Experimental leaching results can be expressed by mathematical models. The models describing kinetics of leaching reactions can be grouped based on their shape in four different categories: acceleratory, deceleratory, linear or sigmoidal (Khawam and Flanagan, 2006). In linear models the reaction rate remains constant while in acceleratory and deceleratory models the rate increases or decreases as the reaction proceeds. Sigmoidal models present the relationship between the reaction rate and the conversion factor, α through a shape of the letter S.
In the comprehensive review made by Li et al. (2013) several leaching mechanisms and the kinetic equations used in modeling have been presented.
Khawam and Flanagan (2006) have described in their review the basics and mathematical fundamentals of kinetic models and Brown et al. (1980) have discussed reaction rate equations in more detail. A collection of the mechanisms and kinetic equations of leaching chalcopyrite in different leaching media is
shown in Table V in which k is the reaction rate constant, t is the time taken for leaching and α is the extent of the reaction. The extent of reaction is calculated by dividing the mass of the dissolved valuable at time t by the initial mass of the valuable.
TABLE V Different kinetic equations used to describe the leaching mechanisms in different media (1Brown et al., 1980; 2Khawam and Flanagan, 2006; 3Li et al., 2013,
4Kabai, 1973).
Eq. No. Kinetic equation Mechanism
(20) kt = α2 One dimensional diffusion1 (21) kt =(1-α)ln(1-α)+α Two dimensional diffusion1,2 (22) kt = [1-(1-α)1/3]2 Three dimensional diffusion2,3 (23) kt = [1-(2α/3)]-(1-α)2/3 Three dimensional diffusion1,2
(24) kt = 1-(1-α)1/2 Surface reaction control, contracting area, by SCM*1
(25) kt = 1-(1-α)1/3 Surface reaction control, contracting volume, by SCM*1,3
(26) kt = [-ln(1-α)]1/2 Random nucleation1,2 (27) kt = [-ln(1-α)]1/3 Random nucleation1,2 (28) kt = [-ln(1-α)]1/4 Random nucleation1,2
(29) kt = ln[α/(1-α)] Random nucleation, 1st order1,2
(30) kt = α1/n Constant nuclei growth by multistep nucleation1,2
(31) kt = lnα Constant nuclei growth by single-step nucleation1,2
(32) kt = -ln(1-α) First order nucleation1,2,3 (33) kt = (1-α)-1 Second order nucleation1 (34) kt = (1-α)-2 Third order nucleation1
(35) alnk+alnt= lnln(1/(1-α)) Surface reaction control or diffusion4 *SCM=Shrinking core model
Reaction rate models can also be divided into groups by their mechanistic assumptions (Khawam and Flanagan, 2006). Reactions taking place on the solid-liquid interface can be described by diffusion (equations 20-23), geometrical contraction (equations 24 and 25), nucleation (equations 26-31) or reaction order
(equations 32-34). Equation 35 describes either surface reaction control or diffusion, depending on the constant a.
Investigations show that the controlling factor in a leaching reaction would be the diffusion through a layer of elemental sulfur (Harmer et al., 2006; Aydogan et al., 2006). Diffusion can be described in either one, two or three dimensions. One dimensional diffusion is represented by parabolic graphs. Two dimensional refers to the diffusion for a cylindrical particle whereas three dimensional diffusion refers to the diffusion for a spherical particle. The two 3D-diffusion models, equation 22 by Jander (1927) and 23 by Ginstling and Brounshtein (1950) are both derived from the shrinking core model (equation 25).
Reactions that are described by geometrical contraction are based on the shape of the particles. A solid particle can be assumed having a cylindrical or spherical shape and according to the assumption either equation 24 or 25 can be used. The model used is more commonly called Shrinking core model (Levenspiel, 1999) and is quite often used when describing leaching behavior. As the model is based on the shrinking of a disc (contracting area) or the shrinking of a sphere (contracting volume) particle size has a great effect on the reaction rate. The particle size can be taken into account by inserting ko=k/ro in which ko is the reaction rate coefficient taking the radius of a particle ro into consideration.
Emphasis should thus be made that actually equations 24 and 25 have particle size included in the rate constant (Khawam and Flanagan, 2006) and hence sieving the sample before experimenting is recommended in order to achieve reliable results that are not affected by particle size. For samples having variable particle size the reaction rate curves might be shifted.
Many solid-state reactions can be described by nucleation models in which a new product phase is formed at the reaction sites. Nucleation is described as either single or multistepped. As can be assumed, in single-step nucleation nuclei growth and nucleation both happen during a single step whereas multistep nucleation takes place in several steps in order for the nucleus to grow. The most
known nucleation models are the Avrami models (Equations 26-28), which describe random nucleation.
