ito

, Gustr

Keywords:Temperature-responsive polymersSelective occulationIron ore otationFines otation

ove

study is to investigate the otation performance of anionic PNIPAM for an iron ore sample containingnes compared to sodium oleate, an industrial collector for hematite. PNIPAM conditioned at room tem-

igh-gra

As the medium-grade deposits begin to be exhausted, there hasbeen an increasing interest in recent years in the processing of low-grade deposits to answer the global demand for base metals.However, the nely-disseminated nature of low-grade ore bodiesrequires a high degree of liberation at even ner grinds, producingparticles that are generally below the optimum treatment sizes ofcurrent otation practice. The result is typically a loss of recovery

es dense mediumncentratinused to be

ate deposits requiring grinds ner than 212 lm (King, 200Environmental Protection Agency, 1994). A typical owinvolves either the direct anionic otation of hematitereverse cationic/anionic otation of silica gangue, and is precededby a desliming stage to reduce the detrimental impact ofparticles ner than 20 lm on overall recoveries (Filippov et al.,2014; Ma, 2012). The slimes do not usually undergo further pro-cessing and are sent directly to the tailings, representing a sig-nicant loss in the valuable minerals contained in the discardednes fraction.

Corresponding author.E-mail address: gvfranks@unimelb.edu.au (G.V. Franks).

Minerals Engineering 77 (2015) 6471

Contents lists availab

n

elsthe processing of such particles can be done reliably and effectivelywith the otation process, resulting in the proliferation of otationin ore processing plants worldwide (Houot, 1983; Pryor, 1965;Trahar, 1981).

of valuable nes.In the case of iron ore, current practice utilis

separation as a well-established method of coore liberated as coarse particles, while otation ishttp://dx.doi.org/10.1016/j.mineng.2015.02.0130892-6875/ 2015 Elsevier Ltd. All rights reserved.g ironneci-9; U.S.sheet

or thetionally, the shift to medium-grade deposits over the past few dec-ades has seen the rise of froth otation as the most commonmineral separation technique. Without a doubt, this is becausethe fall in feed grades has necessitated ner grinds to improvethe liberation of the valuable mineral from the unwanted gangue.While these ner grinds are difcult to treat with other techniques,

1976). This is accompanied by an increase in the entrainment ofne gangue into the concentrate, as well as slime coating effectsresulting in poor collector adsorption onto the desired mineral.This has placed the mining industry in a precarious position forthe treatment of future ore, with pressure to develop new tech-niques to either supplement or replace otation in the recovery1. Introduction

With the gradual depletion of hperature (25 C, below the lower critical solution temperature (LCST)) and oated at 50 C (above theLCST) was found to provide improved hematite grade and recoveries for particles above 20 lm, comparedto sodium oleate. This was attributed to the increased selectivity and hydrophobicity of PNIPAM.Turbidity testing conrmed the occulation of nes with PNIPAM, which deslimes the surface of the coar-ser particles. Below 20 lm, the hematite nes were almost completely recovered with PNIPAM. However,this recovery was not selective, attributed to the entrapment of gangue in the hydrophobic aggregates.Furthermore, conditioning of the polymer above the LCST resulted in signicant losses in grade and selec-tivity, as the polymer self-aggregates hydrophobically and precipitates unselectively onto the closestsurface.

2015 Elsevier Ltd. All rights reserved.

de ore bodies interna-

and selectivity, as the valuable nes are often caught in the stream-line of the bubbles and have poor particle-bubble contact(Sivamohan, 1990; Somasundaran, 1980; Trahar and Warren,Accepted 24 February 2015poly (N-isopropyl acrylamide) (PNIPAM). These polymers act as dual-function occulants and collectorsto form hydrophobic aggregates improving particlebubble collision and attachment. The aim of thisFlocculation/otation of hematite nes wresponsive polymer acting as a selective

Wei Sung Ng a,b, Rowan Sonsie a,b, Elizaveta Forbes b

aChemical and Biomolecular Engineering, University of Melbourne, Parkville, Vic 3010, AbCSIRO Mineral Resources Flagship, Clayton, Vic 3168, Australia

a r t i c l e i n f o

Article history:Received 1 January 2015

a b s t r a c t

A novel approach to the rec

Minerals E

journal homepage: www.h anionic temperature-cculant and collector

eorge V. Franks a,alia

ry of valuable nes is the use of temperature-responsive polymers such as

le at ScienceDirect

gineering

evier .com/ locate/mineng

The detrimental loss of nes has prompted the development ofmethods such as hydrophobic (or shear) occulation (Pascoe andDoherty, 1997; Pascoe and Wills, 1994; Song et al., 2001a,2001b; Warren, 1975) and selective occulation (Bagster, 1985;Bagster et al., 1993; Bagster and McIlvenny, 1985; Read, 1972;Sadowski and Polowczyk, 2004; Somasundaran et al., 1996;Sresty and Somasundaran, 1980; Yu and Attia, 1987). Both tech-niques involve the occulation of ne particles to produce largeraggregates that are within the optimum range for otation, whileminimising the negative impact of nes on recoveries.

Shear occulation uses high collector dosages coupled withhigh shear inputs (Subrahmanyam and Forssberg, 1990) to encour-age hydrophobic aggregation of the nes, with the downside beingincreased operating costs. On the other hand, selective occulation

compared to occulation with conventional PAM, as the use ofadditional collectors is not required. The occulating ability ofPNIPAM is well-established in the literature (Deng et al., 1996; Liet al., 2007, 2009; Li and Franks, 2008; OShea et al., 2010;OShea and Talln, 2011; Sakohara et al., 2002), and otation stud-ies thus far with PNIPAM on single-mineral systems have yieldedhigh recoveries (Forbes et al., 2011; Franks et al., 2009, 2008; Liand Franks, 2008). Of particular note is Li and Franks work demon-strating superior otation performance with PNIPAM relative to acombination of PAM and dodecyl amine as a collector in the ota-tion of pure kaolinite (Li and Franks, 2008).

