Search Documents

By H. Rush Spedden, Arvid Thunaes

Pilot plant test work in 1942 and 1943 showed that by a combination of deslim-ing, fine-size classification, and Sullivan deck concentration it is possible to recover heavy minerals such as cassiterite at least as fine as 10 microns in size. This appreciable improvement in gravity concentration practice has been substantiated by several full-sized plants. IN the past, mills treating ores of tin and tungsten by gravity concentration have recovered very little mineral finer than 325 mesh, although some form of slime concentration has been generally attempted by the use of buddies or round tables. This paper describes a series of pilot plant tests made in 1942 and 1943 in which the use of the Sullivan deck was investigated for the recovery of an appreciable amount of the fine values formerly lost. The investigation was initiated by the U. S. Government in an effort to increase the wartime production of tin. Bolivian milling practice at that time included jigging of the coarse sizes, tabling of the intermediate sizes on conventional shaking tables, and the use of buddies or round tables on the finest sizes. Flotation of the pyrite was employed at different stages of the treatment, depending upon the quantity present and the preference of the operator. Usually good recoveries and the bulk of the production were made in the jigging and tabling operations provided that these sections were not overloaded. Buddies and round tables were used for a small additional recovery if it could be shown that they could pay for their high cost of operation. Extensive test work established that by a combination of desliming, classification, and Sullivan deck treatment, cassiterite as fine as 10 microns (1500 mesh) could be recovered when proper conditions were maintained. This test work has been verified by several full-scale installations. The method is generally applicable to the preconcentra-tion of large tonnages of low-grade ores or tailings for the recovery of a small amount of a valuable heavy mineral. The Sullivan deck (or now the Denver Buckman tilting concentrator), which is the essential part of the method, was devised originally for the recovery of tin from tailings of the Sullivan concentrator, Kimberly, B. C. Since the material treated contained only 0.05 pct Sn, it was impractical to attempt to recover cassiterite finer than 500 mesh.*

Jan 1, 1951

By H. Rush Spedden, Arvid Thunaes

R. R. Knobler and F. E. Albertson—Following the testwork done by Thunaes and Spedden, a Sullivan deck plant was built for the Colquiri mill. This plant started to operate in April 1945 and continues in successful operation. Since May 1950 the Colquiri mill has operated with a completely revamped flowsheet at a rate of 1000 tons per day. The old Sullivan deck section was retained with this new flowsheet. The "slime plant" in Colquiri consists of one 12-cell Denver No. 18-Sp. Sub-A flotation machine, one 16-ft and one 12-ft hydroseparator, six five-deck Sullivan-frames and six shaking tables for cleaning up the Sullivan deck concentrates, as well as 12 Deister plus three Plat-O tables on which the fine sands, which are not treated on the S-decks, are concentrated. Only —325-mesh fines are treated on the Sullivan decks. The combined hydroseparator overflows are fed to the S-decks. From 1946 to 1949, i.e. with the "old" flowsheet and an average mill feed of 800 tons per day an average of 45 tons per day was treated on four Sullivan rougher frames. The two additional frames were used as cleaners. At times, up to 90 and 120 tons of slimes were treated successfully on this Sullivan deck installation. The average operating data for the S-deck section in Colquiri for the year 1949 were as follows: The tests reported in this paper indicated a tin recovery of 42 pct. The actual plant results for 1949 show a 54 pct tin recovery from the Sullivan decks. The total average quantity of tin produced from the S-deck section in 1949 was 15 tons of tin per month. This corresponds to 3 pct of the total mill recovery. The difference between the 6 pct overall recovery-increase expected after the tests in 1943 and the actual plant results given for 1949 is explained as follows. According to the tests, a total of 170 tons was expected to be routed to the Sullivan decks, but because of certain changes in the flowsheet and mill amplifications,

Jan 1, 1951

By H. Rush Spedden, Arvid Thunaes

R. R. Knobler and F. E. Albertson—Following the testwork done by Thunaes and Spedden, a Sullivan deck plant was built for the Colquiri mill. This plant started to operate in April 1945 and continues in successful operation. Since May 1950 the Colquiri mill has operated with a completely revamped flowsheet at a rate of 1000 tons per day. The old Sullivan deck section was retained with this new flowsheet. The "slime plant" in Colquiri consists of one 12-cell Denver No. 18-Sp. Sub-A flotation machine, one 16-ft and one 12-ft hydroseparator, six five-deck Sullivan-frames and six shaking tables for cleaning up the Sullivan deck concentrates, as well as 12 Deister plus three Plat-O tables on which the fine sands, which are not treated on the S-decks, are concentrated. Only —325-mesh fines are treated on the Sullivan decks. The combined hydroseparator overflows are fed to the S-decks. From 1946 to 1949, i.e. with the "old" flowsheet and an average mill feed of 800 tons per day an average of 45 tons per day was treated on four Sullivan rougher frames. The two additional frames were used as cleaners. At times, up to 90 and 120 tons of slimes were treated successfully on this Sullivan deck installation. The average operating data for the S-deck section in Colquiri for the year 1949 were as follows: The tests reported in this paper indicated a tin recovery of 42 pct. The actual plant results for 1949 show a 54 pct tin recovery from the Sullivan decks. The total average quantity of tin produced from the S-deck section in 1949 was 15 tons of tin per month. This corresponds to 3 pct of the total mill recovery. The difference between the 6 pct overall recovery-increase expected after the tests in 1943 and the actual plant results given for 1949 is explained as follows. According to the tests, a total of 170 tons was expected to be routed to the Sullivan decks, but because of certain changes in the flowsheet and mill amplifications,

