A statistical approach to the experimental design of the sulfuric acid leaching of gold-copper ore

A statistical approach to the experimental design of the sulfuric acid leaching of gold-copper ore

F.D.Mendes; A.H.Martins^*

Departamento de Engenharia Metalúrgica e de Materiais, Universidade Federal de Minas Gerais, Phone (0055) (31) 3238-1053, Fax: (0055) (31) 3238-1815,Belo Horizonte - Minas Gerais (MG), CEP 30160-030, Brazil, E-mail: [e-mail]

ABSTRACT

The high grade of copper in the Igarapé Bahia (Brazil) gold-copper ore prevents the direct application of the classic cyanidation process. Copper oxides and sulfides react with cyanides in solution, causing a high consumption of leach reagent and thereby raising processing costs and decreasing recovery of gold. Studies have showm that a feasible route for this ore would be a pretreatment for copper minerals removal prior to the cyanidation stage. The goal of this experimental work was to study the experimental conditions required for copper removal from Igarapé Bahia gold-copper ore by sulfuric acid leaching by applying a statistical approach to the experimental design. By using the Plackett Burman method, it was possible to select the variables that had the largest influence on the percentage of copper extracted at the sulfuric acid leaching stage. These were temperature of leach solution, stirring speed, concentration of sulfuric acid in the leach solution and particle size of the ore. The influence of the individual effects of these variables and their interactions on the experimental response were analyzed by applying the replicated full factorial design method. Finally, the selected variables were optimized by the ascending path statistical method, which determined the best experimental conditions for leaching to achieve the highest percentage of copper extracted. Using the optimized conditions, the best leaching results showed a copper extraction of 75.5%.

Keywords: copper, gold, extraction, sulfuric leaching, optimization.

INTRODUCTION

Conventional cyanidation of gold-copper complexed ores may become infeasible, since most copper minerals are soluble in alkaline cyanide medium. In the case of Igarapé Bahia gold-copper ore, 73% of the copper minerals are soluble in this reagent. These side reactions result in a high consumption of the leach reagent, which increases the cost of production as well as decreases the recovery of gold. Ammonia-cyanide, bromine, thiosulfate and thiourea have all been suggested as alternative reagents for gold dissolution in the presence of copper minerals (La Brooy et al., 1994).

In gold ores with significant copper sulfide mineralization, copper may be removed via copper flotation or chemical pretreatment. The most economical alternative is to produce a gold-copper flotation concentrate for metal recovery by smelting. Cyanide and copper may also be recovered from the cyanidation solution to render the treatment feasible without pretreatment for copper removal. However, cyanide losses are considerably high (Sceresini and Richardson, 1991; Muir et al., 1989).

Another alternative approach in the case of the presence of copper oxide is to use a preleaching stage (e.g. ammonia, inorganic acid or chloride reagents) to remove copper minerals prior to cyanidation. Experimental results obtained by Mendes (1999) have shown that the sulfuric acid preleaching of Igarapé Bahia ore could offer the following advantages:

a) high solubility of copper in sulfuric acid solution due to the presence of large amounts of oxidized minerals,

b) technology that is widely available for copper recovery,

c) the low price of leach reagent,

d) technology for solvent extraction/electrowinning which is well known and widely applied in the mineral industry,

e) high level of recovery of gold from the leach residue and

f) regeneration of sulfuric acid after the leaching stage.

The Igarapé Bahia ore contains copper oxide minerals that can be solubilized by sulfuric acid according to Eq. (1) through (5) (Habashi, 1993; Gupta and Mukherjee, 1990; Habashi, 1997).

Malachite:

As shown in Eq. (2), when cuprite reacts with sulfuric acid under standard conditions part of it precipitates in a metallic copper form. However, in the presence of oxidant agents (air or ferric ions), cuprite is fully solubilized, as shown in Eq.(3).

Cuprite:

Cuprite:

Copper extraction by sulfuric acid leaching is limited, since the native copper is not sufficiently reactive to free hydrogen from diluted acid solutions. Because of this, it is not easily dissolved by acidic reagents unless an oxidant is present. Equation (4) indicates that in the absence of oxygen, native copper can be solubilized only in high-temperature systems with concentrated acid.

Native copper:

In the presence of oxygen, this reaction occurs slowly at room temperature, as shown in Eq. (5).

