Abstract
M.Eng. (Extraction Metallurgy)
This study investigated the effects of sulfuric acid concentration, sodium chloride concentration, temperature, particle size distribution, and leaching time on copper and cobalt recovery from a sulfide ore. A representative sample was first characterized using X-Ray Fluorescence (XRF), X-Ray Diffraction (XRD) and Scanning Electron and Microscopy coupled with Energy Dispersive X-Ray Spectroscopy (SEM-EDS) to determine the chemistry, the mineralogy, and the mineral distribution, respectively. The outcomes from the characterization revealed that the ores contain copper and cobalt as valuable minerals associated with iron and disseminated in the gangue minerals which occur principally in a silicate form. After characterization, the sample was subjected to leaching in the presence of a mixt solution containing H2SO4 and NaCl. The thermodynamic and kinetic aspects of understanