Interpretation of the recovery/time curve and scale-up from column leach tests on a mixed oxide/sulfide copper ore

Transcription

1 Interpretation of the recovery/time curve and scale-up from column leach tests on a mixed oxide/sulfide copper ore Ronald J. Roman Leach, Inc N. Placita del Sol Tucson, AZ Jose Hector Figueroa P. and Jorge Enrique Ruiz H. Mexicana de Cananea S.A. de C.V. Av. Juarez S/N Cananea, Sonora Mexico Jorge Helleon G. Mexicana de Cobre, S.A. de C.V. Aptdo 20 Nacozari, Sonora Mexico Efrén Pérez S. University of Sonora Dept. of Geology Hermosillo, Sonora Mexico ABSTRACT The shrinking core model for coarse particle leaching has been generally accepted as describing the leaching of a copper oxide or sulfide ore. However, when a mixed oxide/sulfide ore is leached this model can not be used in its simple form because at least two and possibly three separate leaching processes are occurring simultaneously (dissolution of oxide copper minerals, secondary copper minerals and primarily copper minerals). It has been impossible to isolate their individual leaching curves from the recovery/time curve generated by the column leach test. This paper describes a tests program carried out at the Groupo Mexico, Mexicana de Cobre s La Caridad operation in which the individual recovery/time curves for the leaching of copper oxide mineral, secondary copper mineral and primary copper minerals were developed from standard column leach tests. Once the individual recovery/time curves were developed scale-up of the column leach test results to the commercial heap leaching operation is possible by using the shrinking core model.

2 INTRODUCTION A fifteen-year mine plan is being prepared by the staff at La Caridad. This mine plan will recognize and incorporate the leaching response of the individual ore blocks so the mining sequence can be based on the overall economics of the operation including the recovery of copper from the leach ore as well as the recovery of copper from the milled ore. This paper briefly describes part of the column leach test program that was undertaken at La Caridad and by the metallurgical staff at La Caridad with the assistance of Leach, Inc. The objective of this column leach program was to develop a correlation between the leachability of the ores from the La Caridad pit and the geologic and/or chemical characteristics of the individual ore blocks. The leachability of an ore is defined by the recovery/time curve generated by leaching the ore under plant conditions. The mine plan is made up of thousands of blocks that are contained within the pit limits. In order to determine if an individual block is to be considered waste, flotation feed or heap leach feed the amount of copper which might be recovered from that block needs to be estimated. Normally, to obtain that estimate for the option of heap leaching a column leach test is required. With the column data in hand, the recovery/time curve for the commercial heap leaching operation is projected based on the shrinking core model (1,2). However, conducting thousands of column leach tests, one for each ore block, would be prohibitive from both the cost and the time required. An addition problem is that shrinking core models are based on the rate of movement of the interface between the leach shell and the unleached core of the ore particle. When the ore contains copper in more than one form (oxide, secondary sulfides and/or primary sulfides) then there exists more than one interface. It is therefore necessary to experimentally measure the rate of copper recovery for each of the forms of copper. No experimental technique is available to divide the experimental recovery/time curve from a column leach test into the individual recovery/time curves for the different copper components of an ore. The division of the overall recovery/time curve into its individual components is necessary if the shrinking core model is to be used to project the column leach test results to the commercial heal leach operation. It was a primary objective of this project to develop a technique to divide the experimentally determined recovery/time curve into individual recovery/time curves for each form of copper present in the ore and secondly to find an easily measured characteristic of an ore sample which correlates with the recovery/time curve under the commercial heap leach conditions. Because the copper produced from leaching the ore is recovered over several years, not only does the final recovery needs to be estimated but the complete recovery/time curve needs to be predicted so that the cash flow from the annual copper production from that ore block can be properly discounted when assigning a value to the ore block. 1

3 PROCEDURE The column leach test program was divided into four parts: 1. Determination of the reproducibility of the column leach test through duplicate tests, which were run at the La Caridad lab; and duplicate tests, one run at La Caridad and the other one run in the lab at Mountain States R & D International, Inc. (MSRDI) in Vail, Arizona. 2. Determination of the change in the leachability of an ore resulting in changes in the operating parameters of the column leach test i.e. leach solution irrigation rate, column height, column diameter, leach solution chemistry, leach/rest cycle schedule, etc. This phase of the column test program was used to establish the procedure for the standard column leach test and to demonstrate that the leaching of the ore followed the shrinking core model of leaching. 3. Determination of the leachability of ore samples from throughout the La Caridad deposit and analysis of the leachability in an attempt to correlate the leachability of an ore sample with geological and/or chemical parameters of the ore sample. 4. Projection of the column leach test date to leachabilities for that ore sample when leached on the commercial heaps. This paper reports on the third part of this project: developing a correlation between the leaching recovery/time curve and the geologic and/or chemical characteristics of an ore sample. The La Caridad deposit has been divided into four zones based on both the geological and mineralogical characteristics of the rock. Ore samples for column leaching were collected from each of the four zones. In addition to this classification of the ore blocks, each ore block can be classified based on it copper grade. Two classes based on copper grade were selected: Cu(total) equal or greater than 0.15 percent and equal or less than 0.30 percent, and Cu(total) greater than 0.30 percent. The most prevalent copper minerals in the La Caridad deposit are chalcocite, covelite and chalcopyrite. In addition some oxide copper minerals as well as bornite are present. Because the responses of these minerals to leaching differ greatly, a third classification was established: the percentage of leachable copper. The leachable copper is defined as that copper contained in minerals solubilized by a five percent sulfuric acid solution or a ten percent sodium cyanide solution (3). The fraction of the contained copper which is either acid soluble or cyanide soluble is called the Solubility Index or 2

