An investigation into the flotation of muscovite with an amine collector and calcium lignin sulfonate depressant

Accepted Manuscript An investigation into the flotation of muscovite with an amine collector and calcium lignin sulfonate depressant Christopher Marion, Adam Jordens, Sheelah McCarthy, Tassos Grammatikopoulos, Kristian E. Waters PII: DOI: Reference:

S1383-5866(15)00249-X http://dx.doi.org/10.1016/j.seppur.2015.04.025 SEPPUR 12307

To appear in:

Separation and Purification Technology

Received Date: Revised Date: Accepted Date:

18 October 2014 24 March 2015 13 April 2015

Please cite this article as: C. Marion, A. Jordens, S. McCarthy, T. Grammatikopoulos, K.E. Waters, An investigation into the flotation of muscovite with an amine collector and calcium lignin sulfonate depressant, Separation and Purification Technology (2015), doi: http://dx.doi.org/10.1016/j.seppur.2015.04.025

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

An investigation into the flotation of muscovite with an amine collector and calcium lignin sulfonate depressant Christopher Marion a, Adam Jordens a, Sheelah McCarthy a, Tassos Grammatikopoulos b, Kristian E. Watersa* a

Department of Mining and Materials Engineering, McGill University, 3610 University Street, Montreal, Quebec, Canada H3A OC5

b

SGS Canada Inc., 185 Concession Street, PO 4300 Lakefield, Ontario, Canada K0L 2H0

Author for Correspondence: Waters, K.E.; Department of Mining and Materials Engineering, McGill University, 3610 University Street, Montreal, Quebec H3A 0C5, Canada; Email: [email protected]; Tel: +1 514 398 1454

Abstract Muscovite mica is a naturally occurring phyllosilicate mineral with a wide range of industrial applications. This work investigated the separation of muscovite from unwanted gangue minerals in a mica ore. Electroacoustic zeta potential measurements and single mineral flotation tests were used to determine the effect of Custamine 8113 (collector) and Norlig-H (depressant) on muscovite. These results were then compared to those of gangue minerals (feldspar and quartz). Muscovite showed an isoelectric point (IEP) at pH 3.5. Feldspar and quartz did not have an IEP, having negative zeta potentials over the pH range investigated (pH 3-10). This suggests that effective separation using a cationic amine collector, which relies on electrostatic attraction as the basic mechanism of adsorption onto mineral surfaces, would not be possible without the use of a depressant. Zeta potential measurements and single mineral flotation tests in the presence of both Custamine 8113 and Norlig-H indicate that reagent adsorption is controlled by Custamine 8113 in the case of muscovite and by Norlig-H in the case of feldspar and quartz. This suggests that effective separation of muscovite from the gangue minerals present in the mica ore is possible. These results were used to determine optimal flotation conditions of the ore. Concentrates from the ore flotation were examined by QEMSCAN analysis, which showed that muscovite was effectively separated from gangue minerals.

Keywords: Muscovite; Flotation; Surface chemistry; Zeta potential; QEMSCAN; Electroacoustics; Amine; Calcium lignin sulfonate; Mica

1

1 Introduction 1.1 Muscovite Micas are a group of phyllosillicate minerals distinguished by their close to perfect basal cleavage. They are widespread in igneous, sedimentary and metamorphic rocks and their crystal structure incorporates a large number of elements, leading to a large and diverse mineral group. Of the many different minerals in the mica family, muscovite (KAl2(AlSi3O10)(OH)2) is the most common, and its physical properties render it useful in a wide range of applications, such as: insulators; condensers; plastics; cosmetics; and paints [1-3]. Muscovite is primarily separated from common gangue minerals such as quartz and feldspar, taking advantage of differences in physical properties [3]. These methods, however, prove to be ineffective when dealing with small particle sizes and therefore froth flotation is used [1, 3]. Froth flotation may also be used as a scavenger step on gravity tailings to recover some of the remaining muscovite.

1.2 Surface Chemistry In froth flotation, the electrical double layer at the mineral-water interface governs the adsorption of flotation reagents [4, 5]. An important mineral property in characterizing the electrical double layer is the isoelectric point (IEP), which is the pH value where the zeta potential is zero [6]. The IEP can aid in predicting the sign of the charge on a mineral surface in a given pH range [6]. Understanding the zeta potential of a mineral can help understand mechanisms of collector adsorption and select optimal flotation conditions to effectively separate a valuable mineral from gangue minerals [3, 5-7]. In systems where electrostatic attraction and hydrophobic bonding are major driving forces in collector adsorption, the adsorption process is characterized as physical and the extent of reagent adsorption is controlled by the sign and magnitude of the surface charge [4, 6]. Chemically adsorbing (chemisorbing) collectors form strong covalent or coordinate bonds with surface species and are able to adsorb onto the surface of similarly charged minerals, however a high surface charge can inhibit the adsorption of chemisorbing collectors via electrostatic repulsion [4]. For an introduction to the concept of surface charge and zeta potentials and their application in flotation, interested readers should consult Riley (2009) [8] and Fuerstenau and Pradip (2005) [4]. Zeta potential measurements may be conducted using different methods, with the most widely used technique being electrophoresis [9-12]. Although electrophoretic techniques are well-established, they are restricted in their applicability by the requirements for very fine particles in suspension and very

dilute (<1 %w/w) dispersions [9, 11-15]. The electroacoustic method offers distinct advantages over electrophoresis. As it is a non-optical measurement technique, electroacoustic methods are free of the limitations associated with optical electrophoretic measurements, and have been shown to be effective in analysing sizes ranging from a few nanometers to several micrometers [13, 14, 16, 17], concentrated and complex mineral suspensions [13, 14] in excess of 60 %w/w [16-18], opaque or photosensitive materials [9, 14] and flowing streams [14]. Electroacoustic zeta potential measurements are based on the application of high-frequency alternating electric fields to a suspension of particles, causing charged particles to oscillate and produce a sound wave of the same frequency. The zeta potential of the sample can be calculated from the phase and magnitude of the resulting sound wave. More details of the electroacoustic technique can be found in O'Brien (1990) [19], O'Brien (1995) [20] and O'Brien, et al. (1990) [21]; and a review of the applications can be found in Hunter (1998) [13] and Greenwood (2003) [17].