Native copper:

From this information, it is possible to evaluate the effects that certain variables, such as the level of oxygen saturation, concentration of sulfuric acid and percentage of solids, may have on the extraction response.

All these aspects justified research on a sulfuric acid leaching stage followed by neutralization and cyanidation. Metal oxide leaching, especially of copper oxides (Habashi, 1993; Gupta and Mukherjee, 1990; Habashi, 1997) has already been extensively studied. The copper leaching theory as well as the driving force of the kinetics of those reactions has also been very well established. However, little information was available on the behavior of the Igarapé Bahia ore submitted to a sulfuric acid leaching stage for copper extraction as a pretreatment of the ore for recovery of gold by cyanidation.

This work was aimed at evaluating the experimental conditions for maximum copper removal by sulfuric acid leaching of Igarapé Bahia gold-copper ore. Seven parameters showing a big influence on the maximization of copper extraction were studied. These parameters were selected from the literature dedicated to the leaching of copper ores (Habashi, 1993; 1997; Gupta and Mukherjee, 1990) and preliminary tests performed on a laboratory scale. Due to the large number of tests for evaluation of the influence of so many variables on the experimental response and optimization process, a statistical approach was applied to the experimental design. As well, the statistical approach permitted determination of the numerical values of the variables for maximum copper extraction at the leaching stage. As mentioned before, the copper content in the leach residue has a very important role in evaluating this technological alternative for further recovery of gold by cyanidation.

EXPERIMENTAL PROCEDURE

Mineral samples studied in this work were from a gold transition ore (1.76g/t Au), situated in the intermediate portion between the oxidized and primary ore in the Igarapé Bahia mine (Brazil). This ore had a high grade of copper (1.93%) and the main copper minerals were malachite [Cu₂(OH)₂CO₃], cuprite [Cu₂O] and native copper (Cu). Small amounts of chalcocite, covelite, chalcopyrite and bornite were also found. Table 1 shows the variables and their experimental levels studied using the Plackett-Burman statistical method for selection of the variables that had the largest influence on the experimental response, i.e., percentage of copper extracted (Cassa, 1976; Plackett and Burman, 1946). The variables selected by the Plackett-Burman method were applied to the replicated full factorial design method to evaluate the individualized influence of each variable, as well as the effect of their interactions on copper extraction (Brinck et al., 1996; Box et al., 1978). Table 2 shows the variables and the experimental levels studied with the replicated full factorial design method. The results obtained were also used in the ascending path statistical method to determine the optimum experimental conditions of the selected variables for maximum copper extraction (Duckworth, 1968). It is not within the scope of this paper to depict the basic theory and application procedures of the statistical methods adopted in this work.

The copper leaching experiments were conducted in a 1000 mL three-neck distilling glass flask reactor. A glass impeller with three circular blades, driven by a mechanical device and with speed control, was used to stir the leach pulp. The temperature of the leach solution was kept constant using a heating plate with automatic control.

Table 3 shows the chemical analysis for the mineral sample of Igarapé Bahia ore and for the residue from the sulfuric leaching test carried out under the optimized conditions. Aluminum, calcium, copper, iron, manganese and magnesium were determined by inductive coupled plasma ICP/AES (Spectro, Modula model). LECO combustion was used for sulfur determination. Gold was determined by fire assay.

Mineral species in the sample ore and in the leach residue (see Table 4) were determined by optical microscopy analysis (Zeiss, Axioplan model), scanning electron microscopy - SEM (Philips, XL 30 model) and X rays diffractometry (Philips, X'Pert model).

RESULTS AND DISCUSSION

Selection of Variables by the Plackett-Burman Statistical Method

By applying the Plackett-Burman method, it was possible to select the variables which had the largest influence on percentage of copper extracted at the sulfuric acid leaching stage. Table 5 shows the matrix block of replicated experiments for the Plackett-Burman statistical method. The percentage of copper extracted was the experimental response for each test.

Analysis of the results in Table 6 showed that temperature of leach solution, stirring speed, concentration of sulfuric acid in the leach solution and particle size of the ore were variables with the greatest influence on the experimental response. These results are in agreement with the copper leaching theory. Stirring speed and particle size usually affect leaching velocity when there are external and internal mass transport resistances. This also means that there was a significant difference in the results for percentage of copper extracted when the tests covered the experimental levels of the variables from the highest to the lowest value.