4 S.I. The other copper minerals, which by this definition are not leachable, are chalcopyrite and refractory mixed iron or manganese copper oxide minerals. Each ore block was then classified by its leachable copper content: less than 25 percent, equal or greater than 25 percent but less than 50 percent, equal or greater than 50 percent but less than 75 percent and equal or greater than 75 percent. In those column tests in which different ore samples were being evaluated for their leachability a standardized test procedure was used. The test procedure was designed to allow both the rate of copper solubilization and the ultimate copper recovery to be determined in as simple a test as possible. The standard test procedure was selected after running a preliminary group of 13 column leach tests in which the operating parameters were varied. Parameters varied included column height, column diameter, irrigation rate, cure procedure, rest cycle schedule and leach solution chemistry. In summary, the standardized test leaches a 90 kg ore sample crushed to -38 mm. for 90 days in a 152 mm diameter column 3.0 meters in height. A sample for a size/assay test is split from the ore sample and each size fraction is analyzed for acid soluble copper, cyanide soluble copper and total copper. The leach solution used in the column test is raffinate from the La Caridad heap leach circuit containing approximately 3.0 gpl total iron of which approximately 2.8 gpl is ferric iron. The free acid content of the leach solution is approximately 6 gpl. The column is irrigated at a rate of lps/m 2 and no rest or cure cycles are used. Pregnant leach solution (PLS) samples are collected daily, their volumes determined and the solution assayed for ferrous iron, total iron, free acid and total copper. The ferric iron is calculated from the total iron and ferrous iron assay. REVIEW OF COLUMN DATA AND SIMULATION OF RESULTS The factors, which determine the recovery/time curve, can be divided into two groups: the operating parameters of the heap leach or column leach and the ore characteristics. Operating parameters include all those parameters that the plant operator can independently select or that are a result of one of the parameters under control of the operator. Particle size, lift height and irrigation rate fall into this group. In addition heap porosity (the percentage voids within the heap) also is in this group since it is determined by the particle size, heap height and the manner in which the heap was built. The second group (ore characteristics) consists of parameters that the operator has no control over: ore specific gravity, porosity, mineralogy and copper mineralogy for example. The primary objective of this study was to determine a correlation between the ore characteristics (the second group of parameters) and the recovery/time curve. The relationship between the operating parameters (the first group of parameters) and the 3

5 recovery/time curve will be based on the shrinking core model as described by LEACH (4), a software package for simulating a heap leach operation. All column tests were, therefore, run under as identical condition as possible so that differences in their recovery/time curves would be only the result of differences in the ore characteristics of the individual ore samples use in each column. Analysis of the column leach test results revealed that the recovery/time curves of all of the ore samples could be described by the three ratios: acid soluble copper to the total copper in the sample, cyanide soluble copper to the total copper in the sample and insoluble copper to the total copper in the sample. Table I summarizes the copper chemistry for all of the column tests run with the standard column test procedure. The recovery/time curves for a selected few of these tests are shown in Figure1 in which the recovery of total copper is plotted. In Figure 2 the same column tests are plotted showing the recovery of soluble copper. The copper recoveries in the tests were fitted by least squares regression analysis to an equation of the form: % Rec t Cu(total) = A t x S.I.(A.S.) + B t x S.I.(CN sol.) +C t x (Insol Cu) (1) where: % Rec t Cu(total) = Recovery of total copper at time t, A t, B t and C t = Constants, time variable, S.I.(A.S.) = Ratio of acid soluble copper assay to total copper assay, S.I.(CN sol.) = Ratio of cyanide soluble copper assay to total copper assay, Insol Cu = Ratio of insoluble copper content to total copper assay. Equation 1 states that the copper recovery will be a function of the fraction acid soluble copper content, the fraction cyanide soluble copper content and the fraction insoluble copper content: the basic copper mineralogy of the ore. The equation also presumes that each of the three forms of copper present will leach independent of the amount of the other two forms of copper present and that all of the other characteristics of the ore will have no measurable effect on the recovery/time curve. Although this may be intuitively incorrect, if the variation in these other characteristics among the test samples is small then their effect on the recovery/time curve can be small or masked by the effect of the copper mineralogy. In addition any effect caused by these other characteristics will be indicated by the correlation coefficient of the regression equation. The results of the regression analysis are summarized in Table II. The constants, A, B and C can be interpreted as the recovery of the their respective copper component at the corresponding times. Figure 3 shows the plots of the constants versus time. The adjusted correlation coefficient represents the amount of the change in 4

