother adsorptionat 1000 rpm in the cell. After the conditioning time, the flotationstarted with the injection of air in the flotation cell, monitored witha flow meter. Data were obtained from concentrates taken at 1, 2,3, 5 and 9 min flotation, keeping the volume in the cell constantwith water injection at pH 10.5. Fig. 3 shows a schematic represen-tation and some pictures of the standard mill test configuration.

2.2.2. Concentrate recirculation flotation (CRF)The CRF studies were divided into two-stages. The first stage

was carried out a standard mill test, and the first minute concen-trate (higher flotation probability) was collected. The volume of

ded to the flotation cell at the high intensity conditioning stage.

of a high intensity conditioning (HIC) mill test.

E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206 1203

this concentrate was measured and added to the second stage feed.At this second stage, a new standard mill test was carried out,being its feed composed by a new feed plus the first concentrateof the STD test (first stage), whilst the pulp volume was kept con-stant. Fig. 4 shows a schematic representation of the concentraterecirculation flotation (CRF) tests.

80

70

60

50

40

30

%er

y,

cove

r re

cpp

erC

opC

20

-10

STD HIC CRF CRF HIC

00 2 4 6 8 10

Time, min

Fig. 7. Flotation kinetic curves of a copper sulphide ore, comparative results (STD,CRF, HIC, and CRF–HIC). Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1;[DF250] = 25 g t�1; pH = 10.5; 38% solids by weight; conditioning stage 1000 rpm(HIC and CRF–HIC: 1400 rpm - 3 min).

22

20

18

16

14

%e,

%ra

deer

gop

peC

o

12

10HIC CRF CRF-HIC STD

820 30 40 50 60 70 80

Copper recovery, %

Fig. 8. Copper grade-recovery curves, comparative results (STD, CRF, HIC, and CRF–HIC). Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1; [DF250] = 25 g t�1;pH = 10.5; 38% solids by weight; conditioning stage 1000 rpm (HIC and CRF–HIC:1400 rpm – 3 min).

Table 2Flotation of copper sulphides. Comparative separation parameters between STD (mill stancombined process (CRF–HIC).

Tests Massrecovery (%)

1st minute coppermetallurgicalrecovery (%)

Copper metallurgicalrecovery (%)

Copper gra

Feed calcul

STD 5.8 60.5 1.1CRF 9.6 25 67.0 1.3HIC 6.4 69.0 1.1CRF–HIC 8.6 43 78.0 1.6

2.2.3. HIC (high intensity conditioning)The HIC studies were carried out similarly to the standard mill

test, with the exception than the stage of the conditioning of thepulp with the reagents (collectors and frother), was substitutedfor a high turbulent stage, obtained with the introduction of fourbaffles in a flotation cell equal to the mill flotation test and a Rush-

dard), concentrate recirculation flotation (CRF), high intensity conditioning (HIC), and

de (%) True flotationrecovery (%)

Entrainmentgrade

K (min�1)

ated (%) Concentrate Tailing

11.2 0.4 30 1.7 1.29.1 0.5 38 1.2 1.5

12.0 0.4 51 0.9 2.015.0 0.4 62 0.7 2.9

26

STD HIC CRF CRF HICSTD HIC CRF CRF-HIC

22

18

14

mµm

size

, cl

e s

part

ian

pM

ea

10

60 2 4 6 8 100 2 4 6 8 10

Flotation time, min

Fig. 10. Mean particle size in the concentrate as a function of time, comparativeresults (STD, CRF, HIC, and CRF–HIC). Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1; [DF250] = 25 g t�1; pH = 10.5; 38% solids by weight; conditioningstage 1000 rpm (HIC and CRF–HIC: 1400 rpm – 3 min).