1.3 Physico-chemical properties of muscovite Muscovite has been shown to have permanently negatively charged basal planes [22-27] and edges with an IEP in the pH range 5 to 7.5 [23, 25, 27]. The edges account for approximately 5-10 % of the overall surface charge [28], resulting in an overall mineral IEP commonly occurring at pH <4 [22-24, 29, 30]. Due to the negative surface potential at pH >4, muscovite is expected to show much greater affinity for cationic collectors than anionic collectors, if electrostatic attraction is the mechanism of adsorption. Many papers on the application of collectors in the flotation of muscovite have been published, focusing primarily on cationic amine collectors [5, 22, 24, 31-42]. Electrostatic attraction is the basic mechanism involved in the adsorption of amines onto mineral surfaces, and thus it has been suggested that at higher pH values greater adsorption is observed, however increasing the pH to a point where the amine solubility limit is exceeded may be detrimental to flotation [36, 39, 43]. Conversely, some researchers have found that amine precipitation onto a mineral surface at an elevated pH can occur and improve flotation, but in either case at pH >12 flotation does not occur [43]. In the published literature focusing on muscovite flotation, the maximum flotation response with dodecylamine, a widely used collector, has been shown to occur around pH 8 [36, 44]. Adsorption of anionic and mixed cationic/anionic collectors on muscovite and the corresponding flotation response has also been examined [5, 24, 33, 45, 46]. Using sodium oleate as an anionic collector no adsorption was observed, however in mixed amine/oleate systems the adsorption of anionic and cationic collectors is enhanced due to co-

adsorption, although the underlying mechanism of this co-adsorption has not yet been identified [24, 33]. The zeta potential of common gangue minerals in mica ores, such as feldspar and quartz, have been shown to be negative at pH >3. [22, 47-53]. The lack of an IEP in this range means that silicate minerals will be electrostatically attracted to a cationic collector and it may not be possible to selectively concentrate muscovite from these minerals. Examples of quartz and feldspar flotation with cationic amine collectors indicate that these minerals must be effectively depressed for selective flotation to occur [54-60]. Calcium lignin sulfonate (Goulac), a silicate depressant, is commonly used in muscovite flotation [61, 62]. The objective of this work was to effectively separate muscovite from gangue minerals present in a mica ore through the use of froth flotation with an amine collector and calcium lignin sulfonate depressant. The surface chemistry of the ore minerals in the presence of various flotation reagents was investigated followed by flotation tests on both pure mineral samples as well as the ore itself.

2 Experimental 2.1 Materials Pure muscovite and feldspar used in this work were provided by an industrial minerals operation; quartz was purchased from Daubois (Canada). The muscovite ore used for flotation experiments was obtained from a muscovite mine. In order to obtain a size fraction small enough for zeta potential measurements the muscovite (d50 = 20 μm) was passed through a 2” Mozely hydrocylone with a cut size of approximately 10 μm and operating pressure of 35 psi. The overflow of the hydrocylone was then wet screened at 25 μm to produce a feed (d50 = 5.0 μm) for electroacoustic zeta potential measurements. The underflow of the hydrocylone was used directly for single mineral flotation experiments. Particle sizes, unless otherwise noted, were determined using a LA-920 particle size analyser (Horiba). Feldspar (d50 = 8.7 μm) was used as provided for electroacoustic measurements. The sample was determined by X-ray diffraction (Bruker D8 Diffractometer) to be a mixture of both Na-feldspar and Kfeldspar.

Quartz was prepared for single mineral flotation experiments by wet grinding in a rod mill to produce a 80 % -106 μm feed. For electroacoustic measurements quartz was milled in a Pulverisette 6 planetary monomill (Fritsch, Germany) to produce a particle size with a d50 of 4.1 μm. The muscovite ore was ground wet in a laboratory rod mill at 50 % solids for 22.5 min to produce an 80 % -106 μm flotation feed. The rod mill had an inner diameter of 180 mm and length of 230 mm. A total of 22 rods were used with a length of 190 mm as grinding media, weighing 10724 g and varying in diameter from 164 mm to 339 mm (14 of 164 mm, 6 of 227 mm and 2 of 339 mm). The mill speed was 61 rpm. Reagents obtained from the mica mine are shown in Table 1. Hydrochloric acid (at a concentration of 1 mol/L), used for pH modification in both zeta potential and flotation experiments, was purchased from Fisher Scientific.

2.2 Electroacoustic Zeta Potential Measurement Electroacoustic zeta potential measurements were carried out using a FieldESA (PartikelAnalytik, Germany) equipped with a large volume (220 mL) cell and an automatic titration unit. Samples investigated included muscovite, feldspar and quartz. Preparation of samples for measurement involved suspending the appropriate mass of mineral in 220 mL of 10-3 M KCl (background electrolyte) and then sonicating this suspension for 30 s using a UP400S ultrasonic processor (Hielscher, Germany). Sonication is required to ensure adequate dispersion of the fine mineral particles. The suspension was then allowed to equilibrate until both pH and zeta potential have stabilized at which point a pH titration was run to pH 10 and then back to pH 3 in steps of 0.25 pH units. A delay of 300 s prior to measurement at each new pH point was necessary to allow the suspension to equilibrate. For zeta potential measurements in the presence of a single flotation reagent, the reagent was added in steps of 500 μL to a total addition of 10 mL (20 steps). For zeta potential measurements with two different reagent additions the reagents were added in 500 μL steps to a total addition of 5 mL (10 steps). In the case of dual reagent additions each reagent was added at double the concentration of a single addition such that the total volume added remained consistent at 10 mL. The particle sizes of the three minerals are shown in Table 2, along with the measured surface areas (as determined by the N2 BET technique) and the solids concentration at which electroacoustic measurement was carried out. BET surface area measurements were carried out using a TriStar 3000 analyser (Micromeritics). Each mineral was initially measured at a series of solids concentrations to

determine the solids concentration which minimized the amount of material consumed while still providing a consistent and repeatable electroacoustic signal. The measured electroacoustic zeta potential of muscovite at three different solids concentrations may be seen in Appendix A. The reagent dosages for each mineral are shown in Table 3 on the basis of mineral mass, suspension volume and mineral surface area. The dosage of both Custamine 8113 and Norlig-H was determined for muscovite by measuring the zeta potential of the mineral suspension as a function of reagent addition. The determined values for Custamine 8113 and Norlig-H, 4.10 x 10-3 greagent/gsolid and 4.10 x 10-2 greagent/gsolid respectively, were selected as no further deviation in zeta potential was observed with increased reagent addition. All additions for feldspar and quartz were kept consistent on a mass basis with the dosages used for muscovite. Due to the nature of the data collected from electroacoustic zeta potential measurements (i.e. a large number of data points at many different pH levels) data have been presented as a fitted trendline, with calculated confidence intervals about the trendline. In all cases a third order polynomial was fitted to the data as this was found to be appropriate for muscovite, quartz and feldspar to represent the variation of zeta potential with pH. The confidence intervals indicate the degree of confidence that the fitted trendline accurately represents the mean of the electroacoustic measurements. While this approach adequately describes the experimental observations it was decided to utilize a “prediction” interval about the trendline to be absolutely certain that observed deviations in zeta potential were in fact significant. The “prediction” interval indicates the degree of confidence (95% confidence) with which one may use the trendline to predict a future zeta potential measurement. Figures 3 to 8 are reproduced in Appendix A (Figures A-4 to A-9) with the confidence intervals about the trendlines replaced with “prediction” intervals. Further explanation of the approach used to calculate the prediction intervals is also included in the Appendix.