Selectivity for the surfaces of the desired mineral can be impart-ed using the incorporation of charged groups to encourage attach-ment to counter-ionic particles, as demonstrated by OShea (2011).

is an investigation involving the batch otation tests of a realmixed-mineral ore sample comparing the performance of the

W.S. Ng et al. /Minerals Engineering 77 (2015) 6471 65uses high molecular weight polymers to selectively aggregate thedesired particles through bridging mechanisms. The main exam-ples for iron ore are anionic polyacrylamide (PAM) and starch,which are can be used to selectively occulate hematite nes(Bagster and McIlvenny, 1985; Khangaonkar and Bala Subramani,1993; Read, 1972). However, the polymers used as selective occu-lants are typically hydrophilic in order to ensure proper dispersionin the slurry, forming hydrophilic aggregates (Laskowski andLopez-Valdivieso, 2004). Therefore, a separate collector will berequired to impart the required surface hydrophobicity to the ocs,prior to the otation stage. In addition, the major issue limiting theuse of selective occulation is the entrapment of gangue particleswithin the ocs (Mathur et al., 2000; Yu and Attia, 1987, 1988).Even when the polymer is selective to the desired mineral, gangueparticles can become physically entrapped by the surroundingaggregates.

A novel approach to the nes issue is to use temperature-respon-sive polymers, such as poly (N-isopropyl acrylamide) (PNIPAM).Below the lower critical solution temperature (LCST) ofapproximately 32 C, PNIPAM is hydrophilic, soluble in water, andproduces well-dispersed suspensions (Franks, 2005; Sakoharaet al., 2002; Saunders et al., 1999; Sun et al., 2004). PNIPAM transi-tions into a hydrophobic state above the LCST, precipitating out ofsolution and forming hydrophobic aggregates (Deng et al., 1996;Franks et al., 2010; Li et al., 2007, 2009; Li and Franks, 2008;OShea, 2009).While the bridging aggregation of PAM is irreversible,the hydrophobic occulation of PNIPAM can be easily reversed bycooling below the LCST (Burdukova et al., 2010; Franks et al.,2010; OShea et al., 2010). Hence, PNIPAM is able to serve as a oc-culant, collector, aswell as a dispersant, as shown in Fig. 1. As a poly-mer, it has the added advantage of a modular structure, allowingfurther functionalisation and optimisation (Franks et al., 2009,2008).

The multi-functional nature of PNIPAM represents a simplerand possibly more economic approach to the otation of nesFig. 1. Multi-functional aspect of PNIPAM as a occulant, collector, and dispersant. Filleanionic PNIPAM against an industrial collector. In this study, weexamine the efcacy of anionic PNIPAM as a collector in the directanionic otation of hematite from a Vale iron ore sample, in com-parison to sodium oleate. Particular focus is given to the grade andrecovery of the nes fraction in the concentrate. The present workalso investigates the effect of polymer addition below and abovethe LCST, to reveal any impact on the otation performance.

2. Experimental

2.1. Materials

Iron ore samples containing 56% hematite were obtained fromVale. The samples were deslimed at 10 lm and consisted of 63%hematite in the nes fraction (1020 lm), 77% hematite in themids fraction (2075 lm) and 27% hematite in the coarse fraction(+75 lm), with the remainder being quartz and trace oxide miner-als. 14% of the hematite in the sample was present as nes, withthe majority 67% contained as mids, and 19% locked up as coarsematerial. The majority of the gangue was present as silica, withMicro-otation tests by Forbes and Franks (2013) showed thatanionic PNIPAM was a feasible collector for separating hematitefrom a synthetic mixture with silica. Performance in the midsand coarse regions were comparable between PNIPAM and sodiumoleate, an industrial collector, and improved recoveries wereobserved with PNIPAM in the nes region. However, the nesrecovery was unselective, indicating high levels of gangue entrap-ment with PNIPAM. Nevertheless, this study demonstrated that thepolymer was both insensitive to the presence of slimes and able torecover the valuable nes.

Building upon the work done on the occulation, otation per-formance and selectivity of PNIPAM-based polymers, the next stepd dots represent valuable minerals, while empty dots represent unwanted gangue.

trace quantities of alumina, magnesia, manganese (II) oxide, andsodium oxide.

A temperature-responsive, anionic random co-polymer of poly(N-isopropyl acrylamide) (PNIPAM) with 15% moles of acrylic acid(AA) was used in this study. The polymer was synthesised in ourlaboratories, following procedures as described by OShea (OShea

ed using a pH sensor plugged into a calibrated Radiometer PHM82.

concentrates collected at 0.5, 1, 2, 4 and 8 min of otation. Due

66 W.S. Ng et al. /Minerals Enginet al., 2011) and summarised in Fig. 2, with a molecular weightof 4.5 MDa and LCST of 39 C.

Sodium oleate powder (purity >82%) was purchased fromRiedel-de Haen (Seelze, Germany). Cataot P40, a low molecularweight polyacrylate dispersant (Bulatovic, 2010) was obtainedfrom CECA (Colombes, France). Methyl isobutyl carbinol (MIBC)from Orica Chemicals (Melbourne, Australia) was used as a frother.

2.2. Methods

2.2.1. Sample conditioningConditioning of the sample slurries with the collector prior to

the otation tests was performed to provide sufcient time for col-lector attachment to the mineral surface. The sodium oleate solu-tions were prepared on the day of the experiments throughdissolution of the desired mass in 20 mL of distilled water for30 min with stirring at room temperature (2025 C). Solutionswere prepared separately for each test. Conditioning of the slurrywas then performed within the otation cell at room temperature,through addition of the appropriate batch of sodium oleate solu-tion at 5 min prior to the start of the otation test.

