121

Canadian Metallurgical Quarterly, Vol 39, No 2, pp 121-128, 2000 © Canadian Institute of Mining, Metallurgy and Petroleum Published by Canadian Institute of Mining, Metallurgy and Petroleum Printed in Canada. All rights reserved

A STANDARD MATERIAL FOR STEREOLOGICAL CORRECTION
D. LIN1 and J.A. FINCH2
2 1 Cominco Ltd., P.O. Box 2000, Trail, B.C., Canada, V1R 4S4 McGill University, Department of Metallurgical Engineering, 3610 University St. Montreal, Quebec, Canada, H3A 2B2

(Received March 1999; in revised form September 1999)

Abstract — The determination of mineral liberation by microscopic examination of ore particle sections is stereologically biased. Procedures have been developed to correct this bias, but most have not been thoroughly tested. A standard material, a material of known particle composition and locking characteristics, would allow the proper testing of these procedures. In this work, a method was developed for the creation of a standard material consisting of two phases: glass and lead borate. Simple-locked particles and particles with granular and layered textures can be created. The standard material is versatile and easy to use. Résumé — La détermination de la libération de minéraux par examen microscopique de sections de particule de minerai est biaisée stéréologiquement. Des procédures on été développées afin de corriger ce biais, mais la plupart n'ont pas été entièrement évaluées. Un matériel étalon, un matériel de composition de particules et de caractéristiques d'emprisonnement connues, permettrait une évaluation appropriée de ces procédures. Dans ce travail, on a développé une méthode pour la création d'un matériel étalon consistant de deux phases: verre et borate de plomb. Des particules à emprisonnement simple et des particules avec des textures granulaires et avec couches peuvent être créées. Le matériel étalon est versatile et facile à utiliser.

INTRODUCTION Mineral liberation represents the ultimate limit in the separability of mineral phases. The measurement of mineral liberation is usually accomplished by mounting the particles in a resin matrix, polishing a surface and examining the surface with an optical or electron microscope. Images of the surface are digitised and processed with an image analyser. The image analyser measures the liberation by determining the mineral composition of the individual particle sections. This method of measuring liberation is stereologically biased. Liberated particles will always yield liberated sections whereas locked particles can yield liberated or locked sections. As a result, mineral liberation will always be over-estimated. Methods have been developed to correct the stereological bias [1, 2, 3, 4, 5, 6, 7], but the effectiveness of these methods is difficult to test since the true liberation is not known. Standard materials, that is, materials of known particle liberation and locking characteristics, have been developed in an attempt to allow the assessment of these correction procedures. The present work describes a method of producing a versatile standard material that can be used to test the robustness of stereological correction procedures.

For the standard material, a two-phase (binary) system is all that is required because liberation is measured one phase at a time: while the liberation of one mineral phase is being measured, the rest of the mineral phases in the sample can simply be grouped together as the second phase. A true, two-phase material is essential if density fractionation of the particles is performed to provide the independent and true measure of particle composition. If there are more than two phases, there will not be a direct correlation between particle density and particle composition. Obviously, a significant density difference between the two phases is necessary to provide good resolution to the density splits. A near-perfect separation such as that obtained using heavy liquids is desired to ensure the precision of the splits. For the standard material, there must be a method of distinguishing the two phases. If optical microscopy is used, the two phases must be different colours or have some other differentiating feature. If electron microscopy is used, the two phases must have different grey levels when viewed with backscattered electron imaging. The backscattered grey level of a phase is directly related to its average atomic number such that a low atomic number material will appear dark and a high atomic number material will appear
CANADIAN METALLURGICAL QUARTERLY