All of the most influential variables, described before had a statistical confidence level greater than 70%. The most usual procedure for the Plackett-Burman method is to choose variables with a statistical confidence level greater than 95%. Nevertheless, if this procedure were followed, just one variable (temperature) would be studied in the next stages of this work. Consequently, a 70% statistical confidence level was adopted to incorporate other variables that played an important role in the leaching stage and to increase the volume of technological information on the subject.

However, the variables percentage of solids in the leach pulp, leaching time and injection of oxygen into the leaching system showed very small influences on copper extraction. Leaching time showed a very important influence only during the first 15 minutes, when most copper minerals were dissolved. Indeed, the 1 to 4 hour range of time adopted based on information in the literature (Habashi, 1993, 1997; Gupta and Mukherjee, 1990) was considered too wide to have any significant effect on the experimental response.

The use of lower or upper experimental levels for these three variables did not cause any significant difference in the experimental response. Nevertheless, the negative values in the variable effect column in Table 6 indicate that by decreasing the experimental levels of particle size of the ore and percentage of solids in the leach pulp, the percentage of copper extracted increases.

Determination of the Influence of the Variables and their Interactions by the Replicated Full Factorial Design Method

The four variables selected previously were used in the replicated full factorial design method, which determined the influence of each variable and its interactions with other variables on the percentage of copper extracted. Table 7 shows the matrix block of replicated experiments for the replicated full factorial design method. The experimental levels used were kept the same as those used in the Plackett-Burman method and the experimental response was the percentage of copper extracted.

According to the mean difference column in Table 8, positive values for temperature, stirring speed and the concentration of sulfuric acid in the leach solution mean that the experimental levels of these variables should be raised to their highest values in order to improve copper extraction. Based on the same evaluation, the experimental level of particle size of the ore should be reduced to its lowest value, i.e., finer mineral samples cause an increase in copper extraction. Temperature was the most significant variable and exerted the greatest influence on the experimental response. This result is in agreement with the literature (Habashi, 1993, 1997; Gupta and Mukherjee, 1990) as the activation energies of these reactions are around 10 to 20 kJ/mol, which means that over the temperature range of 25 to 80^oC, one can expect the rate of leaching to double or triple. In the same manner, stirring speed has a large impact below the level at which complete particle suspension is achieved.

A list of influences of the variable on the experimental response given in decreasing order, follows: temperature of the leach solution (A), particle size of the ore (B), stirring speed (C), and concentration of sulfuric acid in the leach solution (E). These variables and the interactions AB and BC had confidence levels greater than 99%. The statistical confidence level in the replicated factorial method may indicate the level of importance of the variable in the expected response. For instance, a 99% statistical confidence level indicates that the variable or interaction has a strong influence on the experimental response.

The individual contributions of A (9.1) and B (-6.8) were positive and negative, respectively (see Table 8). However, interaction AB showed a positive effect on the experimental response. This means that the simultaneous increase in variables A and B from their lower levels to their upper levels improves copper extraction, as shown in the mean difference column in Table 8 for AB interaction (3.5). This performance may be explained by the fact that when particle size is increased, it causes a decrease superficial area available, in limiting contact between the phases during the reaction. However, raising temperature in pulps with coarser particles overcomes this difficulty, since this causes an increase in the vibration of molecules inside the structure until they rupture and are easily attracted by the solvent particles, which results in an increase in copper extraction.

A similar analysis can be made for the BC interaction. The simultaneous increase in particle size and stirring speed values results in an increase in copper extraction. This means that in these small reactors, even if the particles are coarser, an increase in stirring speed makes suspension of the particles easier and causes a continuous renovation of sulfuric acid consumed on the particle surface, improving the external mass transfer characteristics. Nevertheless, the contributions of temperature (A) and stirring speed (C) to the experimental response seem to be much higher than that of particle size of the ore (B).

Some interactions, such as AC, CE, ABC, BE and ABE, showed a negative effect on copper extraction. This means that the simultaneous increase in the values of their experimental levels values causes a drop in the experimental response. Others, such as ACE, BE, CE and BCE, showed a small influence on copper extraction during the leaching stage.

Optimization of the Experimental Results by the Ascending Path Statistical Method

Temperature of the leach solution, particle size of the ore, stirring speed and concentration of sulfuric acid in the leach solution were optimized by the ascending path statistical method in order to determine the best leaching conditions for maximum copper extraction.