Jan 1, 1951

By M. E. Wadsworth, A. S. Peck

Infrared spectra disclose that oleic acid will not adsorb on the surface of pure beryl unless the mineral is first activated with HF. The adsorption of oleic acid on HF activated beryl is attributed to the hydrogen bonding of oleic acid monomers to F-surface bridging sites. These sites are formed as a result of the reaction of HF with chemisorbed water or surface hydroxyl. Characteristic infrared spectra of physically adsorbed oleic acid are correlated with flotation at different pH at temperatures of 25°C and 75°C for colloidal and -60-mesh particles. The usual mode of occurrence of the beryllium mineral beryl, Be3A12Si6O18, is in granite pegmatites as pale green, white, or yellow crystals. A dark green gem variety, emerald, may be found in the wall rock of pegmatite veins or in mica schist. The most important beryl deposits in the United States are in the Black Hills District of South Dakota, and several mountain ranges of central and west-central Utah contain potential commercial sources of this mineral. Several papers have been published concerning the flotation of beryl with different conditioning reagents. Kennedy and O'Meara1 reported on the essential use of HF in the flotation of beryl from different ores. The use of calcium hypochlorite to separate HF activated beryl from feldspar after flotation with an amine collector was made by Runke.2 A report on the activation of beryl with different cations using sodium oleate collector was presented by Viswa-nathan and co-workers.3 Flotation of beryl with sulfonate collector and different activating salts was employed by Fuerstenau and Bhappu.4 The role of the surface charge or zeta potential in the flotation of beryl with different collectors and salts was also extensively investigated.5"7 Previous infrared investigations by the authors8 disclosed that the orthosilicate beryllium mineral, phenacite, reacts with oleic acid to form a chemisorbed oleate monolayer on the mineral surface. Thus, chemisorption is the effective mode of adsorption in the phenacite-oleate acid flotation system. In contrast, neither chemisorption or physical adsorption of oleic acid occur on the surface of pure beryl as evidenced in this study. Efforts were subsequently made to determine the mechanism of activation and collector adsorption for HF pretreated beryl. EXPERIMENTAL PROCEDURES Large, white, terminated crystals of beryl were obtained from Madagascar through Ward's Co., Monterey, Calif. The crystals were broken into -60-mesh fragments in a porcelain mortar. Part of this prepared material was ground to a powder with a Fisher Grinder using an agate mortar and pestle. The powder was transferred to a graduated cylinder and mixed with 500 ml of distilled water. After 5 min, the colloidal suspension was separated from the settled particles by decantation and used as a stock for infrared test determinations. Adsorption tests were performed by first transferring 20 ml of colloidal suspension to 50 ml glass centrifuge tubes. The solids were then centrifuged out and the liquid was discarded. Then, the solids were pretreated with one molar HC1 or one molar HF. After 15 min, the solids were again centrifuged out. The mineral solids were washed free of residual acid by repeated centrifuging and water washing operations. Then, 20 ml of distilled water was added to the solids, and the resultant suspension was used for test purposes. In the preparation of mineral for flotation tests, 1.5 g of -60-mesh beryl was deslimed by removing the colloidal fraction by three successive water washes and decantations. Then, the coarse particles were pretreated with either one molar HCl or one molar HF for 15 min. The mineral was next washed free of acid with distilled water, and 20 ml of distilled water was added to the beryl prior to collector conditioning. The conditioning of mineral with collector was performed in 50 ml centrifuge tubes to which either NaOH or HC1 was added for pH modification. One drop (6.0 mg) of U.S.P. reagent grade oleic acid or 2 ml of 0.01 molar sodium oleate was added to the pulp and stirred at about 1500 rpm for 3 min at a

Jan 1, 1968

By M. E. Wadsworth, L. H. Raby, A. S. Peck

Infrared spec troscopy was used to study the adsorption of oleate collector on three varieties of hematite. Each of the minerals was found to react with either oleic acid or a solution of sodium oleate to form a mineral cation-collector anion chemisorbed layer at the solid-liquid interface. The chemisorption reaction displaced surface hydroxyls and was found to be pH dependent. The infrared band characteristic of the antisymmetrical C = 0 structural group of chemisorbed oleate appears between 1520 and 1540 cm-'. Evidence is presented which shows that molecular filming of the mineral surfaces with un-dissociated oleic acid precedes chemisorption. Infrared absorbance is correlated with the flotation behavior of the mineral. Adsorption processes are usually classified into two types which are distinguished and character- ized by the nature of the bond formed between the adsorbate and adsorbent. The process in which molecules chemically bond to solid surfaces is termed activated adsorption or chemisorption. If only weak van der Waals forces bind the molecules to the surface then the process is termed physical adsorption. Chemisorption is of primary interest in flotation systems since selectivity may be obtained if a single mineral is rendered air-avid by a specific collector mineral adsorption reaction. In previous studies infrared spectroscopy was used to define the mechanisms involved in the adsorption of oleic acid and sodium oleate on fluorite, barite, and calcite. Infrared techniques were particularly useful since adsorbed monolayers were identified in situ. Further, the infrared method permitted identification of reaction products in the solid, liquid, and vapor phases of the systems. The chemisorption reactions displaced F- from fluorite, SO4 from barite, and CO3 from calcite. This investigation was pursued in an effort to characterize and interpret the adsorption processes for synthetic hematite, compact red hematite, and specular hematite (crystalline hematite). Adsorption tests were performed using either oleic acid or

Jan 1, 1967

By M. E. Wadswort, A. S. Peck

Infrared data disclose that phenacite reacts with oleic acid to form a chemisorbed oleate monolayer on the mineral surface. The absorption band characteristic of the antisymmetrical C = 0 oleate structure appears at a spectral wavelength of 1623 cm-'. Chemisorption also results in the displacement of OH- and silicate from the mineral surface. Infrared absorbance at 1623 cm-1 is correlated with float-ability at different pH for -60-mesh phenacite particles. Phenacite, Be2SiO4, is an ore mineral often associated with quartz, microcline, and chrysoberyl in pegmatites and mica schist. Experimental phenacite beneficiating processes have been confined principally to the flotation of low grade ores occur-ing in several mountain ranges of Utah and Nevada. The U.S. Bureau of Mines ' has recently reported on the separation of phenacite from ore deposits of Mount Wheeler, Nev. Infrared techniques were previously applied by the authors to characterize the role of sodium oleate and oleic acid in the flotation of several non-sulfide minerals.23 A parallel study was projected to the flotation of the orthosilicate mineral phenacite with oleic acid. Infrared spectra provided a 'molecular fingerprint' of the mineral surface and permitted identification of the reaction mechanisms. Infrared data were complemented with flotation and chemical tests. EXPERIMENTAL PROCEDURES The experimental procedures used in this investigation, for which data are presented in graphical or spectral form, are described in chronological test order. Phenacite crystals from Mt. Antero, Colo., were obtained from Ward's Co., Monterey, Calif. The crystals were fragmented in a porcelain mortar to -60 mesh and then pretreated with 0.5 molar HC1 for 15 min to remove iron. Then, the mineral particles were washed free of acid with distilled water and dried. Part of the dried material was ground to a fine powder with a Fisher Grinder using an agate mortar and pestle. Next, the powder was placed in a graduated cylinder and mixed with 500 ml of distilled water. The mineral in suspension was decanted from the settled fraction after 5 min, and the colloidal suspension was used for adsorption tests in which a high specific mineral surface area was required for adsorption and infrared analysis. Conditioning was performed in 50 ml glass centrifuge tubes for both infrared and flotation test determinations. Twenty ml of colloidal suspension was used for the infrared adsorption tests. One and a half g of deslimed -60-mesh HCl pretreated phenacite was added to 20 ml of distilled water for flotation conditioning tests. Six mg of oleic acid was used for each infrared adsorption test, and 0.2 mg of oleic acid was used for the flotation conditioning tests. HCl and NaOH were used for pH modification. The pulps were stirred at a speed of about 1500 rpm for 3 min. Then, the solids were separated out and either dried at 50°C or immediately transferred to a float cell. The pH of the residual liquid was determined with a Beckman Zeromatic pH meter. The flotation procedure is described as follows: 1) A 325-1111 glass micro-float cell was filled with distilled water. 2) Conditioned mineral was placed in the cell and the pulp was aerated through the glass fritted bottom at 7 psi pressure. 3) Floated mineral was skimmed from the cell with a glass slide until flotation ceased. (The end point was reached in all tests in less than 2 min.) 4) The flotation products were dried, weighed, and the recovery was calculated. Samples for infrared analysis were made into mulls using either oleic acid or series 13-21 Halocarbon oil obtained from Halocarbon Products Corp., Hacken-sack, N.J. The mulls were placed between two sodium chloride windows. A characteristic phenacite infrared band at 1868 cm-' was used as an internal standard for quantitative determinations in which absorbance at this position was quantified by the baseline-tangent method4 and related to absorbance at 1623 cm-'. The ratio obtained by dividing absorbance at 1623 cm-' by absorbance at 1868 cm-' was defined as an absorbance quotient for use in graphical presentation.