STANDARD MATERIAL PREPARATION A SHORT REVIEW Naturally-occurring standard materials Naturally-occurring standard materials are created from naturally-occurring two-phase ores. The sphalerite/dolomite particles were density fractionated using the Magstream separator (a centrifugal magnetogravimetric separator). Monosize grains of one phase are embedded into a matrix which acts as the second phase. Work performed in this area is described below. The use of a naturally-occurring standard material is limited because the locking texture cannot be controlled. The density of the dolomite was 2. 2. 1. The densities of the silicate and iron oxide phases were 2.10 g/ml. screened and the composition of the particles measured by density fractionation. The particles were separated based on density using heavy liquids (for densities below 3. When the plastic hardens. Both ores were crushed and screened to 417-595 µm and the iron-oxide/silicate particles fractionated using heavy liquids. Although a stereological correction procedure may be effective with one type of locking texture.00 g/ml. The ore is broken. it may not be as effective with others.72 and 5.118)] and there was some variation in the density of the iron minerals. only one interface between the phases in the particle). If there are significant quantities of either of these in the particles then the determination of particle composition from density measurements is not accurate. the mineral which fails to be liberated is most likely concentrated in the next simplest class the simple-locked particle class. 4. The standard material should also be flexible enough to simulate different types of locking textures. Simple locking also creates the most severe stereological bias and is. 3. Work that has been done in this area is described below: 1. FINCH bright. It is desirable that the standard material be able to exhibit simple locking (i.32 g/ml) and a fluidised bed (for densities above 3. Woollacott and Valenta [12] embedded grains of a polyester resin which had been doped with lead oxide powder and black dye into a matrix of the same polyester resin in liquid form. a severe test of any stereological correction procedure [1. In addition. Miller and Lin [10] used a unique approach to determining the true liberation of mineral particles. it is too tedious and expensive to use in regular liberation analyses. The ore was crushed to 90% -1.06 g/ml. The lead oxide powder . They used serial sectioning to determine the liberation of an iron ore and a copper ore. A short review of previous attempts to create standard materials follows. the standard material was designed for use with electron microscopy.65 g/ml and the iron oxides to have a density of 5. this would be a problem if there were variations in the density of either of the two phases. This procedure allows the manipulation of parameters such as the grain and particle size so that there is some control over the locking texture. The composition of each particle was reconstructed based on these sections. Although serial sectioning does provide an unbiased measure of the true liberation distribution. 2.0 mm and screened. a series of parallel sections was gathered for each particle. respectively. There were two problems with this standard material: there may have been significant CANADIAN METALLURGICAL QUARTERLY porosity in the ore [6(p. The block is then crushed producing free and locked particles.2]. These particles are screened into different size classes and a density fractionation is performed. Bole et al.e. There have been several efforts to develop a standard material from natural sources. Particles were screened to 74-105 µm and mounted in resin. it would be instructive to examine the effect of locking texture on the predictive abilities of different liberation models. The siliceous material was estimated to have a density of 2.122 D. Similarly. but there are difficulties in finding a two-phase ore whose phases are free from porosity and impurities. therefore. The locked blocks were crushed after treatment with liquid nitrogen to make the polystyrene more brittle. doped with white dye.A. [9] examined two binary ores: an ironoxide/silicate sample and a sphalerite/dolomite sample. In this work.32 g/ml). By polishing this pellet down by intervals of 18 µm.85 g/ml and that of the sphalerite. A complete density fractionation could not be performed since the highest density that could be reached using these liquids is about 2 g/ml and the density of pyrite is about 5 g/ml. Simple locking is probably the most important of all locking types in mineral processing following the argument that since comminution is aimed at creating free particles. Bagga [11] created a standard material using 600-710 µm pyrite particles as the grain material and polystyrene as the matrix. Artificial standard materials The general procedure to create an artificial standard material is summarised in Figure 1. There are two categories of standard materials: naturally-occurring and artificial. LIN and J.. Several particle sizes were separated by screening and a density fractionation was performed using CaCl2 and ZnCl2 solutions. a two-phase block is formed. Stewart and Jones [8] used a siliceous iron ore which consisted of grains of iron oxides (hematite with magnetite and minor amounts of goethite and limonite) in a siliceous matrix (quartz with minor amounts of other silicates). The matrix phase is usually a plastic or resin in liquid form.