The PNIPAM-co-AA solutions were prepared through dissolu-tion of the desired mass in 20 mL of distilled water for 48 h leadingup to the experiment with stirring at room temperature. Twomethods of PNIPAM-co-AA conditioning were investigated forcomparative purposes:

(a) Addition at room temperature: The polymer solution wasadded to the slurry at room temperature and allowed to con-dition for 5 min in the otation cell before the slurry washeated to 50 C.

(b) Addition at 50 C: The slurry was rst heated to 50 C, thenthe polymer solution was added to the otation cell andallowed to condition for 5 min prior to the start of the testwhile maintaining the temperature at 50 C.

Heating of the slurry was performed with a copper heating coilsubmerged at the back end of the otation cell. The coil wasattached to a recirculating pump in a water bath at a set point of50 C. A temperature sensor was used to conrm the temperatureof the slurry prior to the otation tests. Reagent dosage was per-formed on a per gram of ore basis, and the addition rates werebased on preliminary testing to produce concentrates with a rangeof recoveries.

2.2.2. FlotationThe otation tests were performed in a stainless steel 3 L bot-

tom-driven CSIRO mechanical otation cell as described in detailby Koh, Manickam and Schwarz (Guy, 1992; Koh et al., 2000).500 g of the hematite-containing ore were mixed with 2 L ofFig. 2. Synthesis of anionic PNIPAM-co-AA, adapted from OShea (2010).to low masses in the last two products, the fourth and fth concen-trates were combined into one sample for the purpose of thisstudy. The concentrate and tails were then dried in an oven at60 C.

Three collector addition regimes were investigated: a baselinetest with no collector, otation with sodium oleate, and otationwith PNIPAM-co-AA. The otation runs with collector werelabelled with SO for the sodium oleate oats and PN for thePNIPAM-co-AA oats, followed by a number indicating the addi-tion of collector on a g/t ore basis. The baseline and sodium oleateoats were conducted at room temperature, as with conventionalotation tests that do not require heating. PNIPAM-co-AA oatswere conducted at 50 C to induce the required hydrophobic tem-perature switch for PNIPAM.

Some of the otation tests were repeated to evaluate the repro-ducibility of the experimental data. For these tests, the grade andrecovery values presented have been averaged across allexperimental runs under the same conditions. The error bars pro-vided represent 95% condence intervals calculated based on thestandard deviation of the grade and recovery values (Weisberg,2013).

2.2.3. SizingTo ensure that the otation concentrate and tail samples were

fully dispersed prior to size analysis, the dried products were madeup to 50% solids by mass in distilled water and rolled for 15 min ina glass jar with 20 ceramic beads. This was performed with 20 mLof 0.5% w/w Cataot P40 dispersant to break up any remainingaggregates. Wet screening was then performed on the sampleswith a 75 lm screen stacked on a 20 lm screen, followed by drysizing at 20 lm and 75 lm with a Ro-tap sieve shaker to obtainne (20 lm), mid (20 to 75 lm) and coarse (+75 lm) fractionsfor each sample. Assaying on pulverised samples was performedusing X-ray uorescence (XRF) general fusion techniques(PANalytical PW 2404 XRF). The grade and recovery of each sizefraction were calculated in terms of hematite rather than iron.

2.2.4. Turbidity testingTurbidity testing was performed on 15 mL tailings samples col-

lected during the otation studies with the use of a Hach 2100ANTurbidimeter. The samples were allowed to settle for 2 min prior toTests were conducted at pH 8, adjusted prior to the oats witheither KOH or HCl. This pH was chosen as:

(a) Hematite is positively-charged below 8.3, facilitating attach-ment with anionic PNIPAM (Shibata and Fuerstenau, 2003).

(b) Silica is well-dispersed between pH 3 and 12, reducing theentrapment of gangue during occulation (Bagster andMcIlvenny, 1985).

(c) This is sufciently close to pH 9, where the recovery ofhematite with sodium oleate is maximised (Shibata andFuerstenau, 2003), without altering the charge characteris-tics of hematite.

The otation tests were performed following procedures asdescribed in detail elsewhere (Forbes et al., 2014) to produce 5distilled water containing 102 M of KCl as electrolyte, prior tothe addition of the appropriate collectors for conditioning. Duringconditioning, the cell was topped up to the 3 L mark with addition-al distilled water containing 102 M of KCl. pH levels were record-

eering 77 (2015) 6471measurement, and the turbidity result was recorded after 30 s inthe instrument. Testing was conducted at room temperature forthe sodium oleate and baseline tailings, whereas the PNIPAM-co-

AA samples were maintained at 50 C to approximate the otationconditions.

3. Results and discussion

3.1. Effect of collector type

To conrm that the otation work for sodium oleate andPNIPAM-co-AA done at different temperatures is comparable, aset of preliminary otation tests for sodium oleate were conductedat room temperature and 50 C. No signicant differences werefound in the graderecovery curves of the sized product for sodiumoleate at the two temperatures studied, as shown in Figs. A1A3 inthe Supplementary Data section available online, validating theapproach used in the current work.

The results of the individual otation grade and recovery testsare presented in Figs. A4A6 in the Supplementary Data sectionavailable online. Fig. 3 depicts the averaged grade-recovery curvescomparing the otation performance of PNIPAM-co-AA and sodi-um oleate at different reagent concentrations. As seen in Fig. 3a,

W.S. Ng et al. /Minerals EnginFig. 3. Grade-recovery curves of the concentrate oated with PNIPAM-co-AA andsodium oleate for (a) the coarse size fraction (+75 lm), (b) the mids fraction (+20/

75 lm), and (c) the nes fraction (20 lm). SO denotes sodium oleate oats,PN denotes PNIPAM-co-AA, while the number denotes the g/t ore addition of thereagents. Error bars indicate 95% condence intervals of the mean values.for the coarse fraction, PNIPAM-co-AA performs better as a collec-tor relative to sodium oleate, with superior grades at matchinglevels of recovery. Poor selectivity for hematite was observedunder high dosages of sodium oleate, with little improvement inconcentrate grades accompanying recovery of almost all the coarsehematite. While higher grades are achievable with lower dosagesof oleate, the trade-off in recovery is signicant and likely to beuneconomical, with a resulting concentrate grade (44%) that is onlyslightly better than the most dilute concentrate found withPNIPAM-co-AA (38%). Except for PN 150, very little coarse hematitewas recovered past the rst minute of otation (represented by thesecond data point on each curve, which covers the rst two con-centrates) under all conditions. This suggests that the readily-oat-able coarse material rapidly attaches to the otation bubbles, aidedby the high probability of particle-bubble contact due to the largeparticle sizes and a lower chance of being caught in the streamlineof the bubbles.

