Selective sulphidation and flotation of nickel from a nickeliferous laterite ore

Minerals Engineering xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

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

Selective sulphidation and flotation of nickel from a nickeliferous laterite ore C.T. Harris, J.G. Peacey, C.A. Pickles ⇑ Robert M. Buchan Department of Mining, Queen’s University, Kingston, Ontario, Canada K7L 3N6

a r t i c l e

i n f o

Article history: Available online xxxx Keywords: Selective sulphidation Nickeliferous limonitic laterites Flotation High temperature experiments TGA/DTA with EGA

a b s t r a c t The sulphidation of a nickeliferous lateritic ore was studied at temperatures between 450 and 1100 °C and for sulphur additions of 25–1000 kg of sulphur per tonne of ore. The experiments demonstrated that the nickel could be selectively sulphidized to form a nickel–iron sulphide. It was found that both the grade and the sulphidation degree largely depended upon the temperature and the sulphur additions, with temperatures above 550 °C exhibiting the highest nickel sulphidation degrees and grades. A DTA/ TGA with mass spectrometer was used to further elucidate the nature of the phase transformations that occurred upon heating of the ore in the presence of sulphur. It was found that at low temperatures, the Fe–Ni–S phase was submicron in nature and heating to temperatures between 1050 and 1100 °C allowed for the growth of the particles, due to the increased sulphide mobility associated with the formation of a liquid sulphide matte phase, containing dissolved oxygen. Flotation studies conducted on 60 g samples showed that the sulphides responded to flotation with maximum grades of up to 6–7 wt.% nickel being achieved. Recoveries were approximately 50% on a sulphide basis and it was determined that the low nickel grades were due to the entrainment of magnetite fines. Ó 2013 Published by Elsevier Ltd.

1. Introduction The nickel industry is currently facing a significant transition from a sulphide ore based production stream to one that will require the successful processing of nickel-bearing oxidic ores, known as laterites. The conventional high-grade sulphur deposits are becoming both depleted and more difficult to access, and therefore new deposits with lower grades and different ore types, which previously were not competitive with the treatment of sulphide ores, are now being considered as options. While nickel extraction has historically been dominated by the pyrometallurgical treatment of sulphide ores, the relatively high abundance of nickel laterites is now beginning to translate into increased production from this oxidic ore type, which account for approximately 70% of estimated resources, but only 40% of current production (Dalvi et al., 2004). In the past, the commercial treatment of sulphide ores has relied on the successful production of a nickel concentrate, with a significant increase in the nickel grade, compared to that of the ore. The production of a nickel concentrate would result in a much lower mass load in the high temperature and energy consuming smelting process. The majority of recent sulphide mines do not have a smelter located on site, as the infrastructure costs associated with the instal-

⇑ Corresponding author. E-mail address: [email protected] (C.A. Pickles).

lation of a smelter are not justified. This decoupling of concentrate production and smelting, allows flexibility in processing, either as an integrated producer or through toll smelting arrangements. In contrast, thus far, nickel laterites have not been amenable to significant physical upgrading, and this inability to produce a nickel concentrate from a lateritic ore, results in large quantities of material, which must be treated in the higher energy consuming pyrometallurgical processes or the high acid consuming hydrometallurgical processes. Additionally, the processing of the laterite ores will likely incur higher environmental costs than the sulphide deposits (Mudd, 2009, 2010). It is therefore logical to try to attempt to develop a process that can produce a nickel concentrate from a lateritic ore without proceeding through a complete smelting route, particularly as the high grade electrical energy for the smelting process often requires the construction of a power plant at the mine site. While there have been limited studies examining the selective sulphidation of nickeliferous ores, there have been numerous studies which have examined the selective reduction of nickel oxide to metallic ferronickel. It has been demonstrated that under reducing conditions, a high degree of nickel conversion to a ferronickel alloy can be obtained, while at the same time leaving much of the iron as an oxide (DeGraaf, 1979; Li et al., 2010; Zhu et al., 2012). Although this does not ensure successful sulphidation, it does indicate that the nickel within the limonitic ore is amenable to selective chemical conversion. Additionally, studies of the dehydroxylation of a limonitic ore (Landers et al., 2009a,b), suggest that a highly reactive combination of iron and nickel oxides will be available for reaction at

0892-6875/$ - see front matter Ó 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.mineng.2013.02.016

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

2

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

relatively low temperatures (<300 °C). A thermodynamic analysis of the selective sulphidation process (Harris, 2011) reveals that there is an operating window, where a nickel sulphide phase can be produced with a nickel grade on the order of 20 wt.% nickel, while achieving reasonable nickel sulphidation degrees (>80%). Provided the nickel can be successfully converted into a sulphide phase of sufficient particle size, the flotation of these particles to produce a concentrate with between 10 and 14 wt.% nickel should be feasible. This paper represents an extension of previous research on the sulphidation process as described by Harris et al. (2011). A series of high temperature sulphidation experiments, thermogravimetric/ differential thermal analysis and flotation experiments were performed, in order to assess the potential of such a treatment route, and to define the key chemical reactions, phase changes and general phenomena which would impact such a process. The purpose of the sulphidation experiments was to examine the behaviour of a lateritic ore under sulphidizing conditions and to quantify the degree of selectivity that could be achieved for a variety of conditions. Additional information on the chemical reactions was obtained through thermogravimetric differential thermal analysis coupled with evolved gas analysis (TGA/DTA with EGA). Proof of concept flotation studies were also done in order to assess the potential of this treatment option, as well as to determine the critical issues that would benefit from further research in this area. 2. Experimental 2.1. Raw materials The ore used for all of the experiments was a limonitic ore from the Ivory Coast, which was provided by Xstrata. The composition of the ore was determined by X-ray fluorescence (XRF) and is shown in Table 1. Assuming that the iron was present as goethite, the composition was converted to an approximate mineralogical composition as shown in Table 2. This composition agrees well with the X-ray diffraction pattern, which indicated that the only identifiable mineral was goethite. The ore was pulverised to 100 mesh and kept in an oven at 65 °C. The limonitic ore and reagent grade elemental sulphur were combined in varying proportions to provide a total sample mass of 3 g. The sample was then compacted into a briquette (1.27 cm in diameter and 1.15 cm in height), which increased the contact area of the reactants and also provided a highly reproducible sample size and surface area. 2.2. Sulphidation experiments Fig. 1 shows a schematic diagram of the bench scale experimental set-up used for all the sulphidation experiments. First, the briquettes were placed onto a quartz sample carrier in the cool end of the tube furnace. Experiments at or above 1000 °C required that

Table 1 Chemical analysis of the limonitic ore samples. Component (wt.%)

Ni

Fe

Co

MgO

SiO2

Sample A Sample B

1.24 1.23

59.8 58.6

0.11 NA

0.26 NA

2.66 NA

Table 2 Approximate mineral composition of the limonitic ore (dry basis). Mineral

NiO

FeOOH

Mg3Si2O5(OH)4

SiO2

Other

Wt.%

1.58

95.13

0.58

2.35

<1

the quartz carrier be lined with sintered alumina powder in order to prevent interaction. The furnace was purged with high purity nitrogen (5 purge volumes) to remove any residual air within the furnace tube. The sample was then introduced into the hot zone of the furnace and allowed to react for the desired length of time (between 30 min to 6 h) under a flowing nitrogen atmosphere. At the end of the test, the reacted sample was moved back into the cool zone of the furnace for 20 min. In order to prevent oxidation of the sample, the briquette was removed from the furnace and immediately placed in ethanol, following the procedure described by Chen et al. (2009). 2.3. Product analysis The quantity and the composition of the nickel and iron sulphides produced during the furnace experiments were determined using a combination of a bromine methanol diagnostic leach, and LECO analysis for sulphur content. Digestion of a sample containing oxides and sulphides into a solution containing 5% bromine and 95% methanol, selectively dissolves the sulphides, while leaving the oxides as a solid. A review of this procedure has been given by Harris (2011). The mineralogies of the raw materials and the reacted samples were characterised through the use of powder X-ray Diffraction (XRD), using a Xpert Pro Philips powder diffractometer and Cu ka radiation. Also, selected briquette samples were examined by optical and Scanning Electron Microscopy (SEM). These samples were impregnated with epoxy and sectioned vertically down the middle of the briquette. At temperatures at or above 1000 °C it was found that there were well-defined grains of oxides. In order to reveal the grain structure a variety of etchants were examined. It was found that concentrated hydrochloric acid with an etching time of two to three seconds was able to successfully reveal the grain structure of the oxide minerals formed at high temperatures. Additionally, a JXA JEOL-8900L electron microprobe (20 kV) with 5 wavelength dispersive X-ray detectors was used for characterisation of selected samples. In order to further understand the chemical reactions occurring during sulphidation, studies were conducted using simultaneous Thermogravimetric Differential Thermal Analysis coupled with Evolved Gas Analysis via mass spectrometry (TGA/DTA with EGA). These studies were conducted on a Netzsch STA 449, with a QMS 403C Aeolos quadrupole mass spectrometer. 2.4. Flotation studies Material for the flotation tests was obtained from the sulphidation experiments. Four 20 g briquettes were prepared and charged to the furnace to produce approximately 60 g of reacted sample. This was done three times under identical conditions to prepare enough bulk material for the flotation tests. The bulk sample was pulverized to 100 mesh and then split into 3 individual samples. Samples were prepared at two different levels of sulphur addition in order to assess the impact of bulk nickel sulphide grade on the flotation performance. Prior to flotation, each sample was mixed with water to achieve a pulp density of 66 wt.% solids and ground in a 0.5L Netzsch ISA mill M-V with 1200 g of ceramic grinding media at 1000 rpm for 2.5 min. 50 mg of sodium metasilicate was also added in order to prevent the agglomeration of fine particles (Bulatovic, 2007). The typical d80 of the material was 13– 16 lm. The sample was separated from the grinding media by wet vibratory screening. Flotation studies were conducted in a 1.5 L Denver D-12 flotation cell at a pH range of approximately 9–9.5. The pH was adjusted using either sodium carbonate, if the test was conducted in deionized water or burnt lime if the test was conducted in tap water. Potassium amyl xanthate (PAX) was used as the collector

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

3

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 1. Schematic diagram of the furnace set-up.