Models based on the reaction order are similar to the models used in homogenous reactions and hence they are the simplest models (Khawam and Flanagan, 2006).
The reaction rate is proportional to the amount, concentration or fraction of the remaining reactants raised to a particular power as shown in equation 36. This is the case in the equations 32-34 in Table V for which n=1-3 based on the reaction order. By integrating the equations formed by equation 36 and n the equations in Table V can be obtained. The first-order model is more commonly known as the Mampel model (Khawam and Flanagan, 2006) and all the reaction order based models are special cases of the Avrami-Erofe’ev model (equations 26-28).
( )
ndt k
dα = −α
1 (36)
The Kabai (1973) model shown in equation 35 describes both diffusion and surface reaction control depending on the constant a. By drawing lnln(1/(1-α)) versus lnt the coefficients a and k can be found out. When a<1 the reaction is diffusion controlled whereas if a≥1 the rate determining step is the rate of chemical reaction.
The leaching of chalcopyrite, pentlandite and some other samples has been investigated and models that best describe the leaching behavior have been suggested. An overview of the suggested models is shown in Table VI.
TABLE VI Kinetic dissolution models suggested by previous researchers for the dissolution of copper and nickel samples in different leaching agents.
Leaching
1 Sulfides: Ar=arsenopyrite (FeAsS), Cat=catierite, Cp=chalcopyrite, Lin=linnacite, Mil=millerite (NiS), Pe=pentlandite, Pr=pyrrhotite, Py=pyrite
2 through a porous product layer
3 Cobalt sulfide minerals: catierite (CoS2) and linnacite (Co3S4)
4 SCM=Shrinking core model
5 PROCESSING AT THE CASE MINE
At the multimetal mine about 3.138 million tons of ore was processed and 7.460 million tons ore was mined in year 2012. Mainly two sulfide concentrates are produced from which mainly two valuable minerals are obtained, nickel and copper. The difference between the sulfide concentrates is that the first contains nickel, copper and PGEs whereas the second only contains copper and PGEs. The ore also contains cobalt in minor amounts, but cobalt is not yet being recovered from the process.
The ore deposit contains the following sulfide minerals: pentlandite ((Fe,Ni)9S8), pyrite (FeS2) and chalcopyrite (CuFeS2). Also pyrrhotite (Fe1-x(x=0-0.17)S), millerite (NiS), heazlewoodite (Ni3S2), nickeline (NiAs), maucherite (Ni11As8) and
gersdorffite (NiAsS) are found in small amounts in the ore deposit. Main host minerals that in the end are cumulated as waste rock are olivine (Mg,Fe)2SiO4 and different silicate minerals of clinopyroxenes and orthopyroxenes. The platinum group elements can be found in grains of amphibole (silicate mineral), serpentinite (hydrous silicate mineral containing magnesium and iron) or chlorite (silicate mineral containing magnesium, iron, nickel and manganese) of size smaller than 75 mm. In most cases these grains are located in the rocks next to sulfides (in 38
% of the cases), inside the sulfides (6 %) or then the grains fill the holes of pentlandite minerals.
The mined ore undergoes a series of comminution steps before entering the concentration process step. Particles of the size smaller than 75 µm are passed on to the concentration stage. During this concentration step copper and nickel are recovered by two separate flotation circuits, copper in the first circuit and nickel in the second. After flotation and washing, the concentrates are thickened, filtered and shipped to customers.
During the processing of ores at the mine, about 3.8-4.8 million tons of tailings are produced per year. Two different kinds of tailings are formed of which a major part are formed when concentrating nickel. Since the target in the nickel flotation circuit is to collect as much sulfide minerals as possible, the tailings formed at this stage have low sulfur content. About 3.5-4.3 million tons of low sulfur tailings and 0.4-0.5 million tons of high sulfur tailings are produced per year. The tailings produced during the first cleaning stage of copper and the rougher sized nickel concentrate are combined and reground before nickel concentrate is passed on to the washing stage. The tailings formed in the washing of nickel are of higher sulfur content than the tailings from the flotation of nickel. The low sulfur tailings contain about 0.4 % sulfur whereas the high sulfur tailings contain 7 % sulfur. The tailings are pumped into two different tailing ponds. A process flowsheet of the flotation circuits and the tailings formation is shown in Figure 3.