Fig. 3b illustrates the grade-recovery curves for the mids frac-tion comparing the performance of PNIPAM-co-AA and sodiumoleate at varying reagent concentrations. Similar to the coarse frac-tion, enhanced performance was found with PNIPAM-co-AA rela-tive to sodium oleate, with signicantly cleaner concentrates atall reagent concentrations. Poor selectivity was again observedfor high dosages of sodium oleate, while low dosages resulted inpoor recovery accompanying improved grades. Overall, slowerkinetics and higher recoveries were observed with the mids frac-tion relative to the coarse, as there is a smaller probability of par-ticle-bubble contact but higher proportion of liberated, oatablematerial.

The nes grade-recovery curves observed with PNIPAM-co-AAand sodium oleate are depicted in Fig. 3c. Close to 100% of the nehematite was recovered at all levels of PNIPAM-co-AA addition.However, unlike the coarse and mids region, poor selectivity wasfound with the polymer, producing a nes fraction in the concen-trate that approaches the nes feed grade. As baseline collector-less testing produced mass recoveries of less than 2% at all sizes(refer to Fig. A7 in the Supplementary Data section availableonline), the entrainment of ne particles is not expected to sig-nicantly affect concentrate grades. Furthermore, dilution fromunliberated composites is highly improbable for particles below20 lm. Hence, the entrapment of gangue particles during occula-tion is likely to be the major factor causing the low nes selectivity,as the hydrophobic aggregates of valuable nes are diluted by thepresence of gangue particles trapped within the aggregates.

Furthermore, low reagent dosages are typically associated withlow recoveries in the nes region, due to the large surface area pervolume ratios resulting in high reagent consumptions. However,this was not observed for PNIPAM-co-AA. Despite incompleterecoveries in the coarse and mids fraction indicating reagent addi-tions below the saturation level for PN 150 and PN 200, close tocomplete recovery of the valuable minerals was achieved in thenes. Therefore, a signicant portion of the nes recovery at lowPNIPAM-co-AA dosages is highly likely to have been recoveredthrough entrapment, albeit of valuable particles instead ofunwanted gangue.

Although sodium oleate fares slightly better in the nes region,the grade improvement is much smaller compared to the coarseand mids size fractions. This result is in agreement with the previ-ous otation studies performed with PNIPAM-co-AA, with theentrapment effect severely impacting grades and recoveries inthe nes fraction (Forbes and Franks, 2013). Nevertheless, unlikewith current otation techniques, the presence of nes did notappear to hamper the otation performance of the mids and coarse

eering 77 (2015) 6471 67particles with PNIPAM-co-AA. This represents a huge potentialgain, as the valuable nes do not need to be discarded to the tail-ings, and can instead be recovered into the concentrate.

The activity of PNIPAM-co-AA in the nes region may have alsocontributed to the otation performance in the coarse fraction, asthe occulation and accompanying entrapment of nes effectivelyacts as a desliming process, reducing the amount of slimes avail-able for coating onto the coarser materials. The effect is two-fold;an improvement in grades, with less dilution of the concentratearising from coarse silica coated with hematite nes, and animprovement in kinetics, as the coarse hematite now has a largerfree surface available for sufcient collector attachment.

In addition to the formation of nes aggregates, a occulationstep prior to otation might increase the attachment of hydropho-bic nes to coarser hydrophobic particles. This encourages recoverythrough carrier otation or ultraotation, whereby the coarserhydrophobic particles serve as a vehicle for the nes (Chia andSomasundaran, 1983; Sivamohan, 1990; Subrahmanyam andForssberg, 1990; Wang and Somasundaran, 1980). Bagster et al.(1993) suggested that this would have a higher impact on nes

3.2. Flocculation of nes

Fig. 5 compares the turbidity measurements for the tailingsfrom the PNIPAM-co-AA, sodium oleate, and baseline otationstudies. After 2 min of settling time, a signicantly lower level ofturbidity was found with the PNIPAM-co-AA samples relative tothe baseline, indicating fast settling characteristics which we positto have arisen from a high degree of nes aggregation. This showsthat the polymer conforms to our expectations and functions as anes occulant, and explains the nes recoveries observed.Conversely, particle settling appears to be hindered by the additionof sodium oleate, producing a turbid suspension that sediments

culation (Mathur et al., 2000; Yu and Attia, 1987, 1988), and should

68 W.S. Ng et al. /Minerals Engineering 77 (2015) 6471recoveries relative to the improved otation kinetics of nes arisingfrom the post-occulation upgrade in size. However, it is unlikelyfor carrier otation to have played a major role in the current study,as there is no correlation for the grades and recoveries between thecoarse and nes fractions.

An interesting effect of PNIPAMs dispersive properties belowthe LCST is that it eliminates the need for the addition of disper-sants prior to nes aggregation. Conventional methods of selectiveocculation typically require a pre-conditioning stage with disper-sants to sufciently liberate the particles and prevent hetero-co-agulation, incurring extra costs (Bagster, 1985; Read, 1972;Somasundaran et al., 1996). This is especially important for thetreatment of poly-disperse systems such as iron ore, with thehematite and silica at opposing charges. With PNIPAM-co-AA, thepolymer acts to free up the slimes during conditioning below theLCST, encouraging increased polymer attachment to the slime-freesurface areas, and minimising hetero-coagulation.