Table 3 Summary of conditions in the flotation tests. Test ID

Sulphur addition (kg/ tonne)

Reagents additions

Conditions

FT01 FT02

100 100

Initial test of reagents adjusted based on visual observations Assess the impact of TETA addition, adjusted based on visual observations

FT03

100

FT04

65

FT05

65

Sodium silicate, PAX, DF250,DI water Sodium silicate, PAX, DF250, TETA DI water Sodium silicate, PAX DF250, TETA. DI water Sodium silicate, PAX DF250, TETA, DI water Same as FT04 with tap water

Test with controlled dose of reagents at specific times – assess kinetics Controlled test with specific dose of reagents – used to compare the impact of different sulphur addition Similar to Test FT04 but done in tap water

3. Results and discussion 3.1. Sulphidation experiments Fig. 2 shows the experimental nickel and iron sulphidation degrees as a function of sulphur addition, as determined by the iron and nickel extraction in the bromine–methanol leach. The nickel extraction is much higher than the corresponding iron extraction at all sulphur additions, and reaches a constant value of nearly 80% extraction, whereas the iron extraction continues to increase with increasing sulphur addition. These results demonstrate that nickel is preferentially sulphidized, as expected thermodynamically. A bulk sulphide composition was calculated based upon the overall nickel and iron extraction extents and sulphur content

100 90 80

Metal Extraction (%)

and Dow Frother 250 was used as a frother. Previous studies involving the flotation of nickel sulphides have shown that the addition of triethylenetetramine (TETA) can be beneficial in improving flotation selectivity (Kelebek and Tukel, 1999). TETA was therefore added in the majority of the flotation tests. The agitator speed of the flotation cell was 900 RPM. A total of five flotation tests were performed and the key features of each test are presented in Table 3. Prior to the initiation of flotation, the pulp was conditioned with the collector for 2 min and the frother was added after one and a half minutes. Air was then introduced and the flotation concentrate was manually removed via skimming of the froth layer. Following the specified collection time, the air was turned off, stage wise additions of reagents were made and the pulp reconditioned with collector and frother for a minute prior to flotation of the next concentrate. The product streams, which consisted of between five to ten separate concentrates, were individually vacuum filtered, air-dried and then analysed for nickel, iron and sulphur. The tailings were pressure filtered, and air-dried, prior to analysing for nickel, iron and sulphur. Particle size analysis of selected flotation products was done using a Fritsch analysette 22 laser particle size analyser.

70 60 50 40 30 20 Ni Recovery Fe Recovery

10 0 0

200

400

600

800

1000

Sulphur Addition (kg S/tonne ore) Fig. 2. Nickel and iron sulphidation extent as a function of sulphur addition at 500 °C.

in the calcine, which is shown in Fig. 3. At low sulphur additions, the bulk composition indicates the formation of pyrrhotite, whereas at higher sulphur additions, the ratio of sulphur to metal indicates that a combination of pyrrhotite and pyrite is favoured. X-ray diffraction of select samples supports these results, as shown in Fig. 4. A separate nickel sulphide phase was not detected, indicating that this phase is either below the instrument detection limit, or that it is not crystalline, or is contained in an iron-bearing sulphide phase. An analysis of the peak position of pyrrhotite provides further information on the location of the nickel phase. The crystal lattice spacing of pyrrhotite shifts due to nickel substitution and sulphur composition (Naldrett et al., 1967). The Bragg angle and the associated lattice spacing of the 102 crystal plane dh102i agrees well with the diagnostic leach test results in terms of the overall sulphide composition. The spacing indicates that the

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

4

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

70

100

12

60

Metal Recovery (%)

80 FeS

50 40

FeS2

30

10 60

8 Ni Recovery (%) Fe Recovery (%) Ni Grade (%)

40

6

Ni Grade (%)

% Metal in Sulphide

14

4

20

20

2 10

0 400

450

0 0

200

400

600

800

550

600

650

0 700

TEMPERATURE (oC)

1000

Sulphur Addition (kg S/tonne ore)

500

Fig. 5. Metal recovery and nickel grade of sulphide as a function of temperature for ore mixed with 100 kg S/tonne ore.

Fig. 3. Calculated bulk sulphide composition (mole% basis) as a function of sulphur addition at 500 °C.

pyrrhotite is rich in sulphur (between 39 and 40 wt.%), and the range of nickel compositions is near the experimentally determined nickel grades of 8 and 3 wt.% for the low and high sulphur additions, respectively. In order to gain further insight into the reaction mechanisms, samples with 100 kg sulphur/tonne of ore were reacted at different temperatures to determine the impact of reaction temperature on the nickel extraction and grade as measured using the bromine– methanol leach. Samples with this sulphur addition were chosen as they provided a balance between nickel sulphidation degree and sulphide selectivity. Temperatures between 450 and 650 °C were examined. Fig. 5 shows the nickel and iron extraction as a function of temperature. There is a sharp increase in nickel extraction between 500 and 550 °C, after which the nickel extraction exhibits little change. Fig. 5 also shows the corresponding grade of the sulphides formed, and indicates an increase in nickel grade with increasing reaction temperature. At the larger sulphur addition the two phases detected are pyrrhotite and magnetite (as presented in Fig. 6). The peaks for the pyrrhotite are shifted slightly

from those for pure iron pyrrhotite and the overall lattice spacing for both samples remain similar to the values determined at 500 °C. This overall increase in grade results from three individual contributions. The first is the increasing nickel recovery with increasing temperature, the second is due to the minor decrease in iron recovery with increasing temperature, and the third is due to the decrease in sulphur content of the calcine with increasing temperature.

3.2. TGA/DTA with EGA Fig. 7 shows the TG results for ore reacted with variable amounts of sulphur under a nitrogen atmosphere. For all of the tests, excluding the sample with no sulphur addition, there are three separate mass loss stages. These individual stages are more evident when the evolved gas is examined and this is shown in Fig. 8. Each of the three reaction stages will be discussed in more detail below. The first mass loss occurs between 250 and 360 °C and is associated with the dehydroxylation of the goethite, sulphidation of the

Fig. 4. X-ray diffraction spectra for two different ore–sulphur mixes reacted at 500 °C (Cu Ka) (Py = pyrite, Po = pyrrohtite, H = haematite).

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

5

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 6. X-ray Diffraction spectra for two sulphur/ore ratios of 100 kg/tonne and 700 kg/tonne and reacted at 550 °C (Cu Ka) (Po = pyrrohtite, H = haematite, M = magnetite).

Fig. 7. Mass loss as a function of temperature and sulphur addition.

oxides species and evaporation of sulphur. The gaseous species were determined through the use of evolved gas analysis. Gas peaks for both low and high sulphur additions are shown in Figs. 9 and 10, respectively. The overall ion currents were normalised relative to the maximum ion current in order to allow for the plotting of all three peaks on the same graph. The presence of m/z (mass to charge ratio) of 18 and 64 indicate water and SO2/S2 gas evolution and the presence of m/z of 34 indicates hydrogen sulphide evolution. During this initial sulphidation and dehydroxylation, the crystalline water is removed and the iron oxide is partially converted into a combination of pyrrhotite and pyrite. This is in accord with the furnace results discussed

below, and the following chemical reactions written in general terms for varying stoichiometry:

2Fe1x Nix O1x ðOHÞ1þx ! ð1  xÞFe2 O3 þ 2xNiO þ ð1 þ xÞH2 O RI

RI

Fe2 O3 þ bFe S ¼ cRI Fe Fed1 S þ dFe SO2 R1

R1

NiO þ bNi S ¼ cR1 Ni Nid1 S þ dNi SO2

ð1Þ ð2Þ ð3Þ

The values of the coefficients will depend upon the overall metal to sulphur ratio of the formed sulphide.

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

6

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

Ion Current of m/z 64 per gram of Sample

7x10-9 25 kg S 100 kg S 500 kg S 1000 kg S

6x10-9

5x10-9

4x10-9

3x10-9

2x10-9

1x10-9

0 200

300

400

500

600

700

800

900

1000

Temperature ( oC) Fig. 8. Evolved gas analysis for ore mixed with the indicated amounts of sulphur.

50 45 m/z=18 m/z=34 m/z=64

0.9 0.8

40 35

Mass Loss (%)

Fraction of Maximum Ion Current

1.0

0.7 0.6 0.5 0.4

25 20 15

0.3

10

0.2

TG Data Predicted from Isothermal Furnace Runs

5

0.1 0.0 150

30

0 0 200

250

300

350

400

200

400

600

800

1000

1200

Sulphur Addition (kg S/tonne ore)

Temperature (oC) Fig. 11. TG mass loss for low temperature reaction as a function of sulphur addition.

Fig. 9. Evolved gases during the initial reaction for 100 kg S/tonne ore.

Fig. 11 shows the mass loss during the initial reaction as a function of sulphur addition. The predicted mass loss based upon assay results from the isothermal tube furnace runs are also shown for comparison. Both the predicted mass loss during the isothermal runs and the non-isothermal runs in the TG show similar mass loss, with the furnace runs showing a slightly higher mass loss. This increased mass loss in the furnace runs occurs between 450 and 520 °C. This second mass loss is more evident in the evolved gas analysis results than on the TG signal. The magnitude of the mass loss and the SO2 evolution depends on the sulphur addition, with higher sulphur additions resulting in an increased mass loss. The reaction that occurs is the decomposition of pyrite to pyrrhotite:

Fraction of Maximum Ion Current

1.0 0.9 0.8

m/z=18 m/z=34 m/z=64

0.7 0.6 0.5 0.4 0.3 0.2

2:32FeS2 ¼ 2:65Fe0:877 S þ S2 ðgÞ DGo450 ¼ 49:2 kJ=mol S2

0.1 0.0 150

200

250

300

350

400

Temperature ( oC) Fig. 10. Evolved gases during the initial reaction for 1000 kg S/tonne ore.

ð4Þ

The standard Gibbs Free Energy change for this reaction is positive (Roine, 2006), but in the presence of haematite, this reaction can proceed spontaneously at temperatures above 450 °C. The gaseous sulphur produced from this reaction further reacts with the iron and nickel oxide to produce more sulphides if the temperature

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

7

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

is at or below 500 °C or if the temperature is above 500 °C it will react with the haematite to form magnetite, according to the following reaction:

FeS2 þ 6:5Fe2 O3 ¼ Fe:877 S þ 4:37Fe3 O4 þ SO2 ðgÞ

ð5Þ

This is in agreement with the XRD results and furnace test results, which show that for 100 kg S per tonne of ore, a mixture of pyrite/pyrrhotite/haematite exists at 450 °C, pyrrhotite/haematite at 500 °C and pyrrhotite/magnetite at 550 °C and it is also in agreement with the iron oxide sulphidation experiments by Bolsaitis and Nagata (1980) who found that reduction of haematite to magnetite occurred prior to iron sulphidation at temperatures of 550 °C or above, in a sulphidizing environment. A thermodynamic analysis of the sulphidation process by Harris (2011) also supports these reaction mechanisms. The third mass loss, which is evident on both the TG curve and evolved gas curve, occurs between 550 °C and 700 °C. The mass loss curves and evolved gas analysis indicate further release of sulphur dioxide, as pyrrhotite reacts with residual haematite to form magnetite according to the following reaction written in terms of a variable pyrrhotite stoichiometry:

present as an oxide. Furnace assay results show that the nickel recovery increases by 20–30% over the range of 450–550 °C with a more dramatic increase coinciding with the onset of magnetite formation at temperatures above 500 °C. One major reason for this is the creation of porosity and additional surface area during the conversion of haematite to magnetite. In order to further understand the nickel sulphidation mechanism, the ore and sulphur were mixed in varying proportions and initially reacted at 450 °C, and then allowed to further react at higher temperatures. These experiments provided information regarding the quantity of nickel oxide, which was converted to sulphide via the secondary sulphidation reactions. At higher temperatures there is a decrease in iron present as sulphide as a result of the following solid-state reaction:

NiFe2 O4 þ FeS ¼ NiS þ Fe3 O4

ð7Þ

The overall mass loss and relative magnitudes of the evolved gas analysis relative to the sulphur addition, agree with this mechanism. The peak with the greatest magnitude corresponds to the lowest sulphur addition, as there is more residual haematite when compared to the higher sulphur additions, which will have less haematite to reduce.