the density of the denser of the two phases should not exceed the upper limit of the separation method. By producing a strong bond between the two phases. Grinding the grain-matrix blocks. The separations were performed using a heavy liquid (ZnCl2 solution). grains that are close to each other will almost inevitably create complex locking (locking such that there is more than one interface between the phases in the particle). Fig. creating other locking textures. but this may seriously reduce the amount of locked particles produced. particles with a layered locking texture (a texture consisting of alternating layers of different phases) may result if an ore with a banded structure is crushed. The densities of the two phases were 1. strong interfacial bonding would allow the standard material to provide a test of models of liberation by size reduction. The silica density was 2. were generated by screening and separated using a ZnCl2 solution. a water-soluble heavy liquid. 4. this material was designed for optical microscopy.219 g/ml.62 g/ml. With the current procedure for producing locked particles.36 and 1. this is not necessarily always the case. Density classification.756 and 1. The greater the difference in density.62 g/ml and the resin density was 1. 1 Conceptual procedure for the production of artificial locked particles. However. Also. the two materials should be significantly different in density.35-4. 2. CANADIAN METALLURGICAL QUARTERLY . 2. strengthening the bond between the grain and matrix material. increased the density of the grain material to allow density fractionation and the dyes provided optical differentiation between the grain and matrix phases. In any case. Ideally. there would be a decrease in breakage along the interfaces and an increase in the amount of locked particles.A STANDARD MATERIAL FOR STEREOLOGICAL CORRECTION 123 Refinement of the standard material 1. only granular locking can be created. Lead borate is a leaded glass additive and is more commonly • 2. the locked blocks were crushed and screened to 2. Two particles sizes. there are several parameters that can be changed to enhance the quality and the quantity of the artificial standard material. Size classification of the particles 4. The densities of the grain and matrix phases were 1. developing a strategy for producing only simple-locked particles. RESULTS Selection of the phases for the standard material Two new materials were selected for the production of the standard material: lead borate (2PbO B2O3) as the matrix material and glass particles as the grain material [7].14] created their standard material by embedding grains of silica in a matrix of epoxy resin. 4.80 mm and 3. simple locking can be produced only when the particle size is significantly smaller than the grain size. changing the grain and/or matrix material.36-2. Grains of a polyester resin doped with zinc oxide powder and a coloured dye were embedded in a polyester resin doped with a different concentration of zinc oxide and a different coloured dye. When the matrix resin had hardened. it is very difficult to control the amount of simple locking due to the granular nature of the texture: unless the grains are perfectly dispersed in the matrix.00 mm. While this may reflect the situation in many ores. Lin et al. This is complicated by the fact that it is impossible to measure the degree of simple locking since stereological bias affects the determination of the locking texture. 3. the greater the resolution of the density separations.22 g/ml. Gay [6] created a standard material in much the same manner as Woollacott and Valenta. 3. For instance. In addition. respectively. 3. The particles were separated using sodium polytungstate. The particle size was 75-106 µm. With the current procedure for manufacturing locked particles. Embedding of monosize grains into matrix. The standard material could be improved by: 1.0 mm. [13. Although artificial standard materials have been established. The amount of locked particles produced should be substantial enough so that large amounts of grain and matrix material do not have to be processed. the phases for the standard material should be brittle so that they are easy to crush and possess good polishing properties so the interfaces are clearly delineated when viewed under a microscope.5-4.