6 recovery that is due to ore characteristics incorporated in equation 1. For example, at 20 days into leaching, equation 1 accounts for 87.1 percent of the change in recovery, the other 12.9 percent must be attributed to factors not included in the equation. These would include experimental error, variations in operating parameters which were intended to be held constant in all tests (such as leach solution chemistry, irrigation rate, percent voids in the 5

7 Table I - Column leach test feed characteristics Column Ore Zone Cu(tot), % Cu(A.S.), % Cu(CN sol), % S.I. Acid S.I. Cyanide S.I. Acid/ S.I. Cyanide S.I. Total

8 Figure 1 - Recovery of Total Copper for Selected Column Leach Tests 7

9 Figure 2 - Recovery of Soluble Copper for Selected Column Leach Tests Table II - Results of Regression Analysis 8

10 Time, days A B C Adjusted Correlation Coefficient

11 Figure 3 - Regression Coefficients as a Function of Leach Time column, column height, particle size distribution, etc.) and differences in the leachability of the ore samples as a result of differences in the ore characteristics which effect the recovery/time curve, such as the ore porosity and reagent consumption. Table III contains measured recoveries and the calculated recoveries at selected times for column tests on 32 ore samples. The measured recoveries versus the calculated recoveries are plotted in Figure 4. The numerical values for the C term are interesting from both an academic point as well as a practical point. The negative value suggests that the chalcopyrite is initially acting as a preg robber. With increasing leach time this effect is reduced and the chalcopyrite then contributes to the copper produced by the column. This can be explained by the following chemical reaction: CuFeS 2 + Cu +2 2CuS + Fe +2 (2) Initially the chalcopyrite reacts with copper in the leach solution, precipitating the copper as covelite and releasing iron into solution. As leaching progresses the covelite undergoes dissolution: CuS + 8Fe H 2 O Cu +2 + SO 4 = +8Fe H + (3) 10

12 and this second reaction releases more copper into solution than the first removes from solution. This reaction sequence is the typical A B C reaction where chalcopyrite is A, covelite is B and C represents solubilized copper. Consequently, the constant C is initially negative but eventually becomes positive. This explanation of the role of chalcopyrite was supported by the observation that in some of the leach residues more covelite was found than could be accounted for by the covelite in the sample head plus the covelite produced from leaching half of the copper from the chalcocite. In addition, this reaction sequence is known to produce covelite during the alteration of primary copper deposits. There have been several reports of an induction period in leaching chalcopyrite ores similar to that suggested by the above sequence of reactions. They have been attributed to an acclimatization period needed by the bacteria before taking part in the leaching process. The chalcopyrite eventually contributes to the production of copper from the heap. The column tests were not of sufficient duration to establish a recovery/time curve for the insoluble (chalcopyrite) copper fraction, however, the results of the standard copper mineralogical assay procedure indicated that the insoluble copper did leach under the typical column leach test conditions and data from the commercial heap indicates that some chalcopyrite is leached. Table III - Column leach test measured and calculated recovery Recovery, % (measured/calculated) Column 5 days 10 days 20 days 40 days 60 days 80 days / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

13 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

14 Figure 4 - Comparison of measured and calculated Copper Recovery Once the recovery/time curve under the base case test conditions is determined for any ore block, the recovery/time curve for the ore leached on the commercial heaps can be estimated based on the shrinking core model and a mass balance for the leach solution using the computer program LEACH. This is accomplished by dividing the copper in the ore into its acid soluble and cyanide soluble components, calculating the commercial heap s recovery/time curve for each of the two components independently then adding the recoveries for the two components together to obtain the recalculated recovery for the total ore. The contribution of the chalcopyrite to the total recovery was estimated to be one percent per year (i.e. 1 percent of the copper in the chalcopyrite would solubilized each year the ore was under leach). PROJECTION OF COMMERCIAL HEAP LEACHING RESULTS The high numerical values of the adjusted correlation coefficients for the regression equations imply that the recovery/time curves are almost completely and solely defined by the copper mineralogy of the sample. Neither the ore zone from which the samples originated, the degree of alteration of the ore nor the grade of the ore appears to have any measurable influence on the recovery/time curve. Operating parameters (as opposed to ore characteristics) for the column tests were held constant. Particle size distributions, irrigation rates, column heights etc. were 13