This exchange reaction allows for improved selectivity at higher temperatures, as the system is better able to approach thermodynamic equilibrium. In order to quantify the increased selectivity that could be achieved, samples were prepared with various sulphur additions and reacted at 450 °C for 1 h, followed by reaction at 1000 °C for 1 h. The nickel grades and extraction as a function of sulphur addition are shown in Fig. 12. There is a substantial improvement in the grade and extraction, compared with lower temperatures. This suggests an optimal range of sulphur additions of between 65 and 100 kg sulphur/tonne of ore, which allows a nickel grade of approximately 20% while keeping the nickel conversion at a relatively high value of 80%. The drop in conversion degree at nickel grades of between 25% and 30% also agrees favourably with the thermodynamic predictions (Harris, 2011).

3.3. TGA mass balance model

3.5. Phase transformations, melting and particle growth

In order to provide additional support for the proposed reaction mechanisms, a mass balance was created for the three individual reaction sequences outlined above. To provide additional information about the extent of iron sulphidation and the reduction of haematite to magnetite, the samples were dead roasted in oxygen following the third reaction. The weight change during the dead roast provides information regarding the relative proportion of iron sulphides to oxides. A regression routine was written in MATLAB™ to calculate the mass loss for the above reaction mechanisms, based on the assay results from the tube furnace tests, the detailed procedure is described elsewhere (Harris, 2011). Table 4 shows a comparison of the mass loss predicted from the MATLAB routine, using the reaction scheme above, and the measured mass loss from the TGA. The results are in good agreement.

The sulphide particles formed at lower temperatures were submicron in size, which made physical characterisation difficult but more importantly, precludes any possible physical liberation and subsequent separation. In order to promote particle growth, a series of studies were performed at higher temperatures in order to assess the impact of temperature on particle size and also to determine which additional phase transformations and reactions where possible at these temperatures. The growth of sulphides from submicron to larger sizes can occur either via solid state diffusion mechanisms due to standard particle coarsening (Porter and Easterling, 2004), or through a liquid phase mechanism (Diaz et al., 1993). The driving force for this process is attributed to the reduction in surface energy of the material. The formation of a liquid phase, with the associated higher diffusivities, could provide more rapid growth as compared to the relatively slow solid-state diffusion mechanisms. In order to study the effect of temperature, briquettes were first subjected to sulphidation at 450 °C for 1 h, followed by cooling and then reintroduced into the hot zone of the furnace at temperatures between 950 and 1100 °C for times ranging from 1 to 6 h. Table 5 shows the chemical compositions of the various phases formed under these conditions. Examination of the evolution of the

R3

R3

Fed S þ bM Fe2 O3 ¼ cR3 M Fe3 O4 þ dM SO2

ð6Þ

3.4. Nickel selectivity and exchange reactions While much of the focus has been on the conversion of iron and the associated mineralogy and chemical reactions, the speciation of nickel is also largely impacted by these reactions as well. During the initial sulphidation reactions at lower temperatures only 40– 50% of the nickel is converted to a sulphide, the remainder is still

Table 4 Comparison of measured mass loss vs. predicted mass loss from a mass balance. kg S/tonne ore

Wt loss 1 Wt loss 2 Wt loss 3 Dead roasta a

100 kg S

200 kg S

500 kg S

1000 kg S

Measured

Predicted

Measured

Predicted

Measured

Predicted

Measured

Predicted

16.63 1.81 2.45 1.58

17.42 2.08 2.63 1.92

22.76 2.15 2.01 0.86

22.33 2.45 2.18 0.91

35.78 3.51 1.36 0.25

35.6 3.56 1.73 0.22

48.09 4.71 0.87 1.81

47.65 4.53 0.99 1.63

Dead roast weight is taken as the difference between mass after reaction 3 and the final weight after 30 min dead roast at 1100 °C.

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

8

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

35

100

80 70

25

60 50

20

40 30

15

20 Ni Grade Ni Recovery

10 40

Ni Recovery (%)

Ni Grade of Sulphide t%)

90 30

50

60

10 70

80

90

0 110

100

Sulphur Addition (kg S/tonne ore) Fig. 12. Nickel sulphide grade and recovery as a function of sulphur addition.

Table 5 Bulk composition of the sulphides studied at 1000 °C. Sulphur Addition

Wt.% sulphide

Wt.% Ni in sulphide

Wt.% Fe in sulphide

Wt.% S in sulphide

100 kg S/ tonne ore 65 kg S/ tonne ore

8.9

14.8

53.8

31.4

5.8

21.5

43.9

34.5

microstructures reveals that both the growth of the sulphide phase and the magnetite grains can be explained using liquid phase sintering theory.