The two tailings contain different amounts of copper and nickel. The tailings having lower sulfur content contain 450 mg Cu/kg and 580 mg Ni/kg whereas the
high sulfur tailings contain 3160 mg Cu/kg and 4530 mg Ni/kg. In 2012, 4.322 million tons of tailings were produced of which about 3.5-4.3 million tons are of the low sulfur tailings. In the following calculations a yearly low sulfur tailings formation is assumed to be 4.0 million tons. During the expected mine life of 15 years in total 42.3 million kg of copper and 56.7 million kg of nickel is lost to tailings. With the prices for copper and nickel on 25th of October 2013 (London Metal Exchange, 2013a and 2013b) this would be worth all together 1.5 billion €.
Hence investing to the treatment of tailings and the recovery of nickel and copper from tailings is economically attractive.
FIGURE 3. Process flowsheet of the flotation circuits of the chosen mine.
36
EXPERIMENTAL PART
The objective of the experimental part was to find out the most suitable leaching agents to dissolve nickel and copper out of two different tailings samples. This was done according to the following steps:
1. Tailings characterization
2. Preliminary screening for leaching tests 3. Dissolution studies
4. Kinetic modeling
6 TAILINGS CHARACTERIZATION
Two different samples were obtained from the mine, one with a sulfur content of 7 g/kg and one with 85 g/kg. The tailings samples were first dried overnight in 105 °C according to SFS standard 3008 (1990) after which characterization techniques were applied.
The content of the valuables in the tailings samples was determined by an external laboratory. By inductively coupled plasma optical emission spectrometry (ICP-OES) the concentration of nickel, copper, iron and some other elements in both of the samples were determined and the values can be seen in Table VII.
TABLE VII The concentration of some elements in the two tailings samples measured by ICP-OES.
Sample Concentration, mg/kg
Cu Ni Fe Au Pd Pt S
low sulfur 429 1180 77300 0.05 0.07 0.11 6600 high sulfur 4466 11300 173000 0.34 0.64 0.84 84600
The particle size distributions were obtained using a Malvern Mastersizer 3000 particle size analyzer by using the Fraunhofer optical model. The particle size distributions are shown in Figure 4.
FIGURE 4. Particle size distributions for tailings samples. Barrel 1 having a lower sulfur content and 2 higher. Barrell 2 also contains a considerably higher amount of valuable metals.
Both distributions were unimodal, the peak of barrel 1 roughly being at 100 µm and of barrel 2 at 30 µm. Barrel 1 had slightly larger particles which can also be indicated by the d10, d50 and d90 values shown in Table VIII.
TABLE VIII Values for d10, d50 and d90 obtained from the particle size distribution.
Sample d10, µm d50, µm d90, µm
Barrel 1 2.6 30.9 138
Barrel 2 2.2 20.1 124
The shape of the particles of the two tailings samples were analyzed using a JEOL JSM-5800 scanning electron microscope (SEM) and a Malvern Morphologi G3 particle image analyzer. As is shown in Figure 5 and Figure 6 the solids consist of particles having very different shapes: for example rectangular, needlelike and elongated. The SEM images also reveal that the samples contain iron, which is shown by light particles in Figure 5.
10-2 10-1 100 101 102 103 104
0 0.5 1 1.5 2 2.5 3 3.5 4
Particle size, µm
Volume fraction, %
Barrel 1 Barrel 2
FIGURE 6. Different shapes of particles of the high sulfur tailings sample. Size of particles is not comparable in this picture.
The shape as well as the size of the particles plays an important role in the dissolution of valuables as was described in Chapter 2.1. However, a large part of the tailings sample is rock with no economic value and hence the tailings were also analyzed using a Bruker D8 Advance X-ray diffractometer (XRD) in order to find out in which mineralogical phase the valuables are present. The XRD data was recorded at a step size of 0.01°2θ and a scan speed of 0.02°2θ/s between 5 and 100°2θ. The data obtained from XRD is not however covered in this thesis because of the complexity of the tailings samples mineralogy.
7 LEACHING EXPERIMENTS
In order to find the most suitable leaching agents for further investigation, a series of leaching agents were chosen to be used in the preliminary leaching tests. Metal sulfides are typically rather insoluble and hence most of the leaching reagents were chosen according to previous research made on pentlandite and chalcopyrite leaching in order to assure the leaching of nickel and copper. In addition other
FIGURE 5. Scanning electron microscope images of the low sulfur tailings (left) and high sulfur tailings (right).
agents were chosen on the basis of probable leaching mechanism. Leaching was also studied in water and alkali.