Fig. 4 shows a comparison of the size-by-size and overall ota-tion performance between the samples oated with PNIPAM-co-AA and sodium oleate. It can be seen that the polymer is able to pro-duce a nal concentrate that is higher in grade relative to sodiumoleate at similar or better levels of recovery (lled circles and trian-gles in Fig. 4). This is particularly clear when one compares, forinstance, PN 150 to SO 150, or PN 250 to SO 250 and SO 750.Furthermore, nearly 100% of the valuable nes are recovered at alldosages of PNIPAM-co-AA. Clearly, PNIPAM-co-AA is a multi-func-tional collector that is not only insensitive to the presence of slimes,but is able to collect the valuable nes into the concentrate, whilemaintaining a performance which rivals that of sodium oleate.

Fig. 4. Summary of the otation recoveries and grades obtained for each sizefraction. SO denotes sodium oleate oats, PN denotes PNIPAM-co-AA, while thenumber denotes the g/t ore addition of the reagents. The proportions of each size

fraction are not exactly the same for each of the samples because the individualrecoveries and losses are based on the recalculated hematite distribution for eachsample. Error bars indicate 95% condence intervals of the mean values.be investigated to improve the otation performance of the neswith PNIPAM-co-AA. Also, lower solid densities will help to reducethe probability of gangue particles being trapped between thevaluable particles.

The high grade of the iron ore feed in this study makes entrap-ment a particularly challenging issue. Assuming entrapmentoccurs at a 2-to-1 valuable-to-gangue ratio, the maximum achiev-able grade will then be 66% regardless of the feed grade. Therefore,for a slurry fed at a grade of 50%, only a small improvement can beobtained. The negative impact of gangue entrapment is expected tobe less apparent for the treatment of low-grade minerals such ascopper ores. A signicant upgrade is possible for a slurry with afeed grade of 1%. Hence, the use of PNIPAM-co-AA appears to besuitable for the roughing circuit of a low-grade ore processingplant.

3.3. Effect of PNIPAM addition method and conditioning temperature

Fig. 6 shows the performance of samples oated at a tem-perature above the LCST following conditioning with PNIPAM-co-AA at temperatures below and above the LCST. For the coarse frac-tion shown in Fig. 6a, addition of PNIPAM-co-AA to the slurryabove the LCST was found to yield lower concentrate grades com-pared to addition below the LCST. This can be partially attributedto the behaviour of the polymer above the LCST, with thehydrophobic portions preferring to associate with itself relativeslowly. This suggests that the sodium oleate acts as a dispersantunder the conditions of this study, which is in agreement with pre-vious experiments showing that sodium oleate can act as a disper-sant for hematite in alkaline slurries (Wang and Heiskanen, 1992).

Although the PNIPAM-co-AA is acting as a occulant for thenes fraction, it is not performing in as selective a manner asdesired. As discussed previously, this is most likely due to theentrapment of ne gangue within the ocs. To reduce the entrap-ment effect, the use of higher shear rates (introduced mechanicallyor through ultrasound techniques) to break up and reform theaggregates into cleaner ocs has been suggested for selective oc-Fig. 5. Turbidity measurements (in NTU) of the tailings from the PNIPAM-co-AA,sodium oleate, and collectorless baseline otation studies. Error bars indicate 95%condence intervals of the mean values from 4 repeat measurements.

nginW.S. Ng et al. /Minerals Eto particle surfaces. Therefore, for samples where the polymer isadded to a suspension that is already at a temperature above theLCST, the PNIPAM-co-AA rapidly self-aggregates upon contact withthe hot slurry, resulting in limited dissolution, mixing, and particleadsorption. When the polymer is added to a suspension at tem-peratures below the LCST, the attachment of PNIPAM-co-AA ontohematite surfaces during the conditioning stage provides a plat-form or nucleating site for the surface deposition of additionalpolymer when the temperature is raised above the LCST, withthe now-hydrophobic polymer attaching onto a particle linked byanother pre-adsorbed polymer (OShea et al., 2010, 2011, 2012).

Previous studies (OShea et al., 2010, 2012) on pure PNIPAMhave established that large ocs (particle aggregates) do not formif the polymer is added to the mineral slurry at temperatures abovethe LCST, indicating limited polymer surface adsorption withoutprior conditioning below the LCST. With charged randomPNIPAM copolymers, the ionic components encourage the attach-ment of polymer to particle surfaces even when added at tem-peratures above the LCST, with the downside being some degreeof bridging occulation occurring below the LCST, impeding the

Fig. 6. Grade-recovery curves of the concentrate oated with PNIPAM-co-AAconditioned above and below the LCST for (a) the coarse size fraction (+75 lm), (b)the mids fraction (+20/75 lm), and (c) the nes fraction (20 lm). PN denotesPNIPAM-co-AA, the number denotes the g/t ore addition of the reagents, whileHOT and COLD denote conditioning above (50 C) and below (2025 C) theLCST respectively. Error bars indicate 95% condence intervals of the mean values.dispersive properties of PNIPAM (OShea et al., 2011). The coarserecoveries seen are a result of this compromise, with PNIPAM-co-AA functioning as a collector following conditioning above theLCST, albeit with a slightly weaker performance relative to the coldaddition method.

Although a grade-wise comparison reveals decreased perfor-mance for the hot addition samples, signicantly higher recoverieswere observed with PN 150 HOT relative to PN 150 COLD. One pos-sible explanation is that the bulk of the added polymer in the PN150 COLD sample had been effectively removed from the solutionvia selective attachment to particle surfaces following conditioningbelow the LCST and subsequent precipitation onto adsorbed poly-mers above the LCST. For the PN 150 HOT sample, a large amountof PNIPAM-co-AA remains in the solution, self-aggregating or pre-cipitating unselectively onto any particle surface to reduce contactwith water. The latter results in the occulation of as many parti-cles as the polymer can cover, producing hydrophobic aggregatesat the feed grade.