For temperatures below 950 °C the microstructure is shown in Fig. 13. The structure is quite porous with much of the original particle structure still intact. This porous structure provides evidence of the gaseous products released during sulphidation and reduction. The sulphides are the fine reflective phase distributed throughout the darker matrix (magnetite). The small particles and many voids increase the overall energy of the system owing to both the magnetite/vapour interfaces and the sulphide/magnetite interfaces. Despite this high energy, there is little evidence of sintering or particle coarsening. SEM examination also revealed no evidence of well-defined magnetite grains. The formation of a Fe–Ni–S–O liquid above 950 °C has a large impact on the microstructure of the sample. Evidence of this liquid phase formation is provided by the DTA peaks, which indicated that melting occurred between 960 and 1010 °C, depending upon the sulphide composition. The initial liquid formation creates strong capillary forces between neighbouring particles, drawing the particles together and rapidly increasing the density of the pellet. This was also supported by visual observation of the pellets and volume measurements, which showed rapid densification above this temperature. The magnetite forms a much more cohesive and connected structure when compared with the structure at 950 °C. The pores that remain are much larger in nature and the sulphides are also larger. SEM analysis revealed that while the overall density has improved, many of the initial reacted particles remain largely intact. There is also evidence of very fine magnetite grains beginning to nucleate within several particles. Following the rearrangement of the particles due to liquid capillary forces, the particles that are connected via liquid bridges provide a means of grain growth, as large grains grow at the expense of smaller grains. Evidence of grain growth is provided by examination of Fig. 13, which shows the etched grain structure of samples reacted at 1050 and 1100 °C. Well-defined magnetite

950 C

1000 C

1050 C

1100 C

Fig. 13. Optical micrograph of etched samples reacted at various temperatures for 1 h (scale bar is 10 lm).

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

9

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

grains are located throughout the microstructure with the sulphides existing at the grain boundaries. Samples were also reacted for 3 and 6 h and the microstructure was similar but with larger grains and sulphide particles. Further evidence for solute reprecipitation as a means of grain growth is provided by a comparison of the shape of the grains throughout the sample compared with shapes predicted by the different mechanisms occurring during liquid phase sintering. Three classical signs of grain growth, as outlined in German et al. (2009) are shown in Fig. 14 (labelled a–c). The first, (a) involves contact flattening and results from the dissolution of the spherical cap of the grain and transport of material to the edges, the second, (b) involves the dissolution of small grains and reprecipitation of the material on larger grains, the third, (c) involves solid state bonding resulting from contact of two particles and transport to the particle edges. Longer heating times also showed evidence of mass movement of sulphides through the pellet with drainage of the sulphides towards the lower portion of the pellet. Presumably as the liquid coalesced, the forces due to gravity overcame the capillary forces and allowed mass movement throughout the particles.

Table 6 Sulphide phases identified in samples heated at 1050–1150 °C.

a

Phase

Ni (wt.%)

Fe (wt.%)

S (wt.%)

Troilite Matte Metallicsa

2.11 ± 0.51 19.39 ± 7.23 46.5

60.25 ± 0.80 45.95 ± 4.8 41.0

35.05 ± 0.90 30.68 ± 2.19 13

Only 1 sample was found that was large enough for microprobe analysis.

3.6. Crystallization path and sulphide compositions To this point, the sulphides have been treated as homogenous in chemical composition. SEM backscattered imaging however reveals that there is chemical gradient within the solidified sulphides. Chemical variation within the sulphides was first identified using backscattered imaging followed by composition determination via microprobe. Due to the small particle size compared with the beam of the microprobe, line scanning was not possible and on the smallest zones there was also some overlap between chemical zones. Three distinct zones were identified and their composition is shown in Table 6. These phases also agree well with the XRD data from samples prepared and reacted at 1000 °C. Troilite, magnetite and an amorphous phase were identified. The different chemical phases are clearly visible in a backscattered micrograph and are identified in Fig. 15. Their formation can be explained by examining the crystallization path of the sulphides buffered with magnetite. Initially at 1100 °C, the entire liquid is composed of a homogenous oxygen-enriched matte. Thermodynamic calculations using the model of Kress (2007) show an overall oxygen solubility in the liquid of 10–12 wt.% (Harris, 2011), which

Fig. 14. Backscattered SEM micrograph of reacted sample. The classical signs of grain growth are. labelled (a), (b) and (c) and are discussed in the text.

Fig. 15. Micrograph showing various sulphide phases formed during solidification (1 – large stable magnetite grain, 2 – troilite, 3 – secondary magnetite, 4 – nickel rich matte, 5 – magnetite).

agrees favourably with the 10 wt.% value on the Fe–O–S phase diagram at 1120 °C (Naldrett, 1969). As the sample cools, the solubility of oxygen decreases slightly requiring the precipitation of magnetite. This decrease in oxygen solubility is small and can be accommodated by precipitation on existing magnetite particles. As the liquid continues to cool, it becomes saturated with pyrrhotite, which begins to precipitate from solution. The onset of pyrrhotite precipitation is strongly influenced by the sulphur content of the liquid, with sulphur rich mattes beginning to precipitate pyrrhotite at 1050 °C whereas sulphur deficient mattes will favour precipitation of troilite between approximately 930–1000 °C (Naldrett, 1969). This precipitation of troilite also requires much more rapid precipitation of magnetite to keep the oxygen solubility of the remaining liquid in equilibrium. Much of this magnetite precipitation can be accommodated by precipitation on existing sites, but there is evidence of some homogenous nucleation of magnetite precipitation in some of the larger troilite particles. The pyrrhotite that initially precipitates is low in nickel content. This agrees well with the Fe–Ni–S phase diagram (Waldner and Pelton, 2004), which shows a low nickel content pyrrhotite in equilibrium with a metal-rich matte phase. This solidification of pyrrhotite also requires an oxygen adjustment in the liquid phase, as the increased amount of nickel relative to iron reduces the solubility of oxygen in the matte. The remaining nickel-rich matte will continue to exsolve magnetite as it cools until there is almost no oxygen present, instead a metal rich matte will exist. Evidence of this additional rejection of magnetite can be found at the interfaces between the troilite and the nickel-rich matte (see points 5a and 5b in Fig. 15). The overall remaining composition falls in the ferronickel, matte, troilite phase field. Ferronickel is the first to precipitate and can precipitate on existing solids as well as via homogenous nucleation within the matte. The last phase to solidify is the nickel rich matte.