Jan 1, 1968

By O. Vartianen, R. T. Hukki

A preliminary investigation of the collecting effects of various fatty acids in tall oil has led to a conclusion that the collecting power of fatty acids used in flotation increases with increasing un- saturation of the hydrocarbon chain. Linoleic acid was found to be an excellent collector for oxide minerals. TALL oil is a mixture of fatty and resin acids extracted as byproducts in the sulphate pulp industry. In the field of flotation a well-known example of this type of product is Reagent No. 708 of the American Cyanamid Co. Its successful use as a flotation collector with many nonsulphide minerals has been described by Falconer and Crawford.' The composition of crude tall oil varies within wide limits.' " By a number of methods the crude oil may be refined to produce a variety of refined products. This investigation presents a preliminary comparative study of collecting properties of the fatty acids in tall oil. In Finnish tall oils these constituents are primarily palmitic, oleic, linoleic, and linolenic acids.:' The first is a saturated fatty acid having 15 carbon atoms in its hydrocarbon chain, plus one in the carboxyl group. The others have 17-carbon-atom chains and are unsaturated. It is of particular interest to note that palmitic, oleic, lino-leic, and linolenic acids represent increasing degrees of unsaturation with 0, 1, 2, and 3 double bonds, respectively. To test the collecting properties of the four fatty acids separately, the vacuum flotation technique described by Schuhmann and Prakash4 was used. By this method the boundary between flotation and non-flotation conditions was approximately delineated, with the use of ilmenite and several other minerals. Collector concentration and pH were the principal variables. On the basis of information obtained, actual flotation tests were run on an ilmenite-bear-ing ore in a laboratory flotation cell. Through the courtesy of P. Kajanne from the Department of Chemistry, Finland Institute of Technology, the following reagents were obtained for this investigation: 1—Palmitic acid, of Judex, had a purity of 95 pct as determined by parallel iodo-metric and spectrophotometric measurements. 2— Oleic acid, product of May and Baker Ltd., indicated purity of 98 pct. By iodometric and spectrophoto-metric determinations the oleic acid content was found to be 83.8 pct. In addition, it contained 6 pct linoleic, 1 pct linolenic, and 9.2 pct saturated fatty acids.' 3—Linoleic acid was prepared from soybean oil by low-temperature crystallization. Its linoleic acid content was 95 pct by the methods indicated above. 4—Linolenic acid was prepared from linseed oil by low temperature crystalIization. Its linolenic

Jan 1, 1954

By L. W. Pommier, F. B. Brien

The Rheological properties of pulps are non-Newtonian in character and more than one viscosity parameter is necessary to describe their behavior, therefore, the single term 'apparent viscosity' which has been used in the past may be misleading. The study reported in this paper uses a dimensional analysis based on the Ostwald-de Waele shear stress-rate of shear model for non-Newtonian liquids to develop velocity and resistance equations for falling spheres. An electronic monitoring apparatus was assembled for determining the velocity of metallic bodies as they fall through opaque liquids and suspensions. By means of this apparatus the behavior of the viscosity parameters of the Ostwald-de Waele model and the applicability of the proposed equations were examined for a number of non-Newtonian liquids. The validity of the equations for these liquids was confirmed. Preliminary data collected using solid-liquid suspensions established the validity of the form of the basic equations for these systems. The differential movement of solid particles through solid-liquid pulps and suspensions is very important LO many branches of engineering, such as minerals processing, soil mechanics, ceramic slip processing, oil well drilling-mud technology, materials-handling in pipelines, and many others. There is a dearth of fundamental knowledge on the flow resistance encountered by immersed bodies moving through solid-liquid suspensions. The resistance offered by Newtonian liquids to the movement of free settling spheres is well described for laminar and turbulent conditions by Stokes' and Newton's Laws, respectively. However, problems are encountered when attempts are made to extrapolate these simple relationships into solid liquid suspensions. The rheological properties of pulps are such that they are non-Newtonian in character;'" that is, the applied shearing stress and resulting rate-of-shear are related by non-linear functions, and thus, in the simplest case two viscosity parameters are necessary to describe their rheological properties. For this reason the single term 'apparent viscosity' which has been commonly used to explain the flow resistance of suspensions and non-Newtonian liquids may be misleading. It is realized that a rigorous analysis of the rnechanisms of momentum transport of liquid-solid systems is complex indeed. There are many variables that affect the rheological properties of the systems such as density, size, shape, and surface energy of the suspended solid and density and viscosity of suspending liquid, as well as the solid to liquid ratio. However, it was felt that a useful contribution to the knowledge of the behavior of suspensions could be made by examining the systems in terms of the basic fundamentals of nowNewtonian flow. This has led to the development, by means of dimensional analysis, of velocity and resistance equations, using two viscosity parameters to describe the velocity of penetrating bodies and the flow resistance encountered by these bodies, as they pass through non-Newtonian systems. THEORY In order to examine the effects of the rheological properties of solid-liquid suspensions on the movement of immersed bodies it is appropriate to start with non-Newtonian liquids and develop equations, analogous to those of Stokes and Newton, for laminar and turbulent flow respectively, from which quantitative information may be obtained. Further investigation would then be directed at the complex suspension systems, and the effects of the many variables on the velocity and resistance equations. Liquids are classified on a basis of viscosity which may be defined as the relationship between applied shear stress and resulting rate of shear?-'* Pure single phase liquids composed of small light molecules behave as Newtonian liquids, having a single viscosity parameter since the shearing stress and rate-of-shear are related by a linear function. Bingham plastics are similar but have a yield stress. Of particular importance in our investigation are dilatant (shear thickening) and pseudoplastic (shear thinning) non-Newtonian liquids, with and without yield values.