2.3 Single Mineral Flotation Pure mineral flotation experiments were conducted on both muscovite and quartz (the feldspar as provided was too fine for flotation to be effective). All flotation tests were carried out in a 1.5 L Denver flotation cell at an airflow rate of 4.5 L/min with concentrate collected after 30 s, 60 s and 300 s. The frother Cp-102a was added as needed to create a stable froth. The concentrate was filtered, dried and weighed to calculate the flotation recovery. The mass of mineral used in each test was 50 g for muscovite and 100 g for quartz. The initial volume of water added for each test was 1.3 L.

The conditions examined for muscovite included pH (pH 4, 7.5 and 10) as well as collector dosage (100, 250 and 500 g/t). In each of these conditions the depressant dosage was kept constant at 1500 g/t. After selecting pH 7.5 as the optimal pH of flotation a series of additional tests were conducted to assess the effect of varying Norlig-H dosage (0, 500 and 1500 g/t). Pure mineral flotation experiments on quartz were carried out at pH 7.5 to investigate the effects of varying dosages of Custamine 8113 (100, 250 g/t) and Norlig-H (0, 500, 1000 and 1500 g/t).

2.4 Mica Ore Flotation Flotation experiments on the mica ore were carried out in a 3 L Denver flotation cell at an air flow rate of 4.5 L/min using Cp-102a as frother. For each test a fresh batch of ore (500 g) was rod milled wet and then transferred directly to the flotation cell where the pulp level was adjusted to a set height by adding tap water. Caustic was added to adjust the pH to 8 followed by the addition of Norlig-H. The ore was then allowed to condition for 1 min before Custamine 8113 was added, followed by another 1 min conditioning step. The air was then turned on with frother added as required to produce a stable froth. Froth was collected after 30 s, 60 s and 300 s to produce three different concentrates. Throughout each flotation test the cell level and pH were continuously adjusted to ensure consistent conditions. The reagent combinations tested in the ore flotation experiments may be seen in Table 4.

2.5 QEMSCAN In order to quantify the flotation behaviour of the mica and associated minerals of the ore body the mineralogy of concentrate and tailings fractions should be measured. Quantitative evaluation of materials by scanning electron microscopy (QEMSCAN) was used in this work to determine the mineral abundance of muscovite and gangue minerals in the feed and in the flotation concentrates. QEMSCAN is a mineralogical characterisation technique employing an EVO 430 automated scanning electron microscope equipped with four light-element energy-dispersive X-ray spectrometers and iDiscover software for data and image processing. QEMSCAN measures (and the iDiscover software processes) data from every pixel across a sample with a pixel size defined based on the scope of the analysis. The software assigns each pixel a mineral name based on 1,000 counts of energy dispersive X-ray spectral data and backscatter electron intensities. If the minerals or constituent phases comprising the sample are chemically distinct, QEMSCAN is capable of reliably discriminating and quantifying minerals [63] by comparing the data from each pixel to a set of mineral definitions which are validated and refined to fit the particular sample set under investigation.

Samples prepared for QEMSCAN analysis included the feed as well as representative samples of the combined concentrate from each flotation condition used during the ore flotation experiments. Each sample was screened and re-combined into three size fractions (+75µm, -75/+38 μm, -38 μm) for a more detailed mineralogical determination to assess potential metallurgical beneficiation performance. These samples were prepared as polished sections and analyzed by QEMSCAN at the Advanced Mineralogy Facility at SGS Canada (Lakefield). A riffled, representative aliquot was obtained from each fraction for whole rock compositional analysis by X-ray fluorescence followed by preparation of the polished sections required for QEMSCAN analysis. Graphite was added to disperse particles, create random orientation and minimise density-induced particle segregation, settling and preferred orientation. The samples were mounted into an epoxy approximately 5 mm deep, and then backfilled with epoxy. This shallow depth was selected to minimise gravity separation during epoxy hardening. After hardening, the sections were ground, polished, carboncoated, and analyzed utilizing QEMSCAN. Two polished sections were prepared from the +75 and one from each of the -75/+38 μm and -38 μm fractions from each sample. The samples were analyzed using the Particle Mineral Analysis (PMA) QEMSCAN method. The PMA is a two-dimensional mapping analysis aimed at resolving liberation characteristics of a set of particles. A pre-defined number of particles were mapped at a pixel size of 3 μm. The QEMSCAN instrument was operated at a 25 kV accelerating voltage and a 5 nA beam current. A reference mineral list was developed using XRD (primarily to define the major minerals) as well as a scanning electron microscope (SEM) equipped with an energy dispersive spectrometer (SEM-EDS). To verify the broad mineral composition and improve the mineral database SEM-EDS analysis was employed as needed. For data validation purposes, the calculated assays using QEMSCAN were compared to chemical assays using Xray fluorescence (XRF). The chemical analyses completed were conducted at SGS Canada (Lakefield).

2.6 X-Ray Diffraction (XRD) Analysis XRD analyses were carried out on the coarse fraction from the feed and concentrates for quality control purposes using a BRUKER AXS D8 Advance Diffractometer at SGS Canada (Lakefield). Analytical conditions were Co radiation, 40 kV, 35 mA, with a regular scanning step at 0.02°, step time at 0.2s, and 2θ range at 3-70°. Interpretations were based on the PDF2/PDF4 powder diffraction databases issued by the International Center for Diffraction Data (ICDD) using the DiffracPIus Eva software. Detection limit is 0.5-2% and is strongly dependent on crystallinity. The reference patterns are compiled by the Joint

Committee on Powder Diffraction Standards - International Center for Diffraction Data (JCPDS-ICDD) and released on software as a database of Powder Diffraction Files (PDF).