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

10

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

Table 7 Residual nickel within the magnetite phase.

8 1050 °C 0.13 ± 0.022 91 ± 1.5 86 ± 4.2

Cumulative Ni Grade in Concentrate (%)

7

6

Total Ni Recoverable Ni

6

FT01 FT02 FT03 FT04 FT05

7

1100 °C 0.21 ± 0.027 86 ± 2.5 81 ± 4.2

Ni Grade (%)

Reaction temperature % Ni in spinel Predicted proportion of nickel in sulphide Proportion of nickel determined by leach

100 kg S/tonne ore

5 4 3 2

5

1

4

0 0

2

4

6

8

10

12

14

S Grade (%)

3

Fig. 18. Relationship between nickel grade and sulphur grade for the flotation tests for 65 kg S/tonne ore and 100 kg S /tonne ore.

2 Head grade to float cell

1 1.6 0 0

10

20

30

40

50

60

FT03 FT04

1.4

Gangue Recovery (%)

Cumulative Recovery (%) Fig. 16. Cumulative grade recovery curve for flotation test FT04.

Cumulative Recoverable Ni Recovery (%)

65 kg S/tonne ore

60

50

1.2 1.0 0.8 0.6 0.4 0.2

40

0.0 0

30

1

2

3

4

5

6

7

8

Sulphur Recovery (%) 20

Fig. 19. Gangue recovery as a function of sulphur recovery.

10

on the total nickel content of the ore, while the recoverable nickel is based on the amount of nickel sulphide present after sulphidation. The overall economics and suitability of the process relies on the total nickel recovery, whereas the flotation performance should be assessed on the recoverable nickel, which is approximately 80% of the total nickel. The flotation results indicate that the flotation process is selectively upgrading the nickel sulphides, with the highest-grade material achieved at the beginning of the test. Then, the grade remains nearly constant until decreasing near the end of the test. In order to verify that the sulphides were floating as a bulk sulphide and not as a separate matte or troilite phase, the cumulative nickel recovery was plotted against cumulative sulphur recovery and this is shown in Fig. 17. If the sulphide floats as a bulk sulphide, and not as the separate phases outlined in the preceding sections then the relationship should be linear with a slope of unity. The excellent fit and a slope very close to one confirms bulk sulphide flotation. Similar results were achieved for the other flotation tests. The overall impact of the different feed compositions on the relationship between nickel grade and sulphur grade is shown in Fig. 18. The shift towards a more nickel-rich sulphide with lower sulphur additions is evident in the flotation results, with the same nickel grade being achieved with the lower sulphur grade, when compared to the higher sulphur addition.

0 0

10

20

30

40

50

60

Cumulative S Recovery (%) Fig. 17. Cumulative nickel recovery as a function of cumulative sulphur recovery for flotation test FT04.

This crystallization scheme agrees with the microstructure as well as the available phase diagrams (Naldrett, 1969; Raghavan, 2000). Table 7 shows the nickel content in the spinel as determined using the microprobe, along with the predicted recovery of nickel to the sulphide based upon the initial head grade. The recovery, based on residual nickel in magnetite, agrees well with the results from the diagnostic leach.

3.7. Flotation studies Fig. 16 shows a typical grade recovery curve from the metered reagent test FT04, showing both absolute nickel recovery and the recoverable nickel recovery. The absolute nickel recovery is based

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016

C.T. Harris et al. / Minerals Engineering xxx (2013) xxx–xxx

In order to determine the reason for the overall low amount of concentrate it is useful to examine the nature of both the nickel and iron recoveries as a function of sulphur recovery. The flotation of magnetite/sulphide assemblages would be indicated by the correlation between the iron recovery and the sulphur recovery, whereas if the low grade is due to entrainment, there should be little relation between sulphur recovery and iron recovery. This becomes more evident if the amount of iron associated with sulphide is estimated based on the bulk composition determined by chemical assay. The remaining mass in the concentrate is then assumed to be gangue. Gangue recovery versus sulphur recovery is shown in Fig. 19 where the weak correlation is more evident. 4. Conclusions High temperature experiments demonstrated that the selective sulphidation of nickel from a lateritic ore is achievable and that the overall extent of conversion and grade of nickel sulphide is strongly dependent upon both the sulphur addition and the reaction temperature. Bulk sulphide grades of between 15 and 20 wt.% nickel were achievable, with a corresponding nickel sulphidation degree of about 80%. Decreasing the sulphur addition to 55 kg S per tonne of ore allowed for an increase in nickel grade to 30%, but resulted in a significant drop in the nickel conversion. The sulphides, which formed at temperatures below about 950 °C, were submicron in size and were not amenable to physical upgrading. At 950–1050 °C it was found that an oxygen-enriched liquid matte phase could form. This resulted in rapid densification of the briquette, due to the capillary forces created by liquid bridging and also provided a means for both magnetite and sulphide growth via a liquid phase sintering mechanism. During cooling, the liquid phase solidified into three distinct nickel bearing phases: heterogeneous nucleation of iron oxide on existing iron oxide particles, precipitated iron-rich pyrrhotite containing 2–3 wt.% nickel and a nickel rich-sulphur deficient matte. Submicron ferronickel particles were also found to have precipitated within the matte. It was found that the nickel sulphides responded to flotation at pH 9 with PAX and TETA additions and that maximum grades of between 6 to 8% were achieved with the average cumulative grade of the nickel concentrates having values of between 4 and 5 wt.% nickel. Cumulative recoveries were low, with values between 50% and 60% on a sulphide basis, which corresponded to a 35–45% recovery on a total nickel basis. An analysis of the correlation between nickel, sulphur and iron grades revealed that the concentrate floated as a bulk sulphide. The low grade of nickel in the concentrate was due to fine particle entrainment and not due to liberation issues. The low recovery of nickel may be due to issues with fine particle grinding and flotation.