Standard deviation in parenthesis Production procedure for standard material with a granular texture The procedure for the creation of the glass/borate standard material with a granular texture was as follows: 1.124 D.18. The crucible was placed in an induction furnace preheated to 600°C. 1 4. The complexity of the complex-locked sections was low. 3. but the use of glass rather than silica allows greater control over the locking texture (as seen by the creation of the layered texture discussed later) and because it has a lower melting point. Subsequent experimentation showed that after contact some glass diffused into the lead borate to give a SiO2 content of 5.006)1 and 2.052 g/ml (±0. This temperature was maintained for 30 minutes. .95% (±1. dark phase is mounting medium). The density of the borate was redetermined as 6. Thirty grams of 1180-1700 µm glass grains were placed in a 40 ml porcelain crucible.19].A. but they were small (<10 µm) and were present in only approximately 1% of the borate sections. FINCH used as a source of lead for glass melts or ceramic glazes. crushed and wet-screened to remove the fines. The density of the lead borate and glass were measured with a water displacement technique using a pycnometer bottle [15] to be 6. 6. respectively. It appeared that there were slightly more simple-locked sections than complex-locked sections. The melting point of lead borate is approximately 500°C while the melting point of glass is approximately 800°C.833 g/ml (±0. Sixty grams of powdered borate was placed on top of the glass grains. The standard material was created by melting the lead borate and embedding the glass particles. 5.) The size range for the standard material was selected as 425–600 µm since 600 µm particles are the largest that the Magstream Model 100 can accommodate and splits on the Magstream are more precise the larger the particle size. LIN and J. but in this work it was used as one of the phases of the standard material. 2. The 425-600 µm fraction was screened out. A large majority of the sections exhibit locking. conventional heavy liquids could not be used to fractionate the standard material so a Magstream Model 100. CANADIAN METALLURGICAL QUARTERLY Production procedure for standard material with layered texture and simple locking To create a standard material with a layered texture. (The principles of the Magstream separator have been described by Walker. The sample was removed from the crucible. Devernoe and co-workers [16. The procedure for the creation of the glass/borate standard material with simple locking was as follows: 1.17. The glass was three Tyler size classes larger than the target particle size to aid in creating simple locking.009) and this was the value used in the density fractionation. This permitted the long glass slides to be parallel to each other and separated by a gap of exactly 1 mm.) Glass was used as the grain material because it has similar brittleness and breakage properties to lead borate. Simple-locked particles were created from this texture by crushing blocks of this material to a size below the thickness of the layers as shown in Figure 3. The glass particles softened slightly and produced a strong bond with the lead borate. grey phase is glass. most of the complexlocked sections contained only two interfaces. Since the glass has a lower density than the borate. 2 SEM backscattered electron micrograph of 425-600 µm glass/borate standard material particles taken at 30x magnification (light phase is lead borate. A SEM (scanning electron microscope) backscattered electron micrograph of this material is shown in Figure 2. Due to the high density of the lead borate.38%) as determined by microprobe analysis. alternating layers of the two phases were bonded together.001). (It was later determined that some glass diffused into the borate which further strengthened the bond. was used. There were a few voids. Silica was considered as the grain material. the glass will be wetted by the borate as it rises to the top. Long glass slides (75 mm x 25 mm x 1 mm) alternated with short glass slides (25 mm x 25 mm x 1 mm) were clamped together at one end. 1 mm Fig. The power was shut off to the furnace and the crucible was allowed to cool to room temperature.502 g/ml (±0. a centrifugal magnetogravimetric separator.