Jan 1, 1968

By C. C. Harris, A. Jowett, S. K. Ghosh

An equation derived from first-order kinetics and accounting for return from froth to pulp has correlated continuous flotation data. The concentration of floatable material in the pulp can be altered either by changing the feed concentration or the feed rate. The latter does not affect the flotation rate constant, the froth to pulp return constant, and the concentration of non-floatable material in both the pulp and concentrate. Both the return constant and the concentration of non-floatable material in the pulp are directly proportional to the feed concentration, while the concentration of non-floatable material in the concentrate is related to the feed concentration. The investigation of the froth flotation process by continuous test methods, previously used by Brown and smith,1 Jowett and Safvi2 and Jowett and West,3 is being continued. Recent tests, primarily concerned with flotation cell design, have yielded kinetic data suitable for testing an equation proposed by Arbiter and Harris4 which postulates that a flotation process is defined by two rate constants: 1) the rate constant for mass transfer from pulp to froth. Previous workers have regarded this as the flotation rate constant. 2) the rate constant for mass transfer from froth to pulp, which has previously been ignored. THEORY Under continuous flow conditions at the steady state in Fig. 1 the equation developed4 is Symbols are defined in Table I. The air-liquid ratio terms4 have been omitted, since they are substantially constant for a given set of conditions. Also, for a system operating at the steady state Vf and V, remain constant and Eq. 1 can be written: Cp/Cc=b'/a,+Qc/a' [2]

Jan 1, 1963

By R. S. Olson

An analysis of the wash thickener-two stage filtration circuit was undertaken to quantitatively describe the circuit performance. On the basis of information developed in this study, the factors of major importance in reducing soluble loss are: filter wash, wash zone size, filter capacity, filter feed pulp dilution, blow back, filter cloth, water to repulp, water to primary wash, water to feed pulp and thickener underflow. The wash thickener-two stage filtration circuit has been used in many hydrometallurgical plants to separate solute from leach solids. As many as 20 to 30 plant operating variables determine the performance of the wash circuit and hence the soluble loss. The complex relationships of these variables, with each other and with soluble loss, make the evaluation of their effect on mill performance difficult. An analysis of this type of circuit was undertaken to quantitatively describe the circuit performance. THE CIRCUIT The diagram in Fig. 1 represents the operation of the wash thickener-two stage filter circuit. This circuit provides for counter current, cross current or for a combination of countercurrent-cross current operation. The feed pulp stream, J, is intended to include any solution or water, other than primary and secondary filtrate, which enters the wash thickener. Except for filter washing, the operations in the circuit are simple mixing operations which can be described by material balance. The concentration of solute in the filtrate and cake solution after application of wash liquor cannot be described by simple material balance. The displacement washing equation, derived by Rhodes1 and extended by Choundhury and Dahlstrom,2 appears to best describe this case of miscible displacement in porous media. In a useful form, it is

Jan 1, 1963

By H. M. Fisher, R. E. Snow, S. C. Sun

SEVERAL constructive and fundamental studies have been made in the analysis of data obtained from experiments carried on with batch ball and rod mills. The operating characteristics of ball milling in small continuous circuits have also been appraised. It is from these analyses that some of the theories of comminution have been developed. Relatively few studies of continuous rod milling have added significantly to the fundamental concepts, because seldom have they yielded sufficiently consistent results. Perhaps they have been too limited in their scope. Careful control of the variables in batch grinding is simple compared with that encountered in a continuous operation. This factor alone has discouraged many investigators. Occasionally results of systematic changes made in industrial rod mill circuits have been published, but usually the data are sketchy and are restricted because of the unwieldiness of the equipment used. The work, in general, has not been comprehensive; nevertheless it has provided empirical relationships that have bridged the gap between postulate and practice so that by proper manipulation of formulae, a mill designer can anticipate mill size and power requirements.14 Although operating variables of a small continuous mill are not so easy to control as with the batch mill, with present day devices, and with careful experimental work, consistent results can be obtained. Nearly four years ago, in the Process Laboratory, Allis-Chalmers Mfg. Co. began a systematic study of the effects of several variables upon the performance of the pilot rod mill. A mill was built in the laboratory to provide the versatility required for the proposed study. It was constructed in sections so that it could be operated, with a few modifications, as a rod mill 30 in. x 8 ft or 30 in. x 4 ft. The discharge end of the shell was flanged so that either an end peripheral discharge or an overflow discharge could be installed. Thus the performance of at least four types of mills could be studied merely by changing the type of discharge or the length of the mill shell. The grinding experiments were designed so that a study could be made of the way in which the mill speed, feed rate, and pulp density influenced the performance of both overflow and end peripheral discharge rod mills. Four sets of experimental data were collected from the four mill arrangements. The mill in each set of experiments was fed at four rates of feed depending on the length of the mill, at four pulp densities, and at five percentages of critical speed. Electrical and mechanical controls were in- stalled to regulate all these independent variables, and auxiliary devices were used to verify the precision of the controls at each point. The dependent variables used to quantify the experiments were the reduction ratio and the hew surface area produced as calculated from sieve analyses. These were incorporated with the energy factor by the calculation of both the new surface produced per unit of energy and the Bond work index.' Rod wear, as a dependent variable, was not studied because of the short period of operation for each run. Exclusive of repeat runs, each set of experiments yielded 80 products, and the total study at least 320 products, all of which were quantified. With the operating information collected, these data presented a bewildering accumulation. Statistical analysis has been invaluable in unraveling the confusion and in presenting a means of establishing the nature and the magnitude of the significant variables. Data presented in this paper are those from the 30 in. x 4 ft end peripheral discharge rod mill, Fig. 1, when limestone was ground at feed rates of 1000, 2000, 4000, and 5000 lb per hr, at pulp densities of 50, 60, 70, and 80 pct solids, and at mill speeds of 50, 60, 70, 80, and 90 pct of the critical speed. These 80 tests have all been run at least twice, and occasionally a third time, to prove that the data obtained were reproducible. The techniques of operation and the methods of quantification of results are described in the following pages and the results analyzed statistically to show the significant variables. The variables are plotted to show the relationships that exist. A massive dolomitic limestone from Waukesha Lime and Stone Co. was used for feed during these experiments because of its availability and its tex-tural uniformity. This limestone analyzed 28.7 pct CaO, 21.0 pct MgO, 6.0 pct SiO2, 0.4 pct A1²O³, and 0.3 pct Fe²O³ and had a loss on ignition of 44.1 pct. It had a rod mill grindability at 14 mesh of 9.6 grams per revolution from which a work index of 13.9 was calculated. The ball mill grindability at