Table 9 shows the steps for variables optimization using the ascending path statistical method. Based on the technological limitation of operating high-temperature pulps in open leach vessels, projected response number 4 was chosen as the best option, since it had a temperature value below 90^oC. The optimum experimental conditions, determined statistically, were temperature of the leach solution (82^oC), concentration of sulfuric acid in the leach solution (98.5kg/t), particle size of the ore (0.044mm), injection of oxygen into the leach system (8ppm) and stirring speed (270rpm). The statistical procedure predicted a maximum copper extraction of 76.5%.

It must be recognized that there is a difference between the total acid consumption of the ore (ore-dependent) and that which reacts with the copper phases (phase-dependent). Of a total of 98.5kg/t acid added, 22.3kg/t was used for copper dissolution. This acid can be regenerated in the next processing stage, which allows conclusion that total consumption of sulfuric acid was 76.2kg/t. Figure 1 shows that the decreasing concentration of sulfuric acid added to the system caused an appreciable decrease in relation to recovery of copper, while a slight increase in this concentration up to 98.5kg/t barely affected these results. Thus, it was concluded that the optimum value of 98.5kg/t acid added determined by the statistical approach could be considered the best value for the sulfuric acid leaching stage.

Optimized values for percentage of solids in the leach pulp and leaching time were determined experimentally, since these variables were not taken into account in the statistical optimization procedure due to their small influence on copper extraction. Their influence on copper extraction was studied separately, while all other parameters were held at the previously established values.

Figure 2 shows the effect of the percentage of solids on percentage of copper extracted. Only a small variation in copper extraction results can be seen for different percentage of solids in the leach pulp, confirming the small influence of this parameter on the experimental response. Percentage of copper extracted was slightly higher in pulps with 40% solids, and for this reason this value was adopted for the sulfuric acid leaching stage.

The influence of the leaching time on solubilization of copper minerals was also evaluated. Figure 3 shows the variation in percentage of copper extracted as a function of leaching time. A small variation in copper extraction can be seen between 15 and 120 minutes. Because of this result, the shortest leaching time (15 minutes) was adopted resulting in lower processing costs. Copper minerals were leached easily in sulfuric acid solution mainly the oxidized species were dissolved quickly in less than 15 minutes and longer time did not affect copper extraction.

Table 10 shows the variables studied in this work that have an influence on copper extraction and their respective optimized values. Two experiments were performed under the optimum conditions to confirm the statistical evaluation. The results obtained were 75.5% copper extraction, while the statistically predicted response was 76.5%. The difference between these two results was smaller than 5.0%, and for this reason, the statistical approach used in this work could be considered valid and reliable.

It was not possible to attain copper extractions greater than 76% in open vessels due to the presence of refractory species, such as native copper and sulfides. After defined the optimum conditions, some sulfuric acid leaching tests which evaluated the total copper and native copper extraction during 96 hours were carried out. This was an attempt to obtain percentages of copper extracted of over 76%. Figure 4 shows that after 96 hours, 79% copper extraction was achieved. This represented a small increase in copper extraction for such a long leaching time. The native copper extraction apparently showed the same behavior, with little increase in the results, confirming the small variation in final percentage of extraction in relation to time.

Therefore, the sulfuric acid leach residue was 0.52% copper, of which 0.30% was undissolved native copper. The remaining copper may be assumed to be a sulfide copper with low solubility in sulfuric acid under the operational conditions. The sulfuric acid leaching residue was neutralized with a sodium hydroxide solution and submitted to cyanidation. At this latter stage, 94% gold extraction was achieved; however copper, mainly native copper, was also solubilized (66%), causing a high consumption of reagent.

CONCLUSIONS

The sulfuric acid leaching stage was studied based on a statistical approach applied to the experimental design. Using the Plackett-Burman method, temperature of the leach solution, stirring speed, concentration of sulfuric acid and particle size of the ore were selected as the variables that had the largest effect on the experimental response. Other variables, such as percentage of solids in the leach pulp, leaching time and injection of oxygen into the leaching system were determined to be less important.

After that, the variables selected and their interactions were evaluated using the replicated full factorial design method, which provided results on the influence of the variables and their interactions and the statistical significance of the percentage of copper extracted. The results showed that, in decreasing order, the variables temperature (A), particle size of the ore (B), stirring speed (C) and concentration of sulfuric acid in the leach solution (E) and interactions AB, AC, AE, BC, ABC, ABE and ABCE, had the greatest influence on the experimental response all of them had statistical confidence levels over 70%.