3 Simple-locked particles can be created if a layered-texture block of material is crushed so that the particle size is smaller than the layer thickness. the particles were separated into different density classes using the Magstream separator. κchalcopyrite ≈ 4 x 10-4 SI units). crushed and wet-screened to remove the fines. for each composition fraction. A theoretical study by Svoboda [23] concluded that the performance of the Magstream was dependent upon particle size and he cautioned against using the Magstream for fine particles. In common with all separators. The particle size of the simple-locked particles was selected to be 425-600 µm because the granular-textured standard material was this size. Bunge and Fuerstenau [20.. The sample was removed from the crucible. Both materials showed no measurable diamagnetism. glass/borate standard material particles taken at 30x magnification (light phase is lead borate. Only simple locking was observed. For instance. As expected. 515.A STANDARD MATERIAL FOR STEREOLOGICAL CORRECTION 125 2.5 vol. A SEM backscattered electron micrograph of these particles is shown in Figure 4.5% and 54.. three separations were performed at both split-points of the fraction..% CANADIAN METALLURGICAL QUARTERLY .% borate. 5. 6. +95 vol. Their diamagnetic properties were also measured. The clamp and the short glass slides were removed. both the glass and lead borate had very low magnetic susceptibility (κglass < 5. The 425-600 µm fraction was screened out.% borate fraction. In this work. 4. In the cleaning steps. To verify this. Fig.95 x 10-5 SI units. any particles that did not re-report to the composition fraction were eliminated.22] observed that the accuracy of Magstream separations decreases with particle size. Also. density fractionation was performed only on the granular-textured standard material particles. The standard material was separated with the Magstream into the following composition fractions: -5. 3. 1 mm Fig. 2. 4 SEM backscattered electron micrograph of 425-600 µm simplelocked. This temperature was maintained for 45 minutes. the particles would have to be the same size. grey phase is glass. If the simple-locked and the granular-textured standard materials were to be combined to form a liberation distribution. The crucible was placed in an induction furnace preheated to 600°C. Simple-locked and layered-texture particles are a useful supplement to the granular-textured standard material. The power was shut off to the furnace and the crucible was allowed to cool to room temperature. Therefore. no sections with complex locking were seen.32 x 10-5 SI units and κlead borate < 1. the short slides and the clamp itself were clear of the powder. separation on the Magstream is not perfect. dark phase is mounting medium). 15-25. for the 45-55 vol. for comparison. the magnetic susceptibility of the two phases of the standard material was measured using a Frantz isodynamic magnetic separator [24]. The products of the first separation (rougher) were cleaned twice. To ensure the accuracy of the separations.21. Density fractionation To determine the liberation. The density fractionation on the Magstream assumed that the particles were non-magnetic.. the split points were set to 45. 85-95. The slides were placed in a crucible containing borate powder so that only the lower portion of the long slides was in contact with the powder. 3. the following measures were taken: 1. 7.5% closer to the midpoint of the fraction. the heavies and lights products must be re-processed (cleaned). The split-points were set 0. To ensure separation accuracy.

805 3.6 106.126 D.% borate) 0. prevention of preferential settling (due to density differences between the particles) and preferential orientation (due to density differences between the phases in the locked particles) is important since this affects the IA grade.). 6. glass.455-4. the grade measured using these methods will be referred to as the pycnometer and IA grades.6 8. 2.850 4.176 3.5 19.7 57.875 +5.680 2. but helps minimise misplaced particles.520-5.532 4.1 .165-5. The results are shown in Table I.865 5. regardless.3 78. FINCH borate.520 5. A small quantity of the resin (0. the grade (i.0 48. It is known that the area grade is not stereologically biased and is identical to the volumetric grade [6(p.4 ml of the sample with an equal amount of a diluent material (425-600 µm graphite particles) and placing the mixture at the bottom of a 32 mm mold.% borate were considered to be free. The resin was prepared by mixing with the hardener and then centrifuging at 1600 rpm for 1 minute to remove any air bubbles.9 98..0 28.0 87.455 4.745 3.035 3. True free borate was created by placing borate powder alone in porcelain crucibles and heating to 600°C for 30 min.22)]. Afterwards. The IA grade is determined by microscopic examination of the particle sections.390 3.100-4. Measurement of the grade of the composition fractions Following Magstream separation.0 49. This leaves a 1% gap in composition between fractions.165 5. It is important to generate random sections.8 20.1 31.502 g/ml) Composition fraction (vol.647 6.8 ml) was added to the particles in the mold and mixed. the pycnometer and IA grades were measured for each composition fraction.035-3. In these samples.517 3. Table I — Pycnometer and IA grades of the composition fractions of the standard material (density of borate.680-3.6 38. the particles in each fraction should have very similar densities and consequently.9 68. For the standard material. precautions were taken to ensure random mounting of the particles. it was removed from the crucible and crushed to 425-600 µm. The sample preparation involved mixing 0. Henceforth.0 IA grade (vol.e. The particles that were -5 vol.% borate) 0.% borate or +95 vol.4 9.2 66..232 4. respectively. The diluent helps to support the sample particles in space (and thus reduce preferential orientation and the incidence of particles touching) and promotes random orientation [25].390-3. Two different methods were used to measure the grade: water displacement in a pycnometer bottle and image analysis.265 Pycnometer grade (vol. the percent borate) of the composition fractions was determined.7 40.875 Pycnometer density (g/ml) 2. But. The mold was then centrifuged at 1600 rpm for 5 minutes to remove any air bubbles that may have resulted from the mixing. 100% glass) were created simply by crushing glass to 425-600 µm.257 5. there should be little segregation.e. The mounting medium was Epofix resin (Struers Inc. Graphite was selected as the diluent since it has a low density and a grey level similar to the mounting medium resin when observed by backscattered electron imaging. In sample preparation. LIN and J.052 g/ml.810-5. True free glass particles (i.A. The two grades should be close to the midpoint. more resin was added to top off the mold.810 4. The total area of the two phases is measured with an image analyser.7 58. After the borate had cooled and solidified.6 77.% borate) -5 5-15 15-25 25-35 35-45 45-55 55-65 65-75 75-85 85-95 +95 CANADIAN METALLURGICAL QUARTERLY Magstream split-points (g/ml) -2.745-4.6 88.522 2. The small amount of resin helps inhibit preferential settling by restricting the space in which the particles can move.100 4.