Jan 1, 1955

By B. H. Bergstrom, Will Mitchell, T. G. Kirkland, C. L. Sollenberger

IN a previous article' the authors outlined a study of the variables in rod milling and also reported data from a series of open circuit grinding tests on a massive limestone in a 30-in. x 4-ft end peripheral discharge rod mill. As a second part of the experimental program, an analysis is now presented for the 30-in. x 4-ft overflow rod mill grinding under identical conditions, except that discharge ports on the periphery of the mill shell have been sealed so that the products from the present series overflowed through a 9-in. diam .opening in the center of the end plate. A variance analysis has been made of the combined data for the two experiments, and performances of the two mills are compared here. Included in the first report' were descriptions of feed preparation, rod mill circuit, instrumentation and controls, and techniques used to evaluate data. Dependent and independent variables were defined, and variance analyses were made to test the relative significance of variables and to establish magnitude of error for the experiment. Significant data were plotted in various combinations, and conclusions were drawn from the graphs. The procedure and analysis in this series of tests follows the first tests and is not repeated. Data from the second series are recorded in Table I. Listed in the first three columns are the independent variables of feed rate (1000, 2000, 3000, 4000, and 5000 1b per hr), mill speed (50, 60, 70, 80, and 90 pct of critical), and pulp density (50, 60, 70, and 80 pct). The dependent variables, Pso, P100, reduction ratio, slope of the log-log sieve analysis curve, power demand, and Bond work index follow. Of these, only the reduction ratio and the Bond work index were analyzed for significance. Production of new surface as calculated from sieve analyses has not been included for this series because of the questionable assumptions that have to be made to satisfy the formulas involved. The large number of products obtained during the runs precluded the use of surface measurement techniques by the gas adsorption methods at this time; however, samples of all products have been stored for future reference. To test the consistency of the reporting of the sieved products, an averaged sieve analysis was calculated from the wet-dry plots obtained from the three product samples of each run. The resulting averaged analysis was plotted and the P80, selected. The relative deviations of the P80's from each of the three product samples with respect to the P80 of the averaged analysis were then calculated. In only two sets were the relative deviations (6.2 and 9.9 pct) considered excessive. In each of these two sets, one sieve analysis was obviously out of line; hence that analysis was ignored and new averages were computed. This reduced the relative deviations to 1.2 and 2.7 pct respectively. The relative deviations of the product analyses with respect to their averages ranged from 0.1 to 1.4 pct at 1000 lb per hr, 0.0 to 1.1 pct at 2000 lb per hr, 0.2 to 3.0 pct at 3000 lb per hr, 0.3 to 4.3 pct at 4000 lb per hr, and 0.5 to 5.2 pct at 5000 lb per hr. The relative deviation of the 80 pct passing point for 96 dry sieve analyses of the feed with respect to that of the averaged analysis was 7.6 pct. This slightly higher percentage can probably be attributed to a greater proportion of tramp oversize in a crusher product than is ordinarily found in a rod mill product. The last column on Table I lists the adjusted work index, which has been used as the measure of efficiency for the various combinations of operating conditions investigated. Efficiency increases as the index becomes lower. It was reported in the previous paper that the work indexes for the Waukesha limestone used in these experiments decreased as the product size decreased (as calculated from Bond grindabilities). That is, this limestone becomes easier to grind as the material becomes finer. This is unusual, because the work index for most materials as calculated from the Bond grindability has remained constant as the product size decreased or has increased slightly. Table II lists the results of Bond grindability tests at all mesh sizes from 3 to 200 and the work indexes calculated from them. To remove this variation of work index with product size from the data so that results would apply to any material of constant work index, the work index values shown in Table II were plotted against product size on log-log paper. From this curve (a straight line function in this case), the expected work index for the product size for each of the runs of the experiment was obtained. The work indexes as calculated from the reduction ratio and energy consumption were then divided by the corresponding expected work index. The results obtained are reported in percentages on Table I as adjusted work index and are actually percentages of the work index for the Waukesha limestone at the size in question. Multiplication of the work index value for a material of constant index by these percentages should allow the application of the adjusted work index curves to the material. Only the adjusted work index values, not the actual experimental values, were used for the variance analyses and for the graphs.

Jan 1, 1956

By G. H. Wilson

INTRODUCED in 1946 to serve a need in power-plant operation, closed-circuit TV has been used by well over 200 organizations in approximately 25 different industries. Known as industrial television, or simply ITV, it can be described as a private system wherein the television signal is restricted in distribution, usually by confinement within coaxial cable that directly connects the TV camera to one or several monitors, Figs. 1, 2. The picture is continuous and transmission is instantaneous, permitting an observer to see an operation that may be too distant, too inaccessible, or too dangerous to be viewed directly. Destructive testing or the machining of high explosives can now be conducted hundreds of feet away by personnel who still have close control through the eyes of the TV camera. It is also possible for one man to control operations formerly requiring the co-ordinated efforts of several workers. For example, at a large midwestern cement plant conveyance of limestone from primary crusher to raw mill and loading into five storage bins once necessitated the work of two men, one having little to do but prevent spilling of material by manually moving the tripper on the belt conveyor as occasion required. TV cameras mounted on the tripper now provide bin level indication to the conveyor operator at the crusher position so he is able to control the entire loading operation remotely, Fig. 3. By means of a switch, the picture from either camera is alternately available on a single viewer, or monitor, Fig. 4. Each camera is mounted on the tripper by means of a simple adjustable support and looks down into the bin, which is identified by the number of cross members on the vertical rod. Each associated power unit is located on a platform above the camera, Fig. 5. This centralized control by means of TV often has produced superior results, and in many instances saving in operating costs has been sufficient to write off equipment costs within six months to a year. Where a key portion of a process may be enclosed or otherwise inaccessible, TV again reduces the likelihood of mistakes and permits closer control by making available to the operator valuable information he might otherwise never possess. An example of this can be found at a strip mine where the coal seam lies 50 ft or more below the overburden, which is removed by a large wheel shovel. From his centrally located position the shove1 operator was unable to judge accurately to what extent the wheel buckets engaged the earth. His chief indication of efficiency was the amount of overburden on the belt conveyor as it passed his control point 75 ft from the wheel. Now, two television cameras mounted on the tip of the boom permit the operator to view the wheel from each side and provide him with a close-up view of the buckets so that he can take immediate and continuous advantage of their capacity, quickly compensating for ground irregularities and avoiding obstructions, Fig. 6. While the word television conjures up visions of highly complex and intricate apparatus such as that employed in modern TV studios and transmitting stations, the term industrial television should indicate compact, straightforward equipment. Most present-day ITV systems contain fewer than 25 tubes including camera and picture tubes. The average home television receiver alone requires at least that many tubes. Equipment like that illustrated in Fig. 1 contains only 17 tubes, of which 3 are in the camera. It can operate continuously and dependably, without protection, in any temperature from 0" to 150°F. It consumes less current than a toaster and weighs under 140 lb. Camera and monitor may be separated by 1500 to 2000 ft and by greater distance with additional amplification. This equipment is designed to withstand vibrations up to 21/16 in. and will operate successfully under more severe conditions of vibration and heat when suitable enclosures are provided. Any number of cameras may be switched to a single monitor, and any number of monitors, within reason, used simultaneously. Two types of applications in the mining industry have already been described. A third under serious consideration by several organizations will make use of ITV for remote observation of conveyor transfer points at copper concentrating plants so that evidence of belt breakdown and plugging of transfer chutes can be spotted immediately and costly overflow of material avoided. A television camera will soon be installed to view a trough conveyor near the exit of an iron-ore crusher to indicate clogging of the crusher as evidenced by reduction or absence of material on the