The recoveries reect this difference between the samples, withPN 150 COLD having less recovery and a cleaner concentrate due toselective multilayer adsorption and localisation of the polymeraround hematite attachment sites, while PN 150 HOT recovers asmuch material as it is able to from unselective attachment andocculation of all particles. As the polymer for PN 150 HOT isadded as a solution, it is expected to remain hydrophilic over theshort time period required for its temperature to increase to thetemperature of the slurry. Hence, the polymer might already bewell-mixed at the point where the conformational switch to ahydrophobic structure occurs. The timescale for this transition islikely slow enough to allow sufcient mixing but too rapid forselective particle attachment.

A similar result can be seen with the mids recoveries, as illus-trated in Fig. 6b. Cleaner concentrates are achieved with PN 150COLD compared to PN 150 HOT, and similarly with PN 250 COLDrelative to PN 250 HOT. Unexpectedly, there is little grade disparitybetween PN 250 COLD and PN 150 HOT at higher recoveries.Arguably, this may be due to the limits of the ideal grade-recoverycurve. Nevertheless, superior performance displayed by the PN 250COLD sample with double the recovery in the rst concentrate atalmost similar grades.

The difference is less pronounced in the nes region, as illus-trated in Fig. 6c. This is possibly because the cold addition samplesalready incorporate a high degree of gangue entrapment, resultingin concentrate grades close to the feed grade. Hence, it is unlikelyfor the hot addition samples to produce a poorer performancegrade-wise. The implication is that under the conditions studied,the strength of the polymer selectivity has little effect on the gradeand recoveries in the nes region because of entrapment. At such ahigh feed grade, the entrapment of particles results in the forma-tion of aggregates of similar grade regardless of whether the oc-culation is selective or unselective.

For both sets of hot- and cold-conditioned samples, it is difcultto quantify the contribution from the entrapment of valuable min-erals on recovery. Nevertheless, taking into account the resultsfrom the other size fractions, it can be argued that a larger portionof the hot-conditioned samples were recovered through entrap-ment, while the bulk of the cold-conditioned samples were collect-ed through true otation. A mass recovery of 99.8% was eventuallyreached in the nal concentrate for PN 250 HOT, with a corre-sponding near-complete recovery of hematite in all 3 size fractions,and a resulting concentrate grade equal to the feed grade.

It is interesting to note that for all 3 size fractions of the PN 250HOT sample, the rst concentrate is relatively clean, followed by a

eering 77 (2015) 6471 69signicant increase in hematite recoveries above 90% and a sub-stantial drop in grades in the second concentrate. This may suggestthat the material oated in the rst concentrate consists of mainly

than the rest of the unselectively occulated particles. Hence, such

providing the iron ore samples studied. We would also wish to

nginthank Dr John-Paul OShea for synthesising the polymer used inthe study and Dr Luke Connal, Kevin Davey, Emma Brisson, andWarren Bruckard for support at various stages of the work.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.mineng.2015.02.013.

References

Bagster, D.F., 1985. Studies in the selective occulation of hematite from gangueusing high molecular weight polymers. Part 2: Physical factors. Int. J. Miner.material is expected to be oated off rst, producing a relativelyclean initial concentrate, followed by the other less hydrophobicaggregates.

4. Conclusions

Batch otation tests on a natural iron ore sample containing amixed-mineral system have been conducted to compare the per-formance of PNIPAM as a otation collector for hematite relativeto sodium oleate. Based on the ndings presented, the followingconclusions can be drawn under the conditions used in the study:

(a) PNIPAM-co-AA performs signicantly better than sodiumoleate at recovering and concentrating hematite particlesabove 20 lm, attributed to the higher selectivity andhydrophobicity of the polymer, and a lower sensitivity tothe presence of nes. Furthermore, the occulation of nes,as evidenced by turbidity testing, effectively acts as a surfacedesliming process, increasing the free surface area availablefor polymer adsorption on larger particles.

(b) PNIPAM-co-AA is capable of providing excellent recovery ofthe nes, collecting nearly all hematite particles below20 lm at all addition rates tested. Although this recoveryis unselective due to gangue entrapment in the hematiteocs, the loss of valuable nes to tailings is signicantlyminimised relative to sodium oleate.

(c) The addition of PNIPAM-co-AA above the LCST results inpoor otation performance relative to conditioning belowthe LCST. The polymer rapidly turns hydrophobic and self-aggregates, recovering the valuable mineral through meansof unselective particle entrapment into hydrophobic ocs.

Acknowledgements

The authors would like to acknowledge the CommonwealthScientic and Industrial Research Organisation (CSIRO) for provid-ing the nancial support required for this work, as well as Vale forselectively attached polymer, whereas from the second concen-trate onwards the collected material is simply unselectivehydrophobic ocs of particles in the slurry. This supports the pre-vious suggestion of a small time window where the freshly-addedpolymer is hydrophilic in the hot slurry, whereby the polymer isdispersed and a small amount of particle attachment occurs.Although relatively limited, these polymer-adsorbed surfaces willlikely attract more polymer attachment once the hydrophobictransition is complete, consequently ending up more hydrophobic

70 W.S. Ng et al. /Minerals EProcess. 14 (1), 2132.Bagster, D.F., Gordon, J.H.R., Linkson, P.B., Quan, J.T.B., 1993. Developments in the

occulation/otation of hematite. In: Proceedings Sixth Conference of the AsiaPacic Confederation of Chemical Engineering. Engineers Australia, Melbourne,pp. 9592 - 9992.

Bagster, D.F., McIlvenny, J.D., 1985. Studies in the selective occulation of hematitefrom gangue using high molecular weight polymers. Part 1: Chemical factors.Int. J. Miner. Process. 14 (1), 120.