Finally, the variables selected were optimized using the ascending path statistical method, which predicted a 76.5% maximum copper extraction. The following optimum experimental conditions were adopted: a 98.5kg/t concentration of sulfuric acid in the leach solution, a 270 rpm stirring speed, a temperature of 80ºC, 40% solids in the pulp, a 0.044mm particle size of the ore, 8 ppm injection of oxygen into the leach system and 15 minutes of leaching.

Two follow-up experiments were carried out under the statistical optimized conditions and the results showed 75.5% copper extraction. The difference between the statistical prediction (76.5%) and the experimental result (75.5%) was less than 5.0%, which validates the statistical approach used in this work.

Sulfuric acid leaching did not provide more than 76% copper extraction because the only slightly native copper that was leached (23%). For this reason, the sulfuric acid leach residue still had 0.30% native copper and 0.52% total copper. From the remaining copper in the leach residue, 66% was extracted at the cyanidation stage, causing high cyanide consumption. Thus, it can be concluded that further efforts should be made to develop alternative procedures to reduce the copper content in the leach residue to make the recovery of gold in the cyanidation stage more economical.

ACKNOWLEDGEMENTS

The authors thank Companhia Vale do Rio Doce (Brazil), Programa de Núcleos de Excelência em Tecnologia Mineral do Ministério da Ciência e Tecnologia (PRONEX-Brazil) and Programa Institutos do Milênio (CNPq/MCT/Brazil) for the financial support that allowed this experimental work to be carried out and to be published.

Received: November 28, 2001

Accepted: April 23, 2003

• Box, G. E. P., Hunter, W. G. and Hunter, J. S., Statistics for Experimenters: An Introduction to Design,Data Analysis and Building. Ed. Chichester, Ny, 653pp (1978).
• Brinck, V., Hauck, G. A. C. And Souza, H. H., Curso e Planejamento e Análise Racional de Experimentos Tecnológicos. Usinas Siderúrgicas de Minas Gerais, Brazil, 61pp (1996).
• Cassa, J. C. S., Seleção de Variáveis de Processo e Programação de Experimentos pela Abordagem Estatística de Plackett-Burman. Anais do Congresso Brasileiro de Engenharia Química, Brazil, July, Nş 1, pp. 27-29 (1976).
• Duckworth, W. E., Statistical Techniques in Technological Research: An Aid to Research Productivity. Ed. Methuen, London, 303pp (1968).
• Gupta, C.K.and Mukherjee, T.K., Hydrometallurgy in Extraction Processes. CRC Press Inc, vol. 1, 225pp (1990).
• Habashi, F. A., A Textbook of Hydrometallurgy. Métallurgie Extractive Québec Enr., Quebec, Canada, 689pp (1993).
• Habashi, F. A., Handbook of Extractive Metallurgy. Departement de Mines et de Metallurgie, Quebec, Canada, 4vol., 2379pp (1997).
• La Brooy, S. R., Linge, H. G. and Walker, G. S., Review of Gold Extraction from Ores. Minerals Engineering, vol. 10, no. 7, pp. 1213-1241 (1994).
• Mendes, F. D., 1999. Tratamento do minério de Ouro-cobre de Igarapé Bahia através de lixiviação sulfúrica seguida de cianetação. Master's thesis, Universidade Federal de Minas Gerais, Brazil (in Portuguese), 116pp.
• Muir, D. M., La Brooy, S. R. and Cao, C., 1989. Recovery of Gold from Copper-bearing Ores. Proceedings of World Gold '89. Reno, Nevada, pp. 363-374.
• Plackett, R. L. and Burman, J. P., 1946. The Design of Optimum Multifactorial Experiments. Biometrika Trust, London, 305pp.
• Sceresini B. and Richardson P.,1991. Development and Application of a Process for the Recovery of Copper and Complexed Cyanide from Cyanidation Slurries. Randol Gold Forum '91. Randol Internation Ltd, Cairns, pp. 265-269.

*

To whom correspondece should be addressed

Publication Dates

• Publication in this collection
  01 Sept 2003
• Date of issue
  Sept 2003

History

• Received
  28 Nov 2001
• Accepted
  23 Apr 2003