STD HIC CRF CRF HICSTD HIC CRF CRF-HIC

%e,

%ra

deer

gr

oppe

Co

Time, min

Fig. 9. Cumulative copper concentrate grades, comparative results (STD, CRF, HIC,and CRF–HIC). Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1; [DF250] =25 g t�1; pH = 10.5; 38% solids by weight; conditioning stage 1000 rpm (HIC andCRF–HIC: 1400 rpm – 3 min).

1204 E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206

ton impeller (Fig. 5a). This high-shear radial impeller is excellentfor particle dispersion.

The HIC were carried out in the same D12 Denver flotation ma-chine, when the stator was taken out, and the rotor was substi-tuted for a Rushton impeller with 1400 rpm speed (transferring2 kWh m�3 of pulp), during 3 min, without air injection, and with50% of solids by weight. After HIC, the solids content was adjustedto 38% solids by weight, the impeller speed was changed back to1000 rpm and the flotation started with the injection of air in theflotation cell, monitored with a flow meter, as in the STD test. Figs.5 and 6 show some details of the HIC system used.

Herein, the studied variable, conditioning time, was quantifiedin terms of energy transferred to pulp. This energy was measuredwith a wattmeter, as described by Valderrama (1997).

4.0

STD HIC CRF CRF-HIC

3.5

3.0

2.5

2.0

s, %

Mas

sM

1.5

1.0

0.5

0.0

2.2.4. CRF combined with HIC (CRF–HIC)The CRF–HIC studies were carried out as a HIC test (stage 1), as

described previously, and the first minute concentrate (higherflotation probability) was collected. The volume of this concentratewas measured and added to the second stage feed (new HIC). Atthis second stage, a new HIC mill test was carried out, (stage 2)being its feed composed by a new feed plus the first concentrateof the HIC test (stage 1), whilst the pulp volume was kept constant.

All concentrates and tailings were filtered, dried, weighed andanalyzed for copper by atomic adsorption (SpectrAA 110-Varian�).

The process efficiency was measured by evaluating separationparameter recovery, ‘‘true” recovery and entrainment grade (as de-scribed by Warren, 1985, 1991), Cu grade and rate constants. TheRinf (ultimate recovery) and k constants for the Klimpel model (bestmodel for the data obtained) were obtained using a nonlinearregression method (least square fitting). In all the fittings the R2 va-lue was about 0.99.

Finally, to assess the performance of flotation in the F and UFfractions, size-by-size copper recoveries were measured in thefractions: >53 lm; 53–35 lm; 35–25 lm; 25–15 lm and <15 lm.

60

STD HIC CRF CRF-HIC

50

40

30

%t,

%ei

ght

We

20

10

0>53 35 25 15 <15

Particle size, µm, µ

Fig. 12. Particle mass recoveries, in each fraction, compared to the mill standard:comparative results (STD, CRF, HIC, and CRF–HIC). Tests conditions: [SF-113] =30 g t�1; [SF-506] = 8 g t�1; [DF250] = 25 g t�1; pH = 10.5; 38% solids by weight;conditioning stage 1000 rpm (HIC and CRF–HIC: 1400 rpm – 3 min).

0.0 0.9 2.4 4.6 9.0 19.0 36.0 66.0 130.0 240.0

Particle size, µm, µ

Fig. 11. Size distribution of flotation concentrates: comparative results (STD, CRF,HIC, and CRF–HIC). Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1;[DF250] = 25 g t�1; pH = 10.5; 38% solids by weight; conditioning stage 1000 rpm(HIC and CRF–HIC: 1400 rpm – 3 min).

3. Results and discussion

Figs. 7 and 8 and Table 2 show comparative results of flotationof copper sulphides at lab scale. Fig. 7 and Table 2 shows that understandard flotation test conditions, less valuable particle are floatedcompared to the other techniques, reaching to recoveries of the or-der of 60% Cu with rate constants (Klimpel model) of about1.2 min�1 and a true recovery or recovery through bubble–parti-cles interactions of 30%.