Bulatovic, S.M., 2010. 20 Flotation of mixed lead zinc sulphide oxide and oxidelead and zinc ores. In: Bulatovic, S.M. (Ed.), Handbook of Flotation Reagents.Elsevier, Amsterdam, pp. 6786.

Burdukova, E., Li, H., Ishida, N., OShea, J.-P., Franks, G.V., 2010. Temperaturecontrolled surface hydrophobicity and interaction forces induced by poly (N-isopropylacrylamide). J. Colloid Interface Sci. 342 (2), 586592.

Chia, Y.H., Somasundaran, P., 1983. Carrier otation of anatase from clay and itsphysiochemical mechanisms. In: Malghan, S.G. (Ed.), Ultrane Grinding andSeparation of Industrial Minerals. AIME, New York, pp. 117130.

Deng, Y., Xiao, H., Pelton, R., 1996. Temperature-sensitive occulants based onpoly(N-isopropylacrylamide-co-diallyldimethylammonium Chloride). J. ColloidInterface Sci. 179 (1), 188193.

Filippov, L.O., Severov, V.V., Filippova, I.V., 2014. An overview of the beneciation ofiron ores via reverse cationic otation. Int. J. Miner. Process. 127, 6269.

Forbes, E., Bradshaw, D.J., Franks, G.V., 2011. Temperature sensitive polymers asefcient and selective otation collectors. Miner. Eng. 24 (8), 772777.

Forbes, E., Davey, K.J., Smith, L., 2014. Decoupling rehology and slime coatings effecton the natural otability of chalcopyrite in a clay-rich otation pulp. Miner. Eng.56, 136144.

Forbes, E., Franks, G.V., 2013. Selective separation of hematite from quartz byotation using a temperature responsive polymer. In: Iron Ore Conference.AusIMM, Perth, Western Australia, Australia.

Franks, G.V., 2005. Stimulant sensitive occulation and consolidation for improvedsolid/liquid separation. J. Colloid Interface Sci. 292 (2), 598603.

Franks, G.V., Burdukova, E., Li, H., Ishida, N., OShea, J.P., 2010. Mechanismresponsible for occulation in poly (n-isopropyl acrylamide) temperatureresponsive dewatering attractive interaction forces induced by surfacehydrophobicity. In: 25th International Mineral Processing Congress. Brisbane,Queensland, Australia, pp. 40574067.

Franks, G.V., Li, H., OShea, J.-P., Qiao, G.G., 2009. Temperature responsive polymersas multiple function reagents in mineral processing. Adv. Powder Technol. 20(3), 273279.

Franks, G.V., Li, H.H., OShea, J.P., Qiao, G.G., 2008. Application of poly (N-isopropylAcrylamide) PNIPAM as a multiple function process aid in mineral otation andsolid/liquid separation. In: Wang, D.D., Sun, C.Y., Wang, F.L., Zhang, L.C., Han, L.(Eds.), 24th International Minerals Processing Congress. China Scientic BookService, pp. 20972106.

Guy, P.J., 1992. The development of laboratory batch otation equipment andpractice at CSIRO, Australia. In: AusIMM Annual Conference, pp. 7579.

Houot, R., 1983. Beneciation of iron ore by otation review of industrial andpotential applications. Int. J. Miner. Process. 10 (3), 183204.

Khangaonkar, P.R., Bala Subramani, K.J., 1993. Flocculation of hematite nes byanionic polyacrylamide polymers. Miner. Eng. 6 (7), 765774.

King, I., 2009. The latest developments in iron ore processing, In Mintek 75: ACelebration of Technology. Mintek, Randburg, South Africa.

Koh, P.T.L., Manickam, M., Schwarz, M.P., 2000. CDF somulation of bubble-particlecollisions in mineral otation cells. Miner. Eng. 13 (1415), 14551463.

Laskowski, J.S., Lopez-Valdivieso, A., 2004. Fine particle aggregation in mineralprocessing circuits. In: Laskowski, J.S. (Ed.), Fifth UBC-McGill BiennialInternational Symposium: 43rd Annual Conference of Metallurgists of CIM:Particle Size Enlargement in Mineral Processing. Met Soc, Hamilton, Canada, pp.1732.

Li, H., Long, J., Xu, Z., Masliyah, J.H., 2007. Flocculation of kaolinite clay suspensionsusing a temperature-sensitive polymer. AIChE J. 53 (2), 479488.

Li, H., OShea, J.-P., Franks, G.V., 2009. Effect of molecular weight of poly(N-isopropylacrylamide) temperature-sensitive occulants on dewatering. AIChE J. 55 (8),20702080.

Li, H.H., Franks, G.V., 2008. Role of temperature-sensitive polymers in hydrophobicaggregation/otation of silicate minerals. In: 24th International MineralsProcessing Congress, Wang, D.D., Sun, C.Y., Wang, F.L., Zhang, L.C., Han, L.(Eds.), China Scientic Book Service, pp. 12611269.

Ma, X., 2012. Froth otation of iron ores. Int. J. Min. Eng. Miner. Process. 1 (2), 5661.

Mathur, S., Singh, P., Moudgil, B.M., 2000. Advances in selective occulationtechnology for solidsolid separations. Int. J. Miner. Process. 58 (14), 201222.

OShea, J.-P., Qiao, G.G., Franks, G.V., 2009. Solidliquid separations enhanced by theaddition of surfactant to a temperature-responsive polymeric occulant. In:Chemeca, 2009.

OShea, J.-P., 2010. Solidliquid separation with poly(N-isopropyl acrylamide)-based temperature responsive occulants. In: Department of Chemical andBiomolecular Engineering. The University of Melbourne, p. 249.

OShea, J.-P., Qiao, G.G., Franks, G.V., 2010. Solidliquid separations with atemperature-responsive polymeric occulant: effect of temperature andmolecular weight on polymer adsorption and deposition. J. Colloid InterfaceSci. 348 (1), 923.