Finch. Powder Technol. Jones. paper 4 (1977). L. 4. J.. The standard material can be used to test the steps in liberation analysis including stereological correction procedures. J. N.S. Lin. Barbery. C. Lin. Most of these models assume negligible breakage along grain boundaries.A. G. In more general terms. Gomez and N. McGill University (1997). thesis.D. 1017-32 (1996).%.B. M. Data correction in two-dimensional liberation studies. thesis. Bole. Lin. 8. Miller. The pycnometer grade of the +95 vol.A. Herbst. 12.. Lin. J. Mineral Process. Simulation of the liberation phenomena in mineral systems. C. 22. Miller and C. Vassiliou.. 14. (1991). Stewart and M. Hill. A standard material for liberation analyses and examination of the robustness of stereological correction procedures.sample preparation and image processing . These distributions can be used as a check on the effectiveness of stereological correction procedures provided the other components of liberation analysis . (Warrendale.L. 72. In this regard. Rowlands and J.. Use of the standard material The standard material can be used to produce different liberation distributions by combining the different composition fractions. there is a layer of borate at the bottom of the crucible (the glass grains float in the borate due to their lower density) that does not come in contact with the glass. Ph. Ph. Lin. McGill University (1990). Int. Treatment of polished section data for detailed liberation analysis. Mineral Process. 9(10).. A. 37.% borate fraction where it belongs. Gay.L. The known true liberation distribution can be compared with the corrected distribution. 11. Valenta. D. McGill University (1991). D.P. Miller. Use of synthetic ore particles to test a transformation function in liberation analysis. 2. Finch. P.A STANDARD MATERIAL FOR STEREOLOGICAL CORRECTION 127 For all the fractions. Process Mineralogy VII. the standard material can also be used to test liberation models.C. We have reported on this comparison elsewhere [26]. S. thesis.P. 351p. J. Liberation modelling using particle sections.D. Since the two phases of the standard material are strongly bonded to each other. 617-32 (1987). a standard material is superior to using computer-simulated particles since it involves all the steps of a liberation analysis. Lin and J. Jones. King and J.Eng. Experimental verification of the PARGEN simulator for liberation analysis. An artificial standard material for liberation analysis. 260 p. CONCLUSIONS A method was developed for the creation of a standard material consisting of glass and lead borate. Applied mineralogy (Oxford: Graham and Trotman Ltd.L. Bagga. J. Meeting B. This can be explained by considering the creation of the standard material.S. 6. Woollacott and M. Ph. Hill. both the pycnometer and IA grades fall in the composition fraction range and in most cases. CANADIAN METALLURGICAL QUARTERLY . although the source of error may be difficult to isolate. The borate density was calculated only for borate that had been in close contact with glass grains and where limited diffusion had occurred.A. C. 7. thesis. Solutions to the trans formation equation for volumetric grade distribution from linear and/or areal grade distributions.O. Sao Paulo. 5. The production of a standard material for liberation analysis. Ph.S. Mineral liberation: measurements. D. 3.L. Mineral Process.D.L. 41-58 (1988). Rowlands. R. Lin.D.L. they are close to the midpoint of the range and the differences in grades are not significant. (1987). Int. C. University of Queensland (1996).D. Although the glass is always in contact with the borate.D. 67. 12th Int. 131-8 (1992). PA: The Metallurgical Society of AIME). G. Applications of two-dimensional image analysis to mineral liberation studies. Proc.D. 15.P.S. J. Miller and J. simulation and practical use in mineral processing (Quebec: Les Editions GB). Powder Technol.are performed correctly.). 103-11 (1991). thesis. REFERENCES 1. 50. Minerals Engineering. Improved transformation technique for the prediction of liberation by a random fracture model. Particles with granular and layered textures were created along with particles exhibiting simple locking.% borate fraction was greater than 100 vol. Powder Technol. Schneider. G. 9. This standard material is flexible and easy to use. P. D. Determining the amounts and the compositions of composite (middling) particles. Hausen and D. 209-21 (1993). All the high density borate reports to the +95 vol. Carbon eds. C. 13. 55-63 (1989). Pennsylvania State University (1983).D. 10. Congress. the standard material would enable the whole procedure (not just stereological correction) to be tested. M.