The concentrate recirculation flotation (CRF) test (1st min con-centrate recirculation), compared with the mill standard test, in-creased the kinetics rate values, enhanced the Cu recovery by6.5% and the recovery by true flotation values by 8%; followed bya decrease in entrainment grade and a slight reduction in copperfinal concentrate grade.

When the high intensity conditioning (HIC) precedes the STDflotation test, the kinetics rate values approximately doubled; theCu recovery increased by 8.5% and the recovery by true flotationby 21%. Also, the entrainment grade reduced to 0.9 and the copperfinal concentrate grade slightly increased. Hence, results show thatthe high intensity conditioning, as a pre-flotation stage, might in-crease both copper grade and recovery revealing, like in other stud-ies, high process selectivity. This specificity appear to bedemonstrated by the enhanced true flotation values, which canonly be explained in terms of the appearance of ‘‘new” valuableand hydrophobic particles in the concentrate. Thus, either particleaggregation or surface cleanliness or both, are the mechanismsinvolved.

Yet, best metallurgical and kinetic results were obtained withconcentrate recirculation flotation combined with high intensityconditioning (CRF–HIC). Results were 17% greater Cu recoveryand 3.6% improved Cu grade at a higher flotation rates (2.4 timesfaster). These were accompanied by an enhancement equivalentto 32.5% in the true flotation values and by a low amount of en-trained copper particles (2.4 times smaller).

A comparative analysis of copper concentrate grade with flota-tion time (Fig. 9) shows similar results for both STD and HIC tests.Concentrate recirculation, however, affected each technique in adistinct manner. Whilst the concentrate recirculation flotation(CRF) decreases the copper concentrate grade the CRF assisted bythe high intensity conditioning (CRF–HIC) increases the copperconcentrate grade.

This general behaviour might be the result of an ‘‘artificial” in-crease in valuable (values) grade followed by an aggregation occur-ring during the HIC, improving overall flotation performance. The

100

STD HIC CRF CRF-HIC90

% 80

y, %

very

70

reco

ver

rop

p

60

Co

50

40>53 35 25 15 <15

Particle size, µm

Fig. 13. Comparative size-by-size flotation copper recoveries: (STD, CRF, HIC, andCRF–HIC). Copper recoveries of concentrate products in >53, 35, 25, 15 and <15 lmfractions. Tests conditions: [SF-113] = 30 g t�1; [SF-506] = 8 g t�1; [DF250] =25 g t�1; pH = 10.5; 38% solids by weight; conditioning stage 1000 rpm (HIC andCRF–HIC: 1400 rpm – 3 min).

E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206 1205

recycled valuable mineral particles would be acting as ‘‘seeds” orcarriers whereby other particles can be attached.

This combination of methods appears to be similar to the mech-anisms proposed for the so-called autogenous flotation (Valderra-ma and Rubio, 1998).

In order to prove the aggregation hypothesis, the mean size offloatable particles (concentrate) were measured comparatively.Fig. 10 shows that bigger particles were formed in the very firstminutes of flotation when that is assisted by the high intensityconditioning. This could therefore suggest the formation of somekind of particle aggregation.

Moreover, the particle size distribution of concentrate productsfor STD, CRF, HIC and CRF–HIC presented in Fig. 11 shows thatwhen flotation is assisted by the high intensity conditioning (bothHIC and CRF–HIC) coarser particles are recovered; which in turnscould indicate that particles acted as carrier for particle aggrega-tion of the fines themselves or with coarser particles.

Fig. 12 shows particle mass recoveries, in each fraction, com-pared to the mill standard. Results show that after HIC and CRF–HIC the recovery of fine (<15 lm) and intermediate (>53 lm) par-ticle size was higher when compared to STD. Highest recovery offine (<15 lm) particles was obtained with concentrate recircula-tion flotation, indicating its higher efficiency for the flotation ofthe finest particle fraction.