OShea, J.-P., Qiao, G.G., Franks, G.V., 2011. Temperature responsive occulation andsolidliquid separations with charged random copolymers of poly(N-isopropylacrylamide). J. Colloid Interface Sci. 360 (1), 6170.

eering 77 (2015) 6471OShea, J.-P., Qiao, G.G., Franks, G.V., 2012. Temperature-responsive solidliquidseparations with charged block-copolymers of poly(N-isopropyl acryamide).Langmuir 28 (1), 905913.

OShea, J.-P., Talln, C., 2011. The effect of salt concentration and pH on the solidliquid separation of silica suspensions with a temperature-responsiveocculant. Sep. Purif. Technol. 82, 167176.

Pascoe, R.D., Doherty, E., 1997. Shear occulation and otation of hematite usingsodium oleate. Int. J. Miner. Process. 51 (14), 269282.

Pascoe, R.D., Wills, B.A., 1994. Selective aggregation of ultrane hematite and quartzunder high shear conditions with conventional otation collectors. Miner. Eng.7 (56), 647656.

Pryor, E.J., 1965. Principles of froth otation. In: Mineral Processing. Springer,Netherlands, pp. 457519.

Read, A.D., 1972. The use of high molecular weight polyacrylamides in the selectiveocculation separation of a mineral mixture. Br. Polym. J. 4 (3), 253264.

Sadowski, Z., Polowczyk, I., 2004. Agglomerate otation of ne oxide particles. Int. J.Miner. Process. 74 (14), 8590.

Sakohara, S., Kimura, T., Nishikawa, K., 2002. Flocculation mechanism of suspendedparticles using the hydrophilic/hydrophobic transition of a thermosensitivepolymer. KONA 20, 246250.

Saunders, B.R., Crowther, H.M., Morris, G.E., Mears, S.J., Cosgrove, T., Vincent, B.,1999. Factors affecting the swelling of poly(N-isopropylacrylamide) microgelparticles: fundamental and commercial implications. Colloids Surf., A 149 (13),5764.

Shibata, J., Fuerstenau, D.W., 2003. Flocculation and otation characteristics of nehematite with sodium oleate. Int. J. Miner. Process. 72 (14), 2532.

Sivamohan, R., 1990. Problem of recovering very ne particles in mineralprocessing. Int. J. Miner. Process. 28 (34), 247288.

Somasundaran, P., 1980. Role of surface chemistry of ne sulphides in theirotation. In: Jones, M.J. (Ed.), Complex Sulphide Ores. IMM, London, pp. 118127.

Somasundaran, P., Das, K.K., Yu, X., 1996. Selective occulation. Curr. Opin. ColloidInterface Sci. 1 (4), 530534.

Song, S., Lopez-Valdivieso, A., Reyes-Bahena, J.L., Bermejo-Perez, H.I., 2001a.Hydrophobic occulation of sphalerite nes in aqueous suspensions inducedby ethyl and amyl xanthates. Colloids Surf., A 181, 159169.

Song, S., Lopez-Valdivieso, A., Reyes-Bahena, J.L., Lara-Valenzuela, C., 2001b. Flocotation of galena and sphalerite nes. Miner. Eng. 14 (1), 8798.

Sresty, G.C., Somasundaran, P., 1980. Selective occulation of synthetic mineralmixtures using modied polymers. Int. J. Miner. Process. 6 (4), 303320.

Subrahmanyam, T.V., Forssberg, K.S.E., 1990. Fine particles processing: shear-occulation and carrier otation a review. Int. J. Miner. Process. 30 (34),265286.

Sun, T., Wang, G., Feng, L., Liu, B., Ma, Y., Jiang, L., Zhu, D., 2004. Reversible switchingbetween superhydrophilicity and superhydrophobicity. Angew. Chem. Int. Ed.43 (3), 357360.

Trahar, W.J., 1981. A rational interpretation of the role of particle size in otation.Int. J. Miner. Process. 8 (4), 289327.

Trahar, W.J., Warren, L.J., 1976. The otability of very ne particles a review. Int. J.Miner. Process. 3, 101131.

U.S. Environmental Protection Agency, 1994. Technical resource document:extraction and beneciation of ores and minerals: Volume 3: Iron. U.S.Environmental Protection Agency, Washington.

Wang, Q., Heiskanen, K., 1992. Selective hydrophobic occulation in apatitehematite system by sodium oleate. Miner. Eng. 5 (35), 493501.

Wang, Y.-H.C., Somasundaran, P., 1980. A study of carrier otation of clay. In:Somasundaran, P. (Ed.), Fine Particle Processing. AIME, New York, pp. 11121128.

Warren, L.J., 1975. Shear-occulation of ultrane scheelite in sodium oleatesolutions. J. Colloid Interface Sci. 50 (2), 307318.

Weisberg, S., 2013. Applied Linear Regression. Wiley.Yu, S., Attia, Y.A., 1987. Review of selective occulation in mineral separations. In:

Attia, Y.A. (Ed.), Flocculation in Biotechnology and Separation Systems. ElsevierScience Publishers, Amsterdam, pp. 601637.

Yu, S., Attia, Y.A., 1988. Entrapment and entrainment in the selective occulationprocess, Part 2: Possible methods for the minimization of entrapment andentrainment. In: Attia, Y.A., Moudgil, B.M., Chander, S. (Eds.), InterfacialPhenomena in Biotechnology and Materials Processing. Elsevier, Amsterdam,pp. 503514.

W.S. Ng et al. /Minerals Engineering 77 (2015) 6471 71

Flocculation/flotation of hematite fines with anionic temperature-responsive polymer acting as a selective flocculant and collector1 Introduction2 Experimental2.1 Materials2.2 Methods2.2.1 Sample conditioning2.2.2 Flotation2.2.3 Sizing2.2.4 Turbidity testing

3 Results and discussion3.1 Effect of collector type3.2 Flocculation of fines3.3 Effect of PNIPAM addition method and conditioning temperature

4 ConclusionsAcknowledgementsAppendix A Supplementary materialReferences