UPGRADING OF LOW-GRADE GOLD ORE SAMPLES FOR IMPROVED PARTICLE CHARACTERISATION USING Micro-CT and SEM/EDX

Transcription

1 UPGRADING OF LOW-GRADE GOLD ORE SAMPLES FOR IMPROVED PARTICLE CHARACTERISATION USING Micro-CT and SEM/EDX Eric A Agorhom*, W. Skinner, Max Zanin Ian Wark Research Institute, University of South Australia, Mawson Lakes, SA 5095, Australia. * agoae001@mymail.unisa.edu.au ABSTRACT There is clearly a major challenge in characterising gold (Au) mineral particles in lowgrade plant samples. This is particularly true for automated mineralogical tools such as X-ray micro-computed tomography (Micro-CT) and scanning electron microscope (SEM), where the need for statistically meaningful numbers of particles requires many sections to be analysed. In this work, an existing Vertical Gas Stream (VGS) apparatus has been modified to Vertical Water Stream (VWS) system for Au particle upgrading to enable efficient characterisation using Micro-CT. The technique is based on a single cylindrical column through which fluidising water is used as the separating medium. The elutriator was calibrated using galena (specific gravity, S.G. of 7.58) and quartz (S.G. of 2.65) particles less than 38 µm as model minerals. From the calibration tests, partition curves as a function of particle size were generated. Using these measurements, theoretical partition curves for Au (S.G. of 19.3) have been calculated. Upgrade tests were carried out on a -38 µm ore plant sample of known Au content, and the concentrate was characterised using Micro-CT and compared with SEM coupled with energy dispersive X-ray (SEM/EDX) analysis of µm VGS concentrate. The Micro-CT analysis of VWS Au concentrate showed that sufficient Au particles can be upgraded using VWS system in the -38 µm size fraction. SEM/EDX results indicate that regrind does not affect changes in free Au particle morphology, aspect radio and frequency of shearing damage in the µm size fraction. Cyclone classification of the regrind mill discharge in the µm size fraction appears to perform surface cleaning by exposing obscuring silver (Ag) surfaces on Au particles in the mill discharge sample. Keywords: gold; galena; quartz; vertical water stream elutriation; microtomography, surface analysis INTRODUCTION Optimisation of Au recovery in Cu-Au flotation requires accurate and complete mineralogical and morphological characterisation to diagnose the possible Au loss routes. Mineralogical tools such as X-ray microtomography (Micro-CT) and scanning electron microscopy coupled with energy dispersive X-ray spectrometer (SEM/EDX) can be used. The major problem associated with Au mineralogical analysis is due to

2 sparsely distributed Au grains and their low concentrations in these ores (i.e. 1-2 g/t) (Bulatovic, 1997; Allan and Woodcock, 2001; Leaver and Woolf, 1931). Hence, there is the need to pre-concentrate or upgrade Au particles and increase grade from any gold plant sample before mineralogical analysis is performed, without changing Au surface chemistry. In recent times, researchers have proposed the use of pre-concentration methods such as heavy medium separation (Zhou & Cabri, 2004), shaking Wilfley table (Laplante et al., 1995) and hydro separation (Lastra et al., 2005) for upgrading Au samples to enhance the identification of a significantly statistical number of gold particles. These pre-concentration methods depend on Au grain size and most of them are expensive to operate and toxic, particularly with regard to the chemicals used in heavy medium separation. The vertical gas stream (VGS) elutriator which is central to this study has been used to successfully upgrade grades of Au plant samples from as low as 0.01 g/t and statistically characterise Au particles on size-by-size basis while preserving surface chemistry (Georgiev et al., 2003). Preliminary studies have shown that dense mineral particles could be concentrated by selectively removing low specific gravity materials using the VGS system. This technique is based on sample pre-sizing and the use of a single cylindrical column through which upward-flowing gas (compressed air) is used as the separating medium. However, low specific gravity of gas makes this technique inefficient for upgrading fine particles (less than 38 µm) due to particle aggregation. Hence, for easy and effective bulk and surface characterisation, there is the need to design a new system for upgrading fine Au particles. In this study, the existing VGS elutriator was redesigned as a vertical water stream (VWS) elutriator to enable upgrading of fine (less than 38 µm) Au particles in Au ore samples. This new VWS system operates using water as separating medium. The VWS system was calibrated using ground galena and quartz as model minerals passing 38 µm and validated with less than 38 µm sized real Au ore sample. The VWS system Au concentrate was characterised using X-ray microtomography for 3D liberation and mineral surface exposure. Also, the effects of particle morphology, surface chemistry and physical factors contributing to the rejection of free Au in the µm size fraction when VGS system is used to upgrade copper-gold flotation circuit samples (Indonesia) has been evaluated. EXPERIMENTAL SECTION Vertical gas stream upgrading (VGS) The VGS process was used to upgrade the µm size fraction of different process streams from the same gold-plant as the VWS samples: regrind mill discharge, regrind cyclone underflow, combined rougher concentrate and tailings. The stream samples were sieved in order to remove the -38 µm fraction until about 100 g remained in the µm size fraction. Each of the sample streams were upgraded using the VGS system in a similar way as used by Georgiev et al.(2003) until about 0.5 g of each sample was remained as VGS final concentrates (heavy fractions). Here, nitrogen gas was employed as the separating medium. Magnetic particles were removed and several SEM stubs were then prepared from the remaining heavy materials. 2