Finally, Fig. 13 shows size-by-size copper recoveries, obtainedin all techniques studied compared to the mill standard. Accordingto expectations (and mechanisms proposed), this figure showsclearly higher recoveries in practically all fractions, especially inthe fine and ultrafine fractions (5% higher to CRF) and in the coar-ser (>35 lm) fraction, 10–15% higher for CRF–HIC. These resultssupport the aggregation hypothesis of the fines themselves or withcoarser particles (‘‘seeds” or carrier). This may explain the resultsobtained, the higher recoveries and concentrate grades at higherprocess kinetics.

4. Conclusions

Results found showed that the concentrate recirculation, to arougher flotation feed, enhanced the metallurgical recovery, gradeand rate flotation of copper sulphides. The best results were ob-tained with high intensity conditioning (HIC), as a pre-flotationstage, combined with concentrate recirculation flotation (CRF),

increasing the kinetics rate values, Cu recovery and concentrategrade. These results were accompanied by an enhancement inthe ‘‘true” flotation and a low amount of entrained copper parti-cles. Particle aggregation occurring after HIC, enhanced by thehigher number of recycled floatable particles, might explain theseresults. This ‘‘artificial” increase in valuable mineral grade resultedin higher collision probability between hydrophobic particles act-ing as ‘‘seeds” or ‘‘carrier”. Size-by-size flotation recoveries appearto confirm this aggregation mechanism. Hence, it is believed thatthese results represent real advances and that this concentraterecirculation based technique may establish a new basis to im-prove the potential of HIC, as an efficient pre-flotation stage.

Acknowledgments

The authors are thankful to all institutions supporting researchin Brazil and Chile and all colleagues from the 31-year old LTM(Environmental and Mineral Technology Laboratory) from our Fed-eral University in Porto Alegre, South Brazil.

References

2002. Report - Proyecto Flotación de finos y ultrafinos – Codelco – Chuquicamata –IM2 (in Spanish).

Bulatovic, S.M., Salter, R.S., 1989. High Intensity Conditioning: A New Approach toImprove Flotation of Mineral Slimes, Conference of Metallurgists, Halifax,Canada, pp. 182–197.

Chen, G., Grano, S., Sobieraj, S., Ralston, J., 1999. The effect of high intensityconditioning on the flotation of a nickel ore. Part II: Mechanisms MineralsEngineering 12 (11), 1359–1373.

Collins, D.N., Read, A.D., 1971. The treatment of slimes. Mineral Science andEngineering 3, 19–31.

Firth, B.A., Swanson, A.R., Nicol, S.K., 1978. The influence of feed size distribution onthe stage flotation of poorly floating coals. The Australasian Institute of Miningand Metallurgy 267, 49–53.

Firth, B.A., Swanson, A.R., Stewart, K.W., 1979. Flotation circuits of poorly floatingcoal. International Journal of Mineral Processing 5, 321–334.

Fuerstenau, D.W., 1988. Flotation science and engineering: advances andchallenges. In: Fossberg, K.S.E. (Ed.), XVI International Mineral ProcessingCongress, pp. 63–80.

Fuerstenau, D.W., Chander, S., Abouzeid, A.M., 1979. The recovery of fine particlesby physical separation methods. In: Ann Arbor, A. (Ed.), Beneficiation of MineralFines: Problems and Research Needs, 1978 – Workshop Report, Sterling Forest,New York, pp. 3–59.

Fuerstenau, D.W., Williams, M.C., Hu, W., 2000. Application of multifeed circuits forcoal and molybdenite flotation. In: XXI International Mineral ProcessingCongress, pp. B9(10)–B9(16).

Hosten, C., Muratoglu, R.A., 1996. A preliminary study on split feed flotation ofTurkish bituminous coal fines. Fuel 75 (6), 651–653.

Hu, W., Huang, K., 1977. Ramification roughing differential cleaning in consistentpulp with aeration new technology on flotation of wolframite slime. Journal ofNonferrous Metals, 16–20.