3 Results and Discussion 3.1 Electroacoustic Zeta Potential Measurement The zeta potential data of muscovite, feldspar and quartz as a function of pH are shown in Figure 1. The data for muscovite indicates an IEP of 3.5, which is in agreement with literature values [22-24, 29, 30]. The data for quartz and feldspar display the expected negative zeta potential across the pH range investigated (3-10) [22, 47-53]. During the course of measuring the zeta potential of feldspar it was noticed that the time required for the mineral suspension to reach equilibrium (prior to any reagent addition) was significantly longer than for the other two minerals (16 hours compared with 1-2 hours for mica and quartz). In order to rule out the presence of any adsorbed organic species on the feldspar surface the mineral was heated to 200 °F and left overnight to remove any volatile organic species from the surface. After heating, the zeta potential of the feldspar showed little variation compared to that of the feldspar shown in Figure 1. The zeta potentials of feldspar before and after heating are shown in the Appendix. Examining the trends of the muscovite, quartz and feldspar it can be seen that selective separation of muscovite from the gangue minerals would be difficult using an amine collector, which relies on electrostatic attraction as the basic mechanism of adsorption [39]. All three minerals are negatively charged for pH >3.5, leaving a small window of pH values where muscovite has a surface charge opposite to that of feldspar and quartz. In industry, flotation at acidic pH (especially at pH <3) can be problematic due to workplace safety concerns, high reagent requirements, and process water treatment considerations. Figures 2, 3, and 4 show the results of muscovite, feldspar and quartz in the presence of Custamine 8113 and Norlig-H. Deviation in the zeta potential of a particle due to the addition of ionic surfactants is an indication of adsorption onto the surface. It can be seen that Custamine 8113 has a significant influence on the zeta potential of all three minerals. The upward shifted curves indicate that the cationic collector has adsorbed on to the negatively charged mineral surfaces. This suggests that without the use of a depressant, muscovite cannot be selectively separated from feldspar and quartz. Examining the zeta potentials in the presence of Norlig-H indicates that adsorption is occurring on the surface of feldspar, however the changes in zeta potential for muscovite and quartz are inconclusive. While Figure 2 and 4

appear to indicate a statistically significant deviation between the pure mineral and the mineral in the presence of Norlig-H, when the prediction intervals shown in Figures A-4 to A-6 are examined, it may be seen that zeta potential data alone is insufficient to determine adsorption of this reagent onto either muscovite or quartz surfaces. Since feldspar and quartz both show affinity for Custamine 8113 it is important to examine the zeta potential when both collector and depressant are present. It can be seen from Figure 5 that the zeta potential trend for muscovite in the presence of both collector and depressant resembles that of muscovite in the presence of Custamine 8113 alone. This indicates that muscovite has a higher affinity for the collector compared to the depressant, whereas Figures 6 and 7 show that feldspar and quartz have a higher affinity for Norlig-H, indicating that selective separation of muscovite from feldspar and quartz is possible. It is also interesting to note that the order of addition of the two reagents shows little change in the zeta potential trends, indicating the strong selectivity of Norlig-H for both feldspar and quartz.

3.2 Single Mineral Flotation Results from the flotation of muscovite can be seen in Figures 8 and 9. They indicate that effective flotation is possible using Custamine 8113 as collector, even in the presence of high Norlig-H dosage. Figure 8 shows the effect of pH on the recovery of muscovite. At pH 7.5 and pH 4 a similar flotation response was observed for all conditions, however at pH 10 a large decrease in recovery is seen. This likely indicates that the solubility limit of the amine collector has been reached and that the Custamine 8113 collector has a solubility limit comparable to other amine collectors, such as dodecylamine, which has a solubility limit near pH 10 [43]. Figure 9 shows a slight reduction in recovery with increasing Norlig-H concentrations; however this reduction is not statistically significant. An increase in recovery can be seen with increasing collector concentration; however this increase becomes much less significant after concentrations of 250 g/t Custamine 8113 are reached.

Single mineral flotation results for quartz were in good agreement with the zeta potential measurements, indicating quartz has a higher affinity for Norlig-H than Custamine 8113. Figure 10 illustrates that Norlig-H effectively depresses quartz in the presence of Custamine 8113, demonstrating that effective separation of muscovite from unwanted gangue minerals may be possible. The results do however indicate that with increased Custamine 8113 additions more Norlig-H is required in order for

the mineral to be depressed. At 100 g/t concentrations of Custamine 8113, 1000 g/t of Norlig-H is sufficient for depressing the majority of the quartz, however when 250 g/t of collector is added 1500 g/t of depressant is needed as quartz recoveries above 50 % are seen at 1000 g/t concentrations. This suggests that for a concentration of Custamine 8113 greater than 250 g/t, Norlig-H additions would need to be greater than 1500 g/t to effectively depress silicate gangue minerals in this system. These initial flotation results indicate that optimal flotation conditions may occur with reagent concentrations of 250 g/t Custamine 8113 and 1500 g/t Norlig-H. These dosages were used as flotation conditions for later experiments in order to avoid very high depressant concentrations. The flotation recovery of both muscovite and quartz as a function of time at these conditions is shown in Figure 11. It can be seen that the majority of muscovite and quartz collection occurs within the first 60 s of flotation (61 % muscovite, 12 % quartz). After 300 s muscovite and quartz recoveries were 78 % and 14 % respectively. The similar flotation kinetics of these minerals indicate that a selective separation based on different flotation times would likely be ineffective.

3.3 Mica Ore Flotation The mass pull results for flotation tests with varying concentrations of Custamine 8113 and Norlig-H can be seen in Figure 12. Examining the effect of the depressant, it was observed that under the presence of Norlig-H the total mass recovery decreases, which may imply that a greater amount of gangue is reporting to the concentrate when no depressant is used. Decreasing collector concentrations in the presence of Norlig-H also showed a decrease in mass recovery. These results suggest that 1500 g/t of Norlig-H is not sufficient in depressing gangue minerals with collector concentrations above 250 g/t, similar to the results obtained from single mineral flotation tests.

3.4 QEMSCAN The mineralogy of the feed sample is provided in

Table 5. The sample consists of quartz (62.2%), K-feldspar (19.6%), muscovite (9.1%), plagioclase (5.6%), and trace/minor amounts of other minerals. The bulk composition of the feed sample was validated by XRD analysis. The liberation and association characteristics of muscovite were examined. For the purposes of this analysis, particle liberation is defined based on 2D particle area percent. Muscovite is well liberated at this K80 (~75 μm) at 90% calculated for the feed sample. The remainder of the mass of the muscovite occurs as ternary middling particles with quartz and feldspars. Liberation is greater than 75% in all size fractions. Particle maps of free muscovite and middling particles of muscovite with quartz/feldspars are shown in Figure 13 and 14, respectively. The results from QEMSCAN analysis of the untreated ore and the combined concentrate from each flotation condition used during the mica ore flotation experiments can be seen in Figure 15, with grades and recoveries calculated from the QEMSCAN results shown in Figure 16. The numerical QEMSCAN results can be seen in the Appendix (Table A1). Comparing the results from the feed sample and the flotation concentrate produced without Norlig-H indicates that selective flotation of muscovite from unwanted gangue minerals using Custamine 8113 as collector is not possible without the use of a depressant. The concentrate with the highest grade of muscovite is obtained when Custamine 8113 and Norlig-H concentrations are 250 g/t and 1500 g/t respectively, further demonstrating that increasing collector concentrations to 500 g/t without increasing the depressant dosage leads to a decrease in grade. Examining mineral concentrations at the three size fractions (+75 μm, 38-75 μm and -38 μm), it can be seen that, in general, muscovite grades are higher at larger particle sizes. For example, when 250 g/t collector and 1500 g/t depressant were used the +75 μm, 38-75 μm and -38 μm size fractions had muscovite grades of 87 %, 49 % and 26 % respectively. This indicates that the majority of the gangue recovery is occurring at finer particle sizes. It is also interesting to note that the recovery of coarse (+75 μm) quartz and feldspar is elevated at high collector dosage in the absence of depressant. This suggests that without the addition of a depressant a significant amount of coarse gangue is recovered, most likely via the flotation of non-liberated particles containing both mica and gangue mineral phases. Combining these data with the high volumes of water being recovered in the concentrates (Figure 17), it is proposed that the fine gangue particles are entering the concentrate via entrainment rather than by true flotation. Adding a de-sliming step to remove fines or adding a cleaner flotation stage may further

concentrate the muscovite. The addition of traditional entrainment treatments such as wash water may also help to reduce the recovery of fine gangue.

4 Conclusions This work investigated the separation of muscovite from unwanted gangue minerals in a mica ore. The investigation included zeta-potential measurements and single mineral flotation tests of pure muscovite and gangue minerals, feldspar and quartz, to understand reagent interactions at mineral particle surfaces and evaluate flotation responses in the presence of an amine collector, Custamine 8113, and depressant, Norlig-H. These results were then compared to the results of bench scale flotation tests performed on the mica ore. The conclusions are as follows: 1. Both Custamine 8113 and Norlig-H are able to adsorb onto the surface of feldspar and quartz, however the results indicate a higher affinity for Norlig-H. Reagent adsorption onto the surface of muscovite is observed with Custamine 8113 but not Norlig-H. 2. Selective separation of muscovite from quartz and feldspar by froth flotation is possible using Custamine 8113 and Norlig-H, however reagent concentrations (particularly depressant dosage) play a strong role in the effectiveness of the separation. Increasing collector concentrations without increasing depressant dosages increases gangue recoveries. Optimal reagent additions were found to be 250 g/t Custamine 8113 and 1500 g/t Norlig-H. 3. Although gangue minerals can be effectively depressed, the recovery of fine (-38 µm) gangue particles reduces the overall product grade. The fine gangue recovery, along with large water recoveries suggests entrainment and not true flotation, as the mechanism responsible for this reduction in grade. The addition of wash water, a cleaner flotation step or a de-sliming step prior to flotation to remove fines could further improve the muscovite grade of the final product.

Acknowledgements The authors would like to acknowledge funding from the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant: Mineral Processing Fundamentals. A. Jordens acknowledges funding from the McGill Engineering Doctoral Award as well as an NSERC Alexander Graham Bell Canada Graduate Scholarship. The authors also gratefully acknowledge helpful discussions with Prof. J.A. Finch and Prof. G.A. Whitmore regarding the display of zeta potential data.

References [1] S.F. Santos, S.C.A. França, T. Ogasawara, Method for grinding and delaminating muscovite, Mining Science and Technology (China), 21 (2011) 7-10. [2] J.L. Pérez-Rodríguez, A. Wiewióra, J. Drapala, L.A. Pérez-Maqueda, The effect of sonication on dioctahedral and trioctahedral micas, Ultrasonics Sonochemistry, 13 (2006) 61-67. [3] L. Wang, W. Sun, Y.-H. Hu, L.-H. Xu, Adsorption mechanism of mixed anionic/cationic collectors in Muscovite – Quartz flotation system, Minerals Engineering, 64 (2014) 44-50. [4] D.W. Fuerstenau, Pradip, Zeta potentials in the flotation of oxide and silicate minerals, Advances in Colloid and Interface Science, 114–115 (2005) 9-26. [5] B. Rai, P. Sathish, J. Tanwar, Pradip, K.S. Moon, D.W. Fuerstenau, A molecular dynamics study of the interaction of oleate and dodecylammonium chloride surfactants with complex aluminosilicate minerals, Journal of Colloid and Interface Science, 362 (2011) 510-516. [6] M.I. Pope, D.I. Sutton, The correlation between froth flotation response and collector adsorption from aqueous solution. Part I. Titanium dioxide and ferric oxide conditioned in oleate solutions, Powder Technology, 7 (1973) 271-279. [7] A.I. Zouboulis, A. Avranas, Treatment of oil-in-water emulsions by coagulation and dissolved-air flotation, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 172 (2000) 153-161. [8] J. Riley, Charge in colloidal systems, in: Colloid Science, Blackwell Publishing Ltd., 2009, pp. 14-35. [9] A.J. Babchin, R.S. Chow, R.P. Sawatzky, Electrokinetic measurements by electroacoustical methods, Advances in Colloid and Interface Science, 30 (1989) 111-151. [10] P.R. Johnson, A comparison of streaming and microelectrophoresis methods for obtaining the ζ potential of granular porous media surfaces, Journal of Colloid and Interface Science, 209 (1999) 264267. [11] N.P. Miller, J.C. Berg, A comparison of electroacoustic and microelectrophoretic zeta potential data for titania in the absence and presence of a poly (vinyl alcohol) adlayer, Colloids and Surfaces, 59 (1991) 119-128.

[12] M. Kaszuba, J. Corbett, F.M. Watson, A. Jones, High-concentration zeta potential measurements using light-scattering techniques, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 368 (2010) 4439-4451. [13] R.J. Hunter, Recent developments in the electroacoustic characterisation of colloidal suspensions and emulsions, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 141 (1998) 37-66. [14] B.J. Marlow, D. Fairhurst, H.P. Pendse, Colloid vibration potential and the electrokinetic characterization of concentrated colloids, Langmuir, 4 (1988) 611-626. [15] P.J. Bruinsma, P.A. Smith, B.C. Bunker, Dynamic mobility spectra of multicomponent colloidal suspensions, The Journal of Physical Chemistry B, 101 (1997) 8410-8417. [16] B. Klein, N.E. Altun, M. Colebrook, M. Pawlik, Electroacoustic measurements of mixed quartz and iron oxide mineral systems, International Journal of Mineral Processing, 110–111 (2012) 12-17. [17] R. Greenwood, Review of the measurement of zeta potentials in concentrated aqueous suspensions using electroacoustics, Advances in Colloid and Interface Science, 106 (2003) 55-81. [18] R. Greenwood, B. Lapčíková, M. Surýnek, K. Waters, L. Lapčík, Jr., The zeta potential of kaolin suspensions measured by electrophoresis and electroacoustics, Chemical Papers, 61 (2007) 83-92. [19] R.W. O'Brien, The electroacoustic equations for a colloidal suspension, Journal of Fluid Mechanics, 212 (1990) 81-93. [20] R.W. O'Brien, The dynamic mobility of a porous particle, Journal of Colloid and Interface Science, 171 (1995) 495-504. [21] R.W. O'Brien, B.R. Midmore, A. Lamb, R.J. Hunter, Electroacoustic studies of moderately concentrated colloidal suspensions, Faraday Discussions of the Chemical Society, 90 (1990) 301-312. [22] S. Nishimura, H. Tateyama, K. Tsunematsu, K. Jinnai, Zeta potential measurement of muscovite mica basal plane-aqueous solution interface by means of plane interface technique, Journal of Colloid and Interface Science, 152 (1992) 359-367. [23] A. Nosrati, J. Addai-Mensah, W. Skinner, Muscovite clay mineral particle interactions in aqueous media, Powder Technology, 219 (2012) 228-238.

[24] L. Xu, H. Wu, F. Dong, L. Wang, Z. Wang, J. Xiao, Flotation and adsorption of mixed cationic/anionic collectors on muscovite mica, Minerals Engineering, 41 (2013) 41-45. [25] H. Zhao, S. Bhattacharjee, R. Chow, D. Wallace, J.H. Masliyah, Z. Xu, Probing surface charge potentials of clay basal planes and edges by direct force measurements, Langmuir, 24 (2008) 1289912910. [26] L. Yan, J.H. Masliyah, Z. Xu, Interaction of divalent cations with basal planes and edge surfaces of phyllosilicate minerals: Muscovite and talc, Journal of Colloid and Interface Science, 404 (2013) 183-191. [27] P.-I. Au, S.-Y. Siow, L. Avadiar, E.-M. Lee, Y.-K. Leong, Muscovite mica and koalin slurries: Yield stress–volume fraction and deflocculation point zeta potential comparison, Powder Technology, 262 (2014) 124-130. [28] M.V. Maslova, L.G. Gerasimova, W. Forsling, Surface properties of cleaved mica, Colloid Journal, 66 (2004) 322-328. [29] P.J. Scales, F. Grieser, T.W. Healy, Electrokinetics of the muscovite mica-aqueous solution interface, Langmuir, 6 (1990) 582-589. [30] P.J. Scales, T.W. Healy, D.F. Evans, The zeta potential of muscovite mica: Counterion complexation by a macrocyclic ligand, Journal of Colloid and Interface Science, 124 (1988) 391-395. [31] A. Blom, G.G. Warr, Structure and composition of cationic−nonionic surfactant mixed adsorbed layers on mica, Langmuir, 22 (2006) 6787-6795. [32] L.O. Filippov, A. Duverger, I.V. Filippova, H. Kasaini, J. Thiry, Selective flotation of silicates and Cabearing minerals: The role of non-ionic reagent on cationic flotation, Minerals Engineering, 36–38 (2012) 314-323. [33] K. Hanumantha Rao, K.S.E. Forssberg, Mixed collector systems in flotation, International Journal of Mineral Processing, 51 (1997) 67-79. [34] P.C. Herder, P.M. Claesson, C.E. Herder, Adsorption of cationic surfactants on muscovite mica as quantified by means of ESCA, Journal of Colloid and Interface Science, 119 (1987) 155-167. [35] E.C. Orhan, İ. Bayraktar, Amine–oleate interactions in feldspar flotation, Minerals Engineering, 19 (2006) 48-55.

[36] R.J. Pugh, M.W. Rutland, E. Manev, P.M. Claesson, Dodecylamine collector — pH effect on mica flotation and correlation with thin aqueous foam film and surface force measurements, International Journal of Mineral Processing, 46 (1996) 245-262. [37] M. Rutland, A. Waltermo, P. Claesson, pH-dependent interactions of mica surfaces in aqueous dodecylammonium/dodecylamine solutions, Langmuir, 8 (1992) 176-183. [38] Ž. Sekulić, N. Canić, Z. Bartulović, A. Daković, Application of different collectors in the flotation concentration of feldspar, mica and quartz sand, Minerals Engineering, 17 (2004) 77-80. [39] Y. Xu, Y.-L. Liu, D.-D. He, G.-S. Liu, Adsorption of cationic collectors and water on muscovite (0 0 1) surface: A molecular dynamics simulation study, Minerals Engineering, 53 (2013) 101-107. [40] P.M. McGuiggan, R.M. Pashley, A study of surfactant solution wetting on mica, Colloids and Surfaces, 27 (1987) 277-287. [41] R.M. Pashley, P.M. McGuiggan, B.W. Ninham, D.F. Evans, Attractive forces between uncharged hydrophobic surfaces: Direct measurements in aqueous solution, Science, 229 (1985) 1088-1089. [42] R.-H. Yoon, S.A. Ravishankar, Long-range hydrophobic forces between mica surfaces in alkaline dodecylammonium chloride solutions, Journal of Colloid and Interface Science, 179 (1996) 403-411. [43] R.W. Smith, J. L Scott, Mechanisms of dodecylamine flotation of quartz, Mineral Processing and Extractive Metallurgy Review, 7 (1990) 81-94. [44] X.P. Zheng, H.K. Lin, Effect of mineralogical properties of synthetic mica on its floatability, Minerals and Metallurgical Processing, 11 (1994) 20-25. [45] K. Hanumantha Rao, J.M. Cases, O. Barres, K.S.E. Forssberg, Flotation, electrokinetic and FT-IR studies of mixed anionic/cationic collectors in muscovite-biotite system, in: S.P. Mehrotra, R. Shekhar (Eds.), Mineral Processing: Recent Advances and Future Trends, Allied Publ. Ltd, New Delhi, 1995, pp. 29–44. [46] K.S. Moon, D.W. Fuerstenau, Surface crystal chemistry in selective flotation of spodumene (LiAl[SiO3]2) from other aluminosilicates, International Journal of Mineral Processing, 72 (2003) 11-24. [47] A. Vidyadhar, K. Hanumantha Rao, Adsorption mechanism of mixed cationic/anionic collectors in feldspar-quartz flotation system, Journal of Colloid and Interface Science, 306 (2007) 195-204.

[48] J. Martinovic, D. Bradshaw, P. Harris, Investigation of surface properties of gangue minerals in platinum bearing ores, in: International Platinum Conference: Platinum Adding Value, The South African Institute of Mining and Metallurgy, 2004, pp. 151-158. [49] G.A. Parks, The isoelectric points of solid oxides, solid hydroxides, and aqueous hydroxo complex systems, Chemical Reviews, 65 (1965) 177-198. [50] A.C.P. Duarte, S.R. Grano, Mechanism for the recovery of silicate gangue minerals in the flotation of ultrafine sphalerite, Minerals Engineering, 20 (2007) 766-775. [51] I. Larson, C.J. Drummond, D.Y.C. Chan, F. Grieser, Direct force measurements between silica and alumina§, Langmuir, 13 (1997) 2109-2112. [52] P. Somasundaran, Cationic depression of amine flotation of quartz, Transactions, Society of Mining Engineers, AIME, 256 (1974) 64-68. [53] H.C. Li, P.L. De Bruyn, Electrokinetic and adsorption studies on quartz, Surface Science, 5 (1966) 203-220. [54] A.C. Araujo, P.R.M. Viana, A.E.C. Peres, Reagents in iron ores flotation, Minerals Engineering, 18 (2005) 219-224. [55] B. Kar, H. Sahoo, S.S. Rath, B. Das, Investigations on different starches as depressants for iron ore flotation, Minerals Engineering, 49 (2013) 1-6. [56] A.M. Vieira, A.E.C. Peres, The effect of amine type, pH, and size range in the flotation of quartz, Minerals Engineering, 20 (2007) 1008-1013. [57] A. Vidyadhar, K.H. Rao, K.S.E. Forssberg, Adsorption of N-tallow 1,3-propanediamine–dioleate collector on albite and quartz minerals, and selective flotation of albite from greek stefania feldspar ore, Journal of Colloid and Interface Science, 248 (2002) 19-29. [58] C. Demir, A.A. Abramov, M.S. Çelik, Flotation separation of Na-feldspar from K-feldspar by monovalent salts, Minerals Engineering, 14 (2001) 733-740. [59] C. Demir, I. Bentli, I. Gülgönül, M.S. Çelik, Effects of bivalent salts on the flotation separation of Nafeldspar from K-feldspar, Minerals Engineering, 16 (2003) 551-554.

[60] M. Gaied, W. Gallala, Beneficiation of feldspar ore for application in the ceramic industry: Influence of composition on the physical characteristics, Arabian Journal of Chemistry, (2011). [61] R. Houot, J.-P. Cuif, Y. Mottot, J.-C. Samama, Recovery of rare earth minerals, with emphasis on flotation process, in: Materials Science Forum, 1991, pp. 301-324. [62] J.S. Browning, R.B. Adair, Selective flotation of mica from Georgia pegmatites, U.S. Dept. of the Interior, Bureau of Mines, Washington, D.C., 1966. [63] P. Gottlieb, G. Wilkie, D. Sutherland, E. Ho-Tun, S. Suthers, K. Perera, B. Jenkins, S. Spencer, A. Butcher, J. Rayner, Using quantitative electron microscopy for process mineralogy applications, JOM, 52 (2000) 24-25. [64] R.A. Johnson, D.W. Wichern, Applied multivariate statistical analysis, Prentice hall Englewood Cliffs, NJ, 1992.

Appendix A The construction of confidence and prediction intervals for electroacoustic zeta potential as a function of the pH level requires the following two assumptions to hold: 1. The mean zeta potential as a function of pH is a third-order polynomial. 2. The measurement errors in readings on the zeta potential function are mutually independent and identically distributed normal random variables. The data set for each interval estimate consists of n pairs of readings on zeta potential and pH. We denote the n-component column vector of zeta readings by Y and consider an n × k matrix X where k = 4. The ith row of matrix X has form (1, xi, xi2 , xi3), i = 1, . . ., n, where xi denotes the pH level for the ith reading of zeta potential. Next, we consider a specific pH reading x0 and the row vector x0 = (1, x0, x02, x03). Finally, we let 1 − α denote the specified confidence level of the confidence or prediction interval and t = t1−α/2,n−k, the 1−α/2 quantile of a t-distribution with n − k degrees of freedom. An apostrophe denotes a matrix or vector transpose. The computational formulae for a confidence interval and prediction interval are, respectively: (1) Here

is the value of the fitted zeta potential function at pH level x 0 and sm and sp are standard errors,

computed as follows: (2) where s2 denotes the estimated error variance: (3) Symbol I denotes the nth order identity matrix. The formulae can be found in standard references; see, for example, Johnson and Wichern (1992) [64]. The confidence and prediction intervals in (1) are readily computed using most statistical packages. The statistical software “Stata 13” was used for this work. The confidence interval in (1) is an interval estimate of ζ 0, the mean zeta potential at pH level x0. In contrast, the prediction interval in (1) attempts to bracket the individual zeta potential measurement on a particular occasion when the pH level is fixed at x0. It can be seen from the expressions in (2) that the prediction interval will be wider than the confidence interval (at the same level of confidence). A

comparison of both prediction and confidence intervals for the zeta potential of muscovite may be seen in Figure A-1, along with a plot of the raw data.

Figure Captions Figure 1 - Zeta potential trend of muscovite, feldspar and quartz in 10-3 mol/L KCl (Error intervals shown are 99% confidence intervals) Figure 2 - Zeta potential trend of muscovite with and without Custamine 8113 and Norlig H (Error intervals shown are 99 % confidence intervals) Figure 3 - Zeta potential trend of feldspar with and without Custamine 8113 and Norlig H (Error intervals shown are 99 % confidence intervals) Figure 4 - Zeta potential trend of quartz with and without Custamine 8113 and Norlig H (Error intervals shown are 99 % confidence intervals) Figure 5 - Zeta potential trend of muscovite with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends (Error intervals shown are 99 % confidence intervals) Figure 6 - Zeta potential trend of feldspar with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends (Error intervals shown are 99 % confidence intervals) Figure 7 - Zeta potential trend of quartz with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends (Error intervals shown are 99 % confidence intervals) Figure 8 - Single mineral flotation results for muscovite with constant Norlig-H concentrations and varying concentrations of Custamine 8113 and pH (Error bars represent 95 % confidence intervals) Figure 9 - Single mineral flotation results for muscovite with varying concentrations of Custamine 8113 and Norlig-H at the natural pH 7.5 (Error bars represent 95 % confidence intervals) Figure 10 - Single mineral flotation results for quartz with varying concentrations of Custamine 8113 and Norlig-H at a constant pH 7.5 (Error bars represent 95 % confidence intervals) Figure 11 - Pure mineral flotation results of muscovite and quartz as a function of time in the presence of 250 g/t Custamine 8113 and 1500 g/t Norlig-H at pH 7.5 (error bars represent 95 % confidence intervals)

Figure 12 - Mass recovery, from muscovite flotation tests, after collecting froth for 30, 60 and 300 seconds (error bars represent 95 % confidence intervals) Figure 13 - Particle maps of free muscovite in the +75 μm fraction Figure 14 - Particle maps of muscovite middlings associated with quartz/feldspars from the +75 μm fraction Figure 15 - QEMSCAN results for the combined concentrates of the muscovite flotation tests for (a) combined particle size and (b) +75 m, (c) 38-75 m and (d) -38 m fractions. Custamine concentrations used were 500 g/t (High) and 250 g/t (Low) with Norlig-H (Depressant) added at a concentration of 1500 g/t Figure 16 - (a) Total Mass recovery, grade (solid line) and recovery (dashed line) of (b) muscovite, (c) quartz and (d) feldspar for three different size fractions. Custamine concentrations used were 500 g/t (High) and 250 g/t (Low) with Norlig-H (Depressant) added at a concentration of 1500 g/t Figure 17 - Cumulative water recovery by volume (error bars represent 95 % confidence intervals) Figure A-1 - Zeta potential data for muscovite as a function of pH along with fitted third order polynomial trendline. Confidence interval [CI] (99 % confidence) and prediction interval [PI] (95% confidence) are shown about the trendline. Figure A-2 - Zeta potential trend for muscovite at three different solids concentrations (Error intervals shown are 99 % confidence intervals) Figure A-3 - Zeta potential trend for feldspar before and after heating (Error intervals shown are 99 % confidence intervals) Figure A-4 - Zeta potential trend of muscovite with and without Custamine 8113 and Norlig-H (Error intervals shown are 99 % prediction intervals) Figure A-5 - Zeta potential trend of feldspar with and without Custamine 8113 and Norlig-H (Error intervals shown are 99 % prediction intervals) Figure A-6 - Zeta potential trend of quartz with and without Custamine 8113 and Norlig-H (Error intervals shown are 99 % prediction intervals)

Figure A-7 - Zeta potential trend of muscovite with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends. (Error intervals shown are 99 % prediction intervals) Figure A-8 - Zeta potential trend of feldspar with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends. (Error intervals shown are 95% prediction intervals) Figure A-9 - Zeta potential trend of quartz with and without both Custamine 8113 and Norlig-H. The order of reagent addition is reflected in the names of the trends. (Error intervals shown are 95% prediction intervals)

Table 1 Reagents from the mica mine Reagent Name

Reagent Type

Manufacturer

Custamine 8113

Collector/Frother

ArrMaz

Norlig-H

Depressant

LignoTech

Cp-102 A

Frother

ArrMaz

Caustic

pH Modifier

OxyChem

Table 2 Particle size, surface area and electroacoustic measurement concentration for muscovite, feldspar and quartz Mineral

d50 (μm)

BET Surface Area (m2/g)

Solids Conc. (wt. %)

Muscovite

5.0

9.67

10.0

Feldspar

8.7

1.88

5.0

Quartz

4.1

3.59

2.5

Table 3 Electroacoustic reagent dosages for muscovite, feldspar and quartz Mineral

Reagent

Dosage (g/g)

Dosage (g/L)

Dosage (g/m2)

Muscovite

Custamine 8113

4.10 x 10-3

4.36 x 10-1

4.24 x 10-4

Norlig-H

4.10 x 10-2

4.36 x 100

4.24 x 10-3

Custamine 8113

4.10 x 10-3

2.07 x 10-1

2.19 x 10-3

Norlig-H

4.10 x 10-2

2.07 x 100

2.19 x 10-2

Custamine 8113

4.10 x 10-3

1.01 x 10-1

1.14 x 10-3

Norlig-H

4.10 x 10-2

1.01 x 100

1.14 x 10-2

Feldspar

Quartz

Table 4 Conditions of mica ore flotation tests Condition

Custamine 8113 (g/t)

Norlig-H (g/t)

1

500

1500

2

500

0

3

250

1500

Table 5 Modal mineralogy of the feed sample Fraction

Combined

+75um

-75/+38um

-38um

49.0

26.0

25.0

37

24

11

Mass Size Distribution (%) Calculated ESD Particle Size

Mineral Mass (%)

Mean Grain Size by Frequency (µm)

20 Sample

Sample

Fraction

Sample

Fraction

Sample

Fraction

Quartz

62.2

33.7

68.7

16.6

64.0

11.9

47.5

Plagioclase

5.58

2.09

4.26

1.79

6.90

1.70

6.78

K-Feldspar

19.6

6.70

13.7

5.52

21.2

7.40

29.6

Oxides

0.37

0.08

0.17

0.13

0.52

0.16

0.62

Muscovite

9.08

5.47

11.2

1.42

5.48

2.18

8.74

Biotite

0.58

0.35

0.72

0.11

0.42

0.12

0.47

Epidote

1.68

0.36

0.73

0.22

0.83

1.11

4.44

Tourmaline

0.80

0.27

0.56

0.16

0.62

0.37

1.46

Other Silicates

0.05

0.01

0.02

0.00

0.01

0.03

0.14

Other

0.08

0.01

0.02

0.01

0.05

0.06

0.26

Total

100.0

49.0

100.0

26.0

100.0

25.0

100.0

Quartz

21

40

25

11

Plagioclase

21

38

26

12

K-Feldspar

18

36

23

11

Oxides

12

13

17

9

Muscovite

14

24

12

7

Biotite

14

21

12

8

Epidote

9

24

9

7

Tourmaline

11

22

11

8

Other Silicates

7

10

6

6

Other

6

9

7

5

Table A-1 QEMSCAN results for the combined concentrates of the muscovite flotation tests Size Distribution (%) Flotation Sample

Feed

High Collector + Depressant

High Collector, No Depressant

Low Collector + Depressant

Mineral Combined

+75um

-75/+38um

-38um

Muscovite

9.08

11.16

5.48

8.74

Quartz

62.17

68.70

63.97

47.51

Plagioclase 5.58

4.26

6.90

6.78

K-Feldspar

19.61

13.67

21.22

29.58

Other

3.57

2.21

2.44

7.39

Muscovite

27.45

62.62

17.36

11.13

Quartz

38.11

21.56

43.78

45.56

Plagioclase 3.82

1.17

4.56

5.06

K-Feldspar

25.14

9.60

28.55

32.61

Other

5.48

5.06

5.74

5.63

Muscovite

11.67

14.74

7.72

7.77

Quartz

58.57

62.78

61.60

49.69

Plagioclase 5.18

4.04

6.14

6.83

K-Feldspar

21.99

16.94

22.12

31.07

Other

2.60

1.50

2.42

4.64

Muscovite

52.68

86.66

48.80

25.95

Quartz

16.57

3.70

16.47

27.24

Plagioclase 2.02

0.04

2.12

3.63

K-Feldspar

20.22

3.02

22.92

33.49

Other

8.51

6.57

9.68

9.69

Highlights   

Zeta potential measurements of muscovite and gangue measured Effect of flotation reagent addition analysed Separation of muscovite from gangue through flotation was related to zeta potential