3 Vertical Water Stream (VWS) Elutriator Design The vertical water stream elutriator was used to upgrade fine Au particles (-38 µm) that could not be upgraded by VGS system. Classification of fine particles is more effective in wet conditions than dry conditions because the particles are well dispersed within the wet system, which enhances efficient separation. The VWS consists of a single cylindrical column through which fluidising water is used as the separating medium. This is similar to the existing VGS system at the bottom part with the top coupled with a micro-flotation unit as shown in Figure 1. The diameter of the main glass column is 3 cm. The working distance between the collector and water inlet is 24 cm. The distance between the inlet tube and the bottom of the water flow tube is 1 cm. Fig. 1: VGS system (left) and VWS system (right) schematic Sample Preparation and VWS Calibration Tests The VWS system was calibrated using galena and quartz as model minerals. Galena and quartz mineral samples were hand ground in a ceramic mortar and pestle and each fractionated to -38 µm by sieving. 10 g of each sample was mixed with 250 cm 3 of water (S.G. of 1). The resulting pulps were elutriated in the VWS system for 1 hour using water at flow rates varying between ml/min. After the elutriation, both the lighter and heavy fractions were collected and weighed as tailings and concentrate respectively. Both concentrate and tailings samples were size analysed using a Malvern Mastersizer X laser diffractometer to obtain the size distribution. An attempt was also made to upgrade a real copper-gold flotation circuit sample to validate the calibration test. The gold plant samples (20 g, -38 µm) with known Au content were upgraded using the VWS system for about 1 hour until less than 1 g remained for each elutriation as concentrate. The upgrade experiment was repeated 10 times and these VWS stage concentrates combined and further upgraded until less than 1 g remained in the final VWS concentrate. Magnetic particles were removed from the final VWS concentrate in a similar way as used by Georgiev et al. (2003) and characterised using micro-computed tomography. 3