15 intentionally kept very nearly constant in all tests. The affect of changes in these parameters on the recovery/time curve can be calculated by the shrinking core model employed by the computer program LEACH. The constants generated by the regression analysis of the column test data physically represent the recoveries of the copper in each of the three copper mineral groups: the acid soluble fraction (copper oxides), the cyanide soluble fraction (secondary copper sulfides) and insoluble fraction (chalcopyrite). The overall copper recovery from a column leach test is found by taking a weighted average of these three curves. Given the recovery/time curves in Figure 3, the leachabilities based on the shrinking core model of the acid soluble and cyanide soluble fractions of the copper in the ore were calculated using the computer program LEACH. Once the leachabilities had been determined, the overall recovery/time curves were calculated for commercial heaps using the operating parameters of the commercial heaps and the computer program LEACH. Simulation of the commercial heap operation give the following results: The estimated recovery of the total contained copper in year 1 of leaching is: % Recovery Cu(total) = 91.1 x S.I.(A.S.) x S.I.(CN sol.) (3) The estimated recovery of the total contained copper in year 2 of leaching is: % Recovery Cu(total) = 8.9 x S.I.(A.S.) x S.I.(CN sol.) (4) The estimated recovery of the total contained copper in year 3 of leaching is: % Recovery Cu(total) = 5.0 x S.I.(CN sol.) (5) The estimated recovery of the total contained copper in year 4 of leaching is: % Recovery Cu(total) = 3.3 x S.I.(CN sol.) (6) The following points should be noted: All of the acid soluble copper is recovered in two years. The recoveries given are incremental recoveries : that is they are the recovery for the year, not cumulative recoveries. The column test does not provide an estimate of the copper recovery from the insoluble copper (chalcopyrite). This has been assumed to be 1 percent of the copper in the chalcopyrite per year and was added to the above equations. The cumulative recovery for any number of years is found by summing the 14

16 coefficients of the equations for all the years over which the cumulative recovery is desired. Core samples of the La Caridad deposit have been collected and assayed for acid soluble copper, cyanide soluble copper and total copper. A block model has been developed from the these core samples. Equations 3 to 6, together with an appropriate present value discount factor have been used to assign a value to each block for the copper that can be recovered by heap leaching. Because of the correlation that has been developed between the copper mineralogy and the copper recovery from heap leaching it is not necessary to run a column leach test on every core sample. CONCLUSIONS This study resulted in three observations. First, the variation in response to leaching of different La Caridad ore samples is primarily the result of variations in copper mineralogy of the different ore samples. While the ultimate copper recovery of each ore sample is a function of the percentage of the copper contained in the oxide and secondary sulfide minerals, the rate of leaching is primarily a function of the ratio of oxide to secondary sulfide minerals present. Other ore characteristics either have a minimal effect on the recovery/time curve or their effects have been masked by the copper mineralogy. Second, given a sufficient number of column leach tests the recovery/time curve can be separated into individual recovery/time curves for the three forms of copper present in the ore: oxide copper, secondary sulfide copper and primary sulfide copper. The recovery/time curve for the individual copper components of the ore can be scaled up to the commercial heap leaching operating parameters based on the shrinking core model. The projected overall recovery/time curve for the commercial heap leaching operation is found to be the weighted average of the recovery/time curves for these components. Once a sufficient number of column leach tests have been conducted it is possible to construct a recovery/time curve for any ore sample from the mineralogical assay of the sample: a column leach test is not needed. Third, chalcopyrite appears to act initially as a preg robber precipitating copper from the leach solution. This reaction converts the chalcopyrite to covelite that eventually dissolves, contributing to the copper production from the ore. 15

17 ACKNOWLEDGMENTS The authors would like to acknowledge and thank the management of Mexicana de Cobra and Mexicana de Cananea for there support during this project and their permission to publish this paper. In addition we would like to thank the many other individuals at both La Caridad and Cananea who participated in the experimental program and in discussions on the column leach program and the plant operations. REFERENCES 1. B.R. Benner and R.J. Roman, Determination of the Effective Diffusivity of H+ Ions in a Copper Ore, AIME Transactions, VOL 256, 1974, (Also see ) 2. R.J. Roman, B.R. benner, and G.W. Benner, Diffusion Model for heap Leaching and Its Application to Scale-Up, AIME Transactions, Vol 256, 1974, (Also see ) 3. G.A. Parkison and R.B. Bhappu, The Sequential Copper Analysis Method geological, Mineralogical, and metallurgical Implications, paper presented at the SME Annual Meeting, Denver, CO, USA, 6-9 March 1995, Preprint No (Also see ) 16