Jan 1, 1955

By C. A. Rowland, R. C. Nealey

Until recently there were very few dry grinding rod mills used. The early installations performed as anticipated. In the last seven years a number of dry grinding rod mills have been made, most of which have been used for grinding iron ore. The recent installations have not operated as anticipated in accordance with practices learned from wet rod milling. The paper does not establish any definite rules for the application of dry grinding rod mills but does attempt to establish some guidelines that can be followed. The paper reviews a number of flowsheets describing the problems encountered and some of the solutions found. Completely dried compared to partially dried feed is reviewed. Problems encountered in scaling up from pilot plant to production-size mills are described. Rod sizing, effect of rod length, rod bending, rod breakage, and rod tangling are discussed. While wet grinding rod mills have been used for 50 years to grind ores and make specifications sands, dry grinding rod mills have only recently come into significant use in the mineral industries. Among the materials now being processed in dry grinding rod mills are (1) lead sinter, lead slag, and limestone; (2) iron ore; (3) silica; (4) cement clinker; and (5) coke. Because the application of wet grinding rod mills was well understood at the time that dry grinding rod mills were introduced, the criteria and guidelines for selecting the former were carried over to the selection of the latter - in some cases with the anticipated results, but in many cases not. It is the purpose of this paper to describe, through a series of flowsheets, some of the recent problems encountered with dry grinding rod mills, to point out the solutions to these problems, and to set forth some guidelines for the selection and application of these mills. When dry grinding rod mills came into use, the following facts about their wet grinding counterparts were already well established: (1) The performance of large-sized mills can be accurately predicted from the performance of rod mills 5 or 6 ft in diameter. Wet grinding mills as large as 14 ft in diameter (inside shell) by 20 ft long with rods 19.5 ft long can operate satisfactorily.

Jan 1, 1970

By J. E. Lawver, G. L. Samsel

The success of many mineral dressing processes, including froth flotation and electrical concentration, depends largely on knowledge of physical and chemical structures of the mineral surfaces involved. As a general rule solids have surface layers, usually invisible by optical tests, that differ from the substratum. To the mineral engineer, therefore, a knowledge of the surface structures of minerals is of paramount importance. During the past decade one of the most powerful tools in the study of froth flotation mechanisms has been the use of radioactive tracers, but this technique usually cannot give definite information regarding the chemical changes that may or may not have occurred at the surface. The electron microscope and electron diffraction camera can be particularly useful instruments in surface studies.. An election microscope permits much closer examination of surfaces than can be obtained with a petrographic microscope. In certain cases, it is possible to identify trace quantities formed by chemical reaction between flotation agents and mineral surfaces by the use of a single-crystal, reflection electron diffraction camera. This article describes several instances in which these instruments have been valuable guides in mineral dressing research at International Minerals & Chemical Corp. One of the first applications of these techniques employed by International Minerals was in conjunction with the study of electrical concentration of potash ore. The first reasonably successful separation of potash ore was made by subjecting a sample of the ore to a methanol wash. Subsequent tests using tagged 6C14 methanol showed that only a minute quantity of radioactivity was detectable after the methanol-washed sample had been dried and heated to about 300°F. It was concluded that although some methanol had been held to the surface by adsorption, this could not account for the improved electrification because there was no evidence of methanol selectivity with respect to NaCl or KCl. Consequently there must have been some other change in the surface. Carbon replicas* of the mineral grains were made before and after the methanol wash. The replicas were separated from the minerals by dissolving in water. This extraction replica technique is particularly attractive because the specimen can first be used with the electron microscope to observe the general structure of the surface and can be re-used as a specimen for electron-diffraction studies to identify any solvent insoluble material present on the mineral surface." Fig. 1 shows the general

Jan 1, 1960

By E. C. Tveter, L. A. Roe

The authors present a general outline of the theory and development of heavy-liquid application to mineral processing. Patent literature and processes are reviewed with special emphasis on liquid recovery systems which have been employed or proposed. Advantages and disadvantages of the process are discussed together with the recent developments which have revived interest in this old concentration method. The most important single factor in this resurgence of interest is concluded to be the narrowing gulf between chemical engineering and mineral dressing which has opened the field for new concepts of mineral plant design. A partial summary of patents on heavy liquid separations is included. In spite of the fact that heavy-liquid separation with organic liquids has been used in the laboratory for over 50 years, this process has never graduated to large scale commercial use for any extended period. However, a variation of this process, the sink-float or heavy-media separation process has found wide acceptance and is used to separate minerals from diamonds to gravel. Materials used to increase the specific gravity of the pulps used as heavy media include sand, clay, barite, magnetite, galena, hematite, atomized lead and ferrosilicon. Because of the greater ease of recovery, the magnetic materials, magnetite and ferrosilicon, are the preferred media today. In the U. S. alone, heavy-media iron ore plants with a capacity of over 10 million tons of concentrate per year are in existence. Heavy-media separation involving use of solid, inorganic particles suspended in water rapidly found a wide range of commercial use with the introduction of magnetic media. The minerals engineer is experienced in working with suspensions and quickly learned to develop and control such media at a cost compatible to the type of separation desired. The natural superiority of a heavy liquid with uniform chemical and physical properties has never been questioned since its first use in laboratory mineral separatory procedures. It offers the only method for gravity separation of fine particles of relatively close specific gravity. The most important early attempt to make heavy liquid separations commercial were made by the DuPont Co. which began experimental work on Virginia limonite in 1904. The appended "Partial Summary of Patents" compiled by W.L. O'Connell of The Dow Chemical Co. demonstrates the quantity and sequence of this work. Both inorganic and organic parting liquids were investigated and numerous patents were issued on the use and recovery of these liquids. The work culminated in the heavy liquid plant of the Weston Coal Co. at Shenandoah, Pa. Chemical engineering design of this plant was the responsibility of Francis I. and Hubert I. DuPont. Heavy-liquid separation of coal almost achieved commercial status at this plant which actually processed over 20,000 tons of coal. The liquids used were tetrabromethane, pentachloroethane and trichlo-roethylene, with liquid losses ranging from 8.9 to 12.4 oz per ton of cleaned coal. The reasons for failure of this plant are not clear but probably involved toxicity problems as well as other problems in chemical engineering. Excellent economics were reported. A more recent pilot plant was built in 1954-55 by Norris Goodwin for the Inerto Co. to treat hectorite clay. This plant employed carbon tetrachloride in jigs with liquid recovery by evaporation. Although good separation was achieved, incomplete removal of the CCl, from the clay prevented commercial operation. The only present operations known to the authors employing heavy liquids for gravity separations are limited to the use of calcium chloride in certain coal washers and bromochloromethane in a small batch operation (one ton solids per day) for the separation of beryllium metal particles from slag materials. This is not strictly a minerals beneficiation problem but it has demonstrated the feasibility of such separations. PROBLEMS Critics of early attempts to commercialize heavy-liquid separation of minerals summarize the draw-