4 Surface Analysis X-ray microtomography (Micro-CT) X-ray computed microtomography is a non-destructive technique which has the ability to produce three-dimensional (3D) image of structures and their internal features at micro-scale levels (Kyle & Ketcham, 2003; Lin & Miller, 1996; Miller & Lin, 2004). This technique has been used widely in medical research and hospitals for scanning soft tissues (bones). It is finding wider application in the mineral processing industry for liberation, mineral surface exposure analysis and analysis of pore structures for fluid flows in heap leaching operations. For liberation analysis, for example, Micro-CT has an advantage over the existing techniques such as QEM-Scan which involves analysis of polished sections to produce one and two-dimensional information. X-ray microtomography technique is non-destructive, less time consuming and its sample preparation does not involve complex polished sections. In this study, the VWS concentrate sample was analysed by microtomography using an Xradia Micro-CT 400 instrument. The VWS concentrate was filled into a Kapton capillary tube (ID of ) and rotated over a full 360 during the X-ray imaging resulting in the production of 2D images which are further treated with the reconstruction software to produce 3D density map of the sample. Scanning Electron Microscopy/Energy dispersive X-ray (SEM/EDX) A Camscan (CS44FE) scanning electron microscope was used equipped with an energy dispersive X-ray spectrometer and a backscattered electron detector. The accelerating voltage was set at 20 KV and stage tilt of 10 and working distance of 35 mm. From each upgraded heavy concentrate, a total of 4 SEM samples were prepared and mounted on aluminium SEM stubs with 1 cm 2 double-sided, electrically conductive tape and carbon coated prior to analysis. SEM/EDX analysis was performed on a subset of stubs during which the Au particles were observed, counted and measured for aspect ratio. Backscattered electron images were acquired and quantitative EDX analysis performed on apparently clean regions of Au particle surfaces and region obscured by attached materials or coatings. Free Au particle counts were used to estimate the head grade for µm fraction of the respective sample streams. These results were compared with fire assay data. The maximum and minimum dimensions of Au particles were measured and converted to a two-dimensional aspect ratio for the rougher concentrate; regrind mill discharge and regrind cyclone underflow. RESULTS AND DISCUSSION VWS Calibration Test Size-by-mass distributions of galena and quartz VWS concentrate and tailings were calculated from the results as shown in Figure 2a (for galena VWS products). From the calibration tests, partition curves for galena and quartz as a function of particle size were generated. The unhindered settling equation (Eq 1) was used to generate a theoretical partition curve of gold taking into consideration the specific gravity of gold particles. The partition curves of galena, quartz and gold are shown in Figure 2b. This clearly showed that fine Au particles can be upgraded using VWS system. From the 4

5 partition curves, it is possible to estimate theoretically the percentage of specific Au particle size ranges that can be upgraded within a particular Au sample. V 2 2gR ( ρ P ρ f ) = (1) 9µ Where: V is the particles settling velocity (m/s) g is the gravitational acceleration (m/s 2 ), 9.81 m/s 2 ρ p is the mass density of the particles (Kg/m 3 ) ρ f is the mass density of the water (Kg/m 3 ) µ is the viscosity of water (Kg/ms), Kg/ms R is the mean radius of each of the Au particle sizes (m) (a) (b) Fig. 2: (a) VWS galena product size distribution and (b) partition curves of quartz, gold and galena As indicated in Figure 2b, the partition coefficient represents the fraction of particles of a given size that can be upgraded into VWS concentrate. For example, for a sample within size range µm, about 50 % of the Au particles in the sample will report to the concentrate after upgrading with the VWS elutriator, while quartz particles will be almost completely rejected. The optimum upgrade of Au particles seems to be achieved in the size range ( µm). In this size fraction, about 70 % of the Au particles can be upgraded effectively by the VWS system while the associated gangue minerals will be almost completely rejected. The behaviour observed in the top portions of the partition curves needs to be investigated further. The apparent jump to high recovery is not easily explained as the particle numbers in these measurements were low. 5

6 VGS Au collection and upgrading An attempt was made to collect and upgrade free gold and gold-rich composites from the regrind mill discharge, regrind cyclone underflow, combined rougher concentrate and tails to ease and enable efficient gold particle characterisation. Table 1 summarises the results for the upgrading of the µm fraction of the four sample streams. It is observed the calculated head grades via the VGS method parallel the fire assay results, although there is a maximum discrepancy of a factor of 2.5. It is likely that a closer correlation would be achieved had the entire VGS concentrate been analysed in each case. The VGS pre-concentration method had not only increased gold grades but also minimised the number of polished sections to be prepared and analysed. This reduces analysis time and become less expensive. Table 1: VGS upgrading, calculated head and fire assay data for the µm fraction of the four streams VWS Collection and Upgrading X-ray Micro-CT analysis Figure 3 shows a microtomography scan of a VWS concentrate obtained from vertical water stream elutriator upgrading of a gold ore sample. The cross sections are twodimensional (2D) images extracted from the microtomography scan. The white features in Figure 3a, b represents Au particles with high X-ray attenuation coefficient (higher density) whereas the light grey features represent the associated gangue minerals with low X-ray attenuation coefficient within the VWS concentrate sample. The black parts of these images represent the background (air). The identified Au particles are sparsely distributed within the sample with the surfaces less coated appearing brighter than the severely coated surfaces. Figure 4 demonstrate how X-ray microtomography can be used to produce 3D spatial distribution of different mineral phases within a particular sample. The cross sections in Figure 4 represent the 3D reconstruction of the VWS system Au concentrate where the contrast threshold is increased from (a) through (c) and (d) to show all phases up to the highest density minerals. Here, the gray scale levels of the images illustrate the relative attenuation coefficients present in the sample. X-ray microtomography can also be used for mineral liberation and surface exposure analysis. From the 3D image analysis as shown in figure 4a, b, c, d, most of the observed free Au particles are coated with extensive patches and insufficiently liberated particularly for the relatively fine free Au particles, can be detrimental for flotation. From this result, it 6