The used leaching reagents during the preliminary tests were H2O, HNO3, H2O2, oxalic acid, citric acid, H2SO4, sodium dithionite, sodium thiosulfate, NaOH, a combination of citric acid and H2SO4 and a combination of H2SO4 with added Fe2(SO4)3. The leaching experiments conducted with a neutral agent were done in reverse osmosis (RO) water. H2SO4, HNO3 and H2O2 having purities of 95-97 %, 65 % and 30 %, respectively, were obtained from Merck. Also solid oxalic acid dihydrate powder with pro analysis grade, sodium thiosulfate powder with 97 % purity and sodium hydroxide pellets having a purity of 97 % were obtained from Merck. Sodium dithionite having a purity of 86 % and iron(III)sulfate hydrate having an iron content of 21-23 % were obtained from Sigma-Aldrich. For the preliminary leaching tests an analysis grade solid citric acid was obtained from Sigma-Aldrich. Due to availability problems with the citric acid used during the preliminary leaching tests, a solid citric acid having a purity of 99.5 % was obtained from Amresco (Leuven, Belgium) for the larger scale leaching tests. All solutions were prepared in RO water. A sulfuric acid purity of 96 % was chosen for calculations.
Preliminary leaching tests were conducted in 100 ml Erlenmeyer flasks with a solution volume of 100 ml and a concentration of 0.5 mol/dm3. The concentration of Fe2(SO4)3 was 0.2 mol/dm3. The scale was held small in order to screen many leaching agents. Both of the previously mentioned tailings samples were investigated. A tailings sample of 0.5 g was added and magnetic stirring was applied. Two samples were drawn after 24 hours of leaching using a syringe equipped with a syringe filter having a pore size of 0.2 µm. One sample was diluted using 14 vol% HNO3 in order to prevent any precipitation whereas the other sample was left undiluted to see if any precipitation occurs. Dilutions of sodium dithionite and sodium thiosulfate were sensitive to precipitation when nitric acid was added and hence the non-diluted samples were analyzed.
The quantity of dissolved Cu, Ni and Fe was analyzed from the samples using a Thermo Scientific iCE 3400 AA Spectrometer. Atomic absorption spectrometry, in this case flame AAS, is based on measuring the absorbance of light from the hollow cathode lamp by the atomized sample. For more information, Higson (2004) and Kivalo et al. (1976) give a more detailed description on the analytical technique of AAS.
The calibration standards used for Ni and Cu were 0.05, 0.1, 1 and 2 mg/dm3, whereas the calibration standards for Fe were 0.05, 1, 3, 5 and 7 mg/dm3. The low concentrations of calibration standards for Ni and Cu were chosen in order to detect even the smallest concentrations of leached valuables. All calibration standards were prepared in 14 vol% HNO3. Some samples needed further dilution and hence these were diluted with 14 vol% HNO3 in order to fit the concentrations of the valuables to the range of the calibration standards. Because of the small scale of the conducted experiments, the recoveries for each metal were not investigated. Instead the dissolution of the valuables was recorded, and leaching agents showing the most promising results were chosen.
Results of the preliminary leaching tests are shown in Table IX and Figure 7. The leaching agents that dissolved the largest quantity of nickel and copper were in decreasing order: H2SO4 with added Fe2(SO4)3, oxalic acid, H2SO4, the combination of H2SO4 and citric acid, sodium dithionite, citric acid and HNO3. Water, sodium hydroxide, sodium thiosulfate and sodium peroxide only recovered minor amounts of nickel and copper according to the preliminary leaching test.
TABLE IX The leaching ability of different leaching agents at room temperature for 24 hours.
Concentrations were 0.5 mol/dm3 for all leaching agents except Fe2(SO4)3 had a concentration of 0.2 mol/dm3. LS is representing low sulfur tailings samples and HS high sulfur tailings samples. A minus sign represents no leaching ability.
Leaching agent Leaching mechanism Cu Ni Fe LS HS LS HS LS HS
*Concentration of iron were not determined because of the addition of Fe2(SO4)3
Sulfuric and citric acid were chosen for further leaching studies. Sulfuric acid was chosen because of its rather low cost and other positive features described in Chapter 2.2. Citric acid was chosen instead of oxalic acid since citric acid leached a significantly lower amount of iron, even though the recoveries for nickel and copper were similar for both acids. The recovery of iron was not a target in this study and the dissolution of iron might even be a disadvantage when recovering nickel and copper from the leach solution. In addition a very promising previous investigation has shown the possibility of using citric acid (Hansen et al., 2005) and the combination of citric acid and H2SO4 when dissolving copper.
FIGURE 7. Preliminary leaching test results. conducted for 24 hours at ambient temperature and atmospheric pressure. The amount of nickel, copper and iron leached into a solution volume of 100 ml having a pulp density of 5 g/dn3 . The concentration of each leaching agent was chosen to be 0.5 mol/dm3 .