11

Acknowledgements The authors thank the Natural Sciences and Engineering Research Council of Canada and Xstrata for funding this research. References Bolsaitis, P., Nagata, K., 1980. Kinetics of sulfidization of iron oxide with SO2–CO mixtures of high sulfur potential. Metallurgical Transactions B 11 (2), 185–197. Bulatovic, S., 2007. Handbook of Flotation Reagents: Chemistry, Theory and Practice. Elsevier, Amsterdam. Chen, J., Jak, E., Hayes, P.C., 2009. Factors affecting nickel extraction from the reduction roasting of saprolite ore in the Caron Process. In: Proceedings of the International Symposium on the Pyrometallurgy of Nickel and Cobalt 2009, pp. 449–462. Dalvi, A.D., Bacon, W.G., Osborne, R.C., 2004. The past and future of nickel laterites. In: PDAC 2004 International Convention, Trade Show and Investors Exchange, March 7–10, pp. 1–27. DeGraaf, J.E., 1979. The treatment of lateritic nickel ores – A further study of the Caron Process and other possible improvements. Hydrometallurgy, 47–65. Diaz, C.M., Vahed, A., Shi, D., Doyle, C.D., Warner, A.E.M., MacVicar, D.J., 1993. Low temperature thermal upgrading of lateritic ores. United States Patent 178 (5), 666, 12th January. German, R.M., Suri, P., Jin Park, S., 2009. Review: liquid phase sintering. Journal of Materials Science 44 (1), 1–39. Harris, C.T., 2011. The Selective Sulphidation and Physical Upgrading of Nickel from a Nickeliferous Lateritic Ore. Ph.D. Thesis. Queen’s University, Kingston, Ontario, Canada. Harris, C.T., Peacey, J.G., Pickles, C.A., 2011. Selective sulphidation of a nickeliferous lateritic ore. Minerals Engineering 24 (7), 651–660. Kelebek, S., Tukel, C., 1999. The effect of sodium–metabisulfite and triethylentetramine system on pentlandite–pyrrhotite separation. International Journal of Mineral Processing 57 (2), 135–152. Kress, V., 2007. Thermochemistry of sulfide liquids III: Ni-bearing liquids at 1 bar. Contributions to Mineralogy and Petrology 154, 191–204. Landers, M., Gilkes, R.J., Wells, M., 2009a. Rapid dehydroxylation of nickeliferous goethite in lateritic nickel ore: X-ray diffraction and tem investigation. Clay and Clay Minerals A57 (6), 751–770. Landers, M., Gilkes, R.J., Wells, M., 2009b. Dissolution kinetics of dehydroxlayated nickeliferous goethite from limonitic lateritic nickel ore. Applied Clay Science 42, 615–624. Li, Q.H., Cui, Y., Zhu, D.Q., Zhu, J.H., Pan, J., Zhang, H.L., Zheng, G.L., 2010. Study on selective reduction and magnetic separation of low grade nickel laterite ore to produce high nickel concentrate. In: XXV International Mineral Processing Congress. AUSIMM, Brisbane, pp. 1549–1556. Mudd, G.M., 2009. Nickel sulfide versus laterite: the hard sustainability challenge remains. In: Proc. 48th Annual Conference of Metallurgists. Canadian Metallurgical Society (August). Mudd, G.M., 2010. Global trends and environmental issues in nickel mining: sulfides versus laterites. Ore Geology Reviews 38, 9–26. Naldrett, A., 1969. A portion of the system Fe–S–O between 900–1080 C and its application to sulfide ore magmas. Journal of Petrology 1 (2), 171–201. Naldrett, A., Craig, J.R., Kullerud, G., 1967. The central portion of the Fe–Ni–S system and its bearing on pentlandite exsolution in iron–nickel sulfide ores. Economic Geology 62 (6), 826–847. Porter, D.A., Easterling, K.E., 2004. Phase Transformations in Metals and Alloys, second ed. CRC Press, Boca Raton. Raghavan, V., 2000. Fe–Ni–S (iron–nickel sulfur). Journal of Phase Equilbria and Diffusion 25 (4), 373–381. Roine, A., 2006. HSC Chemistry 6.0 – User’s Guide. Outokumpu Research Oy, Pori, Finland. Waldner, P., Pelton, A., 2004. Critical thermodynamic assessment and modelling of Fe–Ni–S system. Metallurgical and Materials Transaction B 35 (5), 897–901. Zhu, D.Q., Cui, Y., Veining, K., Hapugoda, S., Douglas, J., Pan, J., Zheng, G.L., 2012. Upgrading low nickel content laterite ores using selective reduction followed by magnetic separation. International Journal of Mineral Processing, 1–7.

Please cite this article in press as: Harris, C.T., et al. Selective sulphidation and flotation of nickel from a nickeliferous laterite ore. Miner. Eng. (2013), http:// dx.doi.org/10.1016/j.mineng.2013.02.016