Jan 1, 1963

By H. H. Dreissen, F. J. Fontein

The operating principle of the classifying cyclone has been described often but some special aspects of the cyclone receive further clarification here. In addition, several uses of the classifiers are summarized and made clear by means of practical examples. In coal and ore washeries, the washing cyclone is widely used for the separation by specific gravity of sizes up to about 12 mm. This application has lately been extended to the separation by specific gravity of particles up to about 50 mm. This use makes special demands on the construction and wear resistance of the cyclone. The sieve bend, a development of the Dutch State Mines which is less well known than the hydrocy-clone, is dealt with at some length. The sieve bend is an apparatus designed for wet screening. Attention is given to the variables influencing the performance of the sieve bend. Hydrocyclones are now so widely applied that it would be difficult to find an industry not, directly or indirectly, using them or knowing how to use them since publications on this subject abound.1-3,5,9-11,13-15 However, because of the large number of variables influencing the operation of classifying cyclones, it is not entirely possible to calculate the results of this type of cyclone. Therefore, a number of lesser known variables will be discussed here along with some special applications of the classifying cyclone. FACTORS INFLUENCING CYCLONE CLASSIFICATION Shape of Particles: The shape of particles may cause a much coarser classification than the one expected, considering the sieve analysis and the cyclone dimensions chosen. This was clearly shown by feeding a mixture of spheres and disks in water into a cyclone. The spheres were 0.36 mm diam and the disks, 3 mm diam and 0.03 mm thick. These dimensions were chosen in order that the disks and spheres might have the same rates of fall under gravity. Both spheres and disks had a specific gravity of 1.05. The results obtained with these particles in a cyclone are summarized in Table I. Test No. 2 shows that the catch of spheres was

Jan 1, 1963

By D. W. Frommer

Processing nonmagnetic taconites by selective flocculation-desliming and flotation requires large volumes of water. If impounded without treatment, these off-process waters require excessively large areas for containment. To discharge the waste water into natural waterways would contribute to stream pollution and likely would not be permitted. In U.S. Bureau of Mines experiments conducted in the Twin Cities Metallurgy Research Center's 900-lb per hr pilot plant, approximately 85% of water requirements for the flotation-based treatment of a Michigan nonmagnetic taconite were met by reclaimed water. Water reclamation of the off-process streams from flotation was accomplished by controlled additions of lime, sodium carbonate, and a synthetic flocculant to reduce turbidities to 51000 ppm equivalent SiO*, while maintaining a Ca(II) content of =16 ppm in the finished effluent. Flotation concentrates of good quality were obtained using the reclaimed water. The cost of chemicals used in water reclamation was approximately equal to the savings in flotation reagents attributed to recycling of the water. Water quality is perhaps as important to flotation as are the reagents used. The character of water is extremely variable, depending on whether the source is a well, lake, or stream, upon the season and temperature, upon prior use, and upon the character of the watershed. All of these factors influence the water hardness and the quantity of other dissolved inorganic salts, turbidity, dissolved and suspended organic matter, dissolved gases, and pH. Frequently, the differences in water quality can measurably influence flotation selectivity, often to the point of spelling success or failure. Water hardness is particularly troublesome in flotation systems employing fatty acids, but other unrecognized constituents may also contribute to peculiarities in flotation behavior. Furthermore, a given water source may be entirely satisfactory in one flotation system, but entirely inappropriate in another. In recent years, society has given more attention than formerly to water use, even in areas where water is plentiful. However, both the demands of a growing population, with increased per capita needs, and also those of industry must be met. As a result, riparian rights must be negotiated with the appropriate government agency or agencies so that consumption of water is often allocated or otherwise controlled. Furthermore, the disposal of off-process industrial and domestic water is coming under the increasing scrutiny of these same governmental units. In these respects, the mineral industry is no exception, so that conservation, water reclamation, and reuse may be expected to assume increasing importance. In 1932, it was stated that, "water reclamation is generally more expensive than the economy in water and reagents resulting from its use. It is employed only if it is urgent to save water."' The economic aspects of this statement may still be true, but the unrestricted use of water is becoming less and less an option of the user. The purpose of this paper is to discuss an investigation conducted by the U.S. Bureau of Mines (USBM), in which water reclamation and treatment were undertaken to develop procedures for, and to assess the effects of, water reuse on the flotation treatment of nonmagnetic taconites. This investigation is a logical extension of previous work described by the author and associates at the Twin Cities Metallurgy Research Center employing selective flocculation-desliming and anionic flotation of silica from low-grade, nonmagnetic iron ores2,8 The requirements for the selective flocculation-de-sliming and the anionic flotation of silica processes determined the direction of the investigation. Previous studies had indicated that pH levels of about 11.0 and 11.8 were required for selective flocculation and flota-tlon, respectively. The calcium content of the water was believed to be of importance in both of these operations, and since Minneapolis tap water with a Ca(I1) content of about 16 ppm had been successfully used, an attempt was made to reclaim the process water at an equal level of dissolved calcium. Additionally, the objective was to nullify or effectively limit the effects of dispersants and fatty acid residuals from prior stages of processing. Last but not least, the system of water treatment had to have the capability of reducing turbidities to workable levels. With about half of the effluent being derived from the selective flocculation-desliming step and carrying about 25,000 ppm of highly dispersed, suspended fines, this last objective appeared formidable at the outset. Procedures for water reclamation were derived, in part, from well-known mineral dressing practices, from past observations and investigations, and from concepts contained in various pertinent publications on water treatment."c These procedures involved: 1) flocculation with lime and poly electrolyte-type flocculants, 2) lime-soda-ash softening, 3) chemical precipitation, and 4) mineral surface adsorption. Control was exercised at various stages of water reclamation by frequent measurements of Ca(11), pH, and turbidity.