W. Magnetogravimetric separation in a rotational device. 23. 105-16 (1991). Fuerstenau. J. Hess. Mag.G.the MC process.L.S. (TMS-AIME). 1999. Separation characteristics of a magnetogravimetric separator. Mineral Process. Lastra and J. 24. 359-67 (1988).L. Devernoe. Proc. W.S.O. 17. Finch. Dresden.A.W. Miner. Cannes. Walker. XVII Int.A.C. Proc. 31-42 (1991). Bunge and D. 307-16 (1985). M. Walker and A. submitted to Trans.L. 743-52 (1996). Mineral separations using rotating magnetic fluids. C. R. Notes on operation of Frantz isodynamic magnetic separator. A. Separation of non-magnetic minerals using magnetic fluids in a flowthrough MHS rotor. Frantz Co.L. R. J. Walker. Process Mineralogy VIII. Andres. Bunge and D. Carson and A.. Congress. 21.W. J. Comparison of stereological correction procedures for liberation measurements using a standard material. Stuart.W.128 D. Metall. CANADIAN METALLURGICAL QUARTERLY . Finch and J. Jan. 22. 209-14 (1990). Morrow. Minerals Engineering. 7(4). Process. R. S. W. Lin. Devernoe.C. A specimen preparation procedure for automated image analysis. Congress. LIN and J. Congress. R. Gomez. XV Int.S. 26. D. Svoboda. Fuerstenau. 7(3). D. Proc. A. Urbanski and G. Urbanski. Wilhelmy. A. M.. Rowlands. Vassiliou. 31. 25.S. Mineral Process. Mineral Process.H. 195216 (1991). Dresden. FINCH 16. eds. SME pre-print 90-192. 9(7).S.-F. Mineral Process. Urbanski and U..R. Salt Lake City (1990). Mainly "gravity" separations using magnetic fluids under rotation. M. 18.C. Electr. Int. XVII Int. M. A contribution to the theory of separation in a rotating ferrofluid. R.A. N.S. 163-84 (1996). A new method for the commercial separation of particles of differing densities using magnetic fluids . Fuerstenau. Sep. Devernoe. H. Devernoe and W. Walker. 19.. Mineral separations with a magnetogravimetric separator. Metall.S. Instn Min. 20. Bunge and D.