Mathur, S., Singh, P., Moudgil, B.M., 2000. Advances in selective flocculationtechnology for solid–solid separations. International Journal of MineralProcessing 58, 201–222.

Rodrigues, W.J., Nankran, M.A., Fonseca, R., Rabelo, L., 2005. Otimização darecuperação de finos no scavenger–cleaner na usina do sossego. XXI EncontroNacional de Tratamento de Minérios e Metalurgia Extrativa 5, 235–242 (inPortuguese).

Rubio, J., 1988. Carvão Mineral – Caracterização e Beneficiamento. vol. 1, Nova LinhaArtes Gráficas (in Portuguese).

Rubio, J., 2002. Unconventional flocculation and flotation techniques. In: Ralston,J.M. Rubio, J. (Eds.), Flotation and Flocculation: From Fundamentals toApplications, Hawaii – US, pp. 17–32.

Rubio, J., Capponi, F., Matiolo, E., Nunes, D., Guerrero, C.P., Berkowitz, G., 2003.Advances in flotation of mineral fines. In: XXII International Mineral ProcessingCongress. Cape-Town, South Africa. pp. 1014–1002.

Rubio, J., Matiolo, E., Rodrigues, R.T., Capponi, F., 2006. Flotación Extensora y conacondicionamento em alta intensidad de finos y ultrafinos de sulfuros de Cu andMo, Congresso Internacional de Metalurgia Extractiva, Saltillo, Coahuila, Mexico(in Spanish).

Sivamohan, R., 1990. The problem of recovering very fine particles in mineralprocessing – a review. International Journal of Mineral Processing 28, 247–288.

Song, S., Lu, S., 2000. Theory and application of hydrophobic flocculationtechnology. In: XXI International Mineral Processing Congress, Rome, Italy,pp. 31–32.

Stassen, F.J.N., 1991. Conditioning in the flotation of gold, uranium oxide and pyrite.Journal of the South African Institute of Mining and Metallurgy 91, 169–174.

1206 E. Tabosa, J. Rubio / Minerals Engineering 23 (2010) 1198–1206

Subrahmanyam, T.V., Forssberg, E.S., 1990. Fine particle processing: shearflocculation and carrier flotation – a review. International Journal of MineralProcessing 30, 265–286.

Tabosa, E., 2007. A flotação com reciclo de concentrados (FRC), uma técnicaalternativa de recuperação de finos de minérios: fundamentos e aplicações.M.Sc. Thesis, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil,122 pp (in Portuguese).

Tabosa, E., Centeno, C., Testa, F., Rubio, J., 2009. Flotação com reciclo de concentrado(FRC), flotação com pré-condicionamento em alta intensidade (CAI) e processoscombinados para recuperação de finos de minério: fundamentos e aplicações.In: XXI Encontro Nacional de Tratamento de Minérios e Metalurgia Extrativa.pp. 309–316 (in Portuguese);

Trahar, W.J., 1981. A rational interpretation of role of particle size in flotation.International Journal of Mineral Processing 2, 289–327.

Tsai, S.C., 1988. Staged Flotation of Fine Coal and the Effects of Mineral Size andDistribution. Industrial and Engineering Chemistry Research 27 (9), 1669–1674.

Valderrama, L., 1997. Estudos de flotação não convencional para o tratamento derejeitos de ouro. Ph.D. Thesis, Universidade Federal do Rio Grande do Sul, PortoAlegre, Brazil, 124 pp (in Portuguese).

Valderrama, L., Rubio, J., 1998. High intensity conditioning and the carrier flotationof gold fine particles. International Journal of Mineral Processing 52, 273–285.

Warren, L.J., 1985. Determination of the contributions of true flotation andentrainment in batch flotation tests. International Journal of MineralProcessing 14, 33–44.

Warren, L.J., 1991. Shear-flocculation. In: Ralston, J., Laskowski, J.S. (Eds.), ColloidChemistry in Mineral Processing. Elsevier, Amsterdam Ch. p. 10.