7 appears that high resolution X-ray microtomography can be used to diagnose Au particle flotation if a correct correlation is developed between the percentage of exposed mineral surface and Au particle size distribution, similarly to the approach by Miller and Lin (2003) for the optimisation of heap leaching operations. 100 µm 100 µm Fig. 3: Vertical water stream elutriator concentrate constituents, gold and other associated gangue mineral phases identified from selected cross sectional image of the 2D reconstruction of an upgraded gold sample. Fig. 4: Micro-CT of a sample of a VWS elutriator concentrate, the contrast threshold is increased from (a) through (c) and (d) to show all phases up to the highest density minerals. The bed of particles was 1mm in diameter. Au particle morphology, composition and surface Au particle morphological properties of the four sample streams upgraded by VGS were characterised by SEM/EDX. Initially, SEM/BSE was used to determine the Au particle morphology and composition. Figure 5 (a, b, c, d) illustrate Au particles found in regrind mill discharge, regrind cyclone underflow, combined rougher concentrate and tail respectively. The observed Au particles in these sample streams are very similar in terms of coverage of attached material, impacted material and shape. The coverage of these attached materials varies between 5-40 % of the visible surface. 7

8 Fig. 5: Backscattered electron images of free gold particles in the (a) Regrind Mill Discharge (b) Regrind Cyclone Underflow (c) Combined Rougher Concentrate and (d) Combined Rougher Tail of µm fraction upgraded by VGS. An aspect ratio plot in Figure 6a shows a similar distribution suggesting that, in this size range ( µm), there appears to be little effect of regrind on Au particle shape within the regrind circuit. Again, comparison with average aspect ratio measured for the rougher concentrate (Figure 6b) also indicated that little or no change in Au particle shape resulted from regrinding. It is apparent that Au surface was coated with patchy material coatings. Fig. 6: (a) 2D aspect ratio (major axis/minor axis) distribution as a percentage of Au particles examined in the MD (90 particles) and CUF (b) Gold particle aspect ratio comparison, including the Rougher Concentrate (feed to regrind circuit) 8

9 Au particle composition and surface chemistry Energy dispersive X-ray analysis (EDX) was performed on apparently clean regions and attached materials found on Au surface in each of the sample streams. Figures 7 (a, b) and 8 showed the EDX spectra and the graphical representation of the statistical data obtained for this size fraction ( µm). The analysis of the unobscured regions reveals low Ag (average < 5 wt %) except for rougher concentrate; contain Cu, Fe, S, Si and Al as impurities with low levels of oxygen. Sulphur content is high (5-11 wt %) for all sample streams. The bulk composition of Au is 70 wt %. The attached material is characterised by Fe, S, Cu and higher levels of oxygen. The higher levels of Fe, S and O in the attached material suggest the presence of iron oxide and discrete pyrite as Au surface coatings. The surface coating coverage varies between 5-40 % of the visible surface. Figure 8 also compares the average elemental exposures on apparently clean Au surfaces. The significant difference is the anti-correlated exposure of Ag with Fe, Si and Al. Fe, Si and Al increase in the mill discharge obscuring Ag due to increased in attached/coatings on Au particle surface but is apparently removed after cyclone separation, re-exposing Ag signals. This reappearance of Ag on cyclone underflow Au particles may be due attrition during cyclone separation. Since the Ag content across all the sample streams is low, the Au particles could only be recovered with hydrophobic collectors such as 2-mercaptobenzothiazole (2-MBT) or dithiophosphates (DTP) Fig. 7: Typical SEM/EDX spectra of (a) clean regions and (b) attached material. 9