Jan 1, 1971

By A. D. Rovig, D. W. McGlashan, Donald M. Podobnik

Crystal-chemical and stntctural properties of sulfide minerals are considered. The information gained is to be used to interpret (I ) freshly broken mineral surfaces, (2) modifications of the mineral surfaces, and (3) reactions at the mineral surfaces. From these basic disciplines, concepts with regard to the changes that surfaces undergo, reactions that might take place, the geometry of the interface, the state of different atoms and ions in the interface, and other physical and chemical properties of an interface must be developed, weighed and applied. The authors first deal with these conceptual considerations from which hypotheses are set forth to describe the environ men tal-interfacial relationships for several sulfide minerals. Qualitative and quantitative explanation of par-ticulate solid separations are unknown entities because of the lack of adequate models and mathematical relationships to explain the activity occurring between the solid and liquid phases. It is a transitional region in which it is difficult to ascertain the mechanisms of adsorption of ions, molecules, etc., on mineral surfaces, as well as other secondary reactions which occur in interfacial regions. Thus, in this paper the authors deal with conceptual considerations from which hypotheses are set forth to describe the environmental-interfacial relationships for sulfide minerals. GENERAL CONSIDERATIONS Interfacial Reactions:Interfacial reactions are the cruxes of flotation schemes as well as other processes such as thickening, filtration and hydrometallurgy. However, as noted by Klassen and Mokrousov: 1 "The problems concerning the surfaces of natural minerals, the laws governing simultaneous adsorption from aqueous solution on these surfaces of a whole series of reagents, the laws of the surface reactions and the properties of water layers separating the minerals, are all known to a first approximation only." An investigator has the privilege of postulating methods of solid-liquid interfacial reactions. REACTIONS WITH WATER - Water consists of hydronium (H3 O 4) and hydroxyl (OH-) ions in the ionized state. That this is so forms the basis for the postulated reaction of the hydrated hydrogen ion with net-negative mineral surfaces as depicted in Fig. 1. In this case water is bonded to mineral surfaces through hydrogen-bridge-type bonds - possibly hydrogen bonds. Although bonding of the van der Waal type may be responsible for this reaction, it is most likely that stronger bonds are involved. Once firmly bonded to the surface, the water layer is known to be quite tenacious. It is further speculated that a shear plane exists at some distance (see Fig. 1) away from the mineral surface. Exact location of this plane is not known, but in all probability it will be positioned across a weak bond of the hydrogen-bond type where the surface-attached water is coordinated to the original hydrated hydrogen ion. REACTION WITH METAL IONS - Klassen and Mokrousov ' state: "The presence of an ion in water leads to an immediate formation around that ion of a highly condensed atmosphere of water dipoles and thus to hydration of the ion." The fact is known that multivalent cations are strongly hydrated, usually by six or eight molecules of water, and that anions are not so strongly hydrated. Conceding the fact that thermodynamics, concentration, pH, and physical considerations are of utmost importance in truly explaining a mechanism of metal ions reaction with a hydrated mineral surface, it does seem logical that the mechanism illustrated in Fig. 2 is feasible. In this reaction, the metal ion (M) is coordinated in one dimension - to the mineral-surface hydration layer. Note also that explanation of this reaction requires that the shear plane move to a less stable bond configuration; that is, the shear plane has moved to a position between the metal ion and the other coordinated water molecules which are more free to dissociate. Reactions as depicted in Fig. 2 should cause the formation of an apparent mineral surface which is net-positive. Immediately, it must be noted that measurements of apparent-mineral surfaces serve - within limits - to indicate a degree of ion-surface reaction capability. The major limiting factor he re is one related

Jan 1, 1970

By T. Szaplonczay, D. E. Light

The auto-oxidation of sulfur dioxide to form sulfuric acid in commercially produced barren and recycle liquors, containing 1.66 and 4.56 gpl Fe respectively, from Elliot Lake was studied. A gas mixture, containing 3.5% SO2 and air was introduced into these liquors through a frit at a rate of 6.0g SO2 per hour per liter of solution. The barren and recycle were heated to 60°C. At the end of the first 24 hours of the process the concentrations of acid produced were 158 and 220 gpl H2SO4 for the barren and recycle respectively. Corresponding conversion efficiencies were 70 and 83%. In the period from 24 to 36 hours the rate of acid production decreased so that the H2SO4 concentrations at the end of 36 hours were 180 gpl for the barren and 260 gpl for the recycle; average conversions for the initial 36 hours were 55 and 68% respectively. Many investigators1'2'3 have considered the oxidation of sulfur dioxide to form sulfuric acid in aqueous solutions containing metallic ions. A summary of the work performed on this subject prior to 1946 has been presented by Keyes4 who described the auto-oxidation process as follows. Keyes also suggested that in this second reaction ferric ion serves as a catalyst for the conversion of sulfur dioxide to sulfuric acid. Neytzell-de Wilde and Taverner5 have reported on the use of auto-oxidation to produce a 4 to 5% sulfuric acid solution for use in a cyclic leach process at South African uranium plants. This paper describes investigations carried out to determine the feasibility of using the auto-oxidation process to produce sulfuric acid solutions of as high concentration as possible using commercially produced uranium plant barren and recycle liquors from the Elliot Lake region of Ontario. These liquors contained 1.66 and 4.56 gpl respectively of iron, chiefly in the ferrous state. This investigation is of particular interest to the uranium producers in Elliot Lake where the ores contain 3 to 4% sulfur, as sulfides. Flotation of these sulfides and roasting of same to produce SO, bearing gases have been successfully demonstrated using conventional methods.6'7 Barren and recycle solutions, as described previously, are also produced in large tonnages by these uranium producers. The experimental method and apparatus used to examine the auto-oxidation process will be described first. Then the tests carried out on laboratory prepared aqueous solutions containing ferric sulfate to determine optimum conditions will be presented. The next section will describe the tests using the commercial barren and recycle liquors. A final section will summarize the conclusions of the work. EXPERIMENTAL APPARATUS AND PROCEDURES Two types of solutions were used in the tests, a laboratory prepared aqueous solution containing definite amounts of ferric sulfate, and commercially produced iron bearing solution obtained from a uranium leaching plant in the Elliot Lake area. Two liters of a solution to be examined were placed in a 3 in. diam by 30 in. long glass cylinder with a cross-sectional area of 7 sq in. A mixture of air, and SO2 supplied from a lecture bottle for convenience, was introduced into the bottom of the vessel through a standard 30 mm glass frit, 700 sq mm in area. The temperature of the apparatus was controlled by a heating tape and variable transformer. Exhaust gases were vented to atmosphere. At the end of the test period the gas flow was shut off and the solution was analysed for H2SO4, Fe ++, and Fe +++, by conventional means. Four variables were examined using the laboratory prepared aqueous solution of ferric sulfate. First, the effect of SO2 concentration, from 1 to 1076, in the feed gas mixture was considered. Second, the temper-

Jan 1, 1964