10 Fig. 8: Comparison of apparently clean surface (i.e. no visible attached material) of gold particles found in the Rougher Concentrate, Mill Discharge and Regrind Cyclone Underflow. CONCLUSIONS The upgrade of fine Au particles (-38 µm) using VWS has been investigated using low grade copper-gold processing plant samples. The upgrading method has been calibrated using galena and quartz model mineral particles. Furthermore, the VGS upgrading method has been used successfully to upgrade the µm size range of the ore sample. Upgrading is critical to increase the statistical significant number of Au particles found by SEM/EDX analysis. This is a major problem in image analysis since the analysis of low grade Au samples using SEM/EDX could not guarantee to return a representative result. X-ray microtomography was used to investigate and examine the surface exposure of free Au particles in the VWS upgraded Au concentrate. The microtomography 3D images obtained show that Au particles are partially liberated with extensive surface patches. The analysis of different plant streams suggests that regrinding does not affect changes in free Au particle morphology, aspect ratio and frequency of shearing damage. Cyclone classification of the regrind mill discharge appears to perform surface cleaning by exposing obscuring Ag surfaces on Au particles in the mill discharge sample. Ag content on the Au particles was generally low; hence, it is only more hydrophobic collectors such as MBT or MBT/DTP blends that can recover Au particles in this size fraction. ACKNOWLEDGEMENTS The financial support from AMIRA International and permission to publish this work is strongly acknowledged. 10

11 REFERENCES Allan, G. C. & Woodcock J. T. (2001) A review of the flotation of native gold and electrum, Miner. Eng. 14, Bulatovic, S.M. (1997) Flotation behaviour of gold during processing of porphyry copper-gold ores and refractory gold-bearing sulphides, Miner. Eng. vol.10, 9, Georgiev, T.V., Skinner, W.M. & Grano S.R. (2003) Size dependent characterisation of gravity recoverable gold from the rougher flotation at OK Tedi Mining Limited, 10 th Balkan Mineral Processing Congress, Verna, Bulgaria, Kyle, J.R. & Ketcham, R.A. (2003) In situ distribution of gold ores using highresolution X-ray computed tomography, Eco. Geol. 98 (8), Laplante, A. R., F. Woodcock & M. Noaparast (1995) Predicting gravity separation gold recoveries, Miner. Metal.Proc. 12, Lastra, R., Price, J., Cabri, L.J., Rudashevsky, V.N. & McMahon, G. (2005) Gold characterisation of a sample from Malartic East (Quebec) using concentration by hydroseparator, Treatment of Gold Ores. In: 5 th International Symposium, 44 th Annual Conference of Metallurgists of CIM, Calgary, Alberta, Canada, MetSoc: pp Leaver, E.S., and Woolf, J.A. (1931) Flotation of minor gold in large-scale copper concentrators, AIME, Technical Publication No.410, p.3 Lin, C.L. & Miller, J.D. (1996) Cone beam X-ray microtomography for threedimensional liberation analysis in the 21 st century, Int. J. Miner. Proc. 47 (1-2), Miller, J.D. & Lin, C.L. (2003) 3D analysis of particulates in mineral processing systems by cone beam X-ray microtomography, XXII International Mineral Processing Congress, Cape Town, 29 September-3 October, 2003, pp Miller, J.D. & Lin, C.L. (2004) Three-dimensional analysis of particulates in mineral processing systems by cone bean X-ray microtomography, Miner. Metal. Proc. 21 (3), Zhou, J. Y. & L. J. Cabri (2004) Gold process mineralogy: Objectives, techniques, and applications, JOM, 56, BRIEF BIOGRAPHY OF PRESENTER Eric Aminartey Agorhom is currently in his first year of a 3-year PhD programme at the Ian Wark Research Institute (IWRI), University of South Australia. He obtained his Bachelor of Science (Hons) in 2009 from the University of Mines and Technology, Tarkwa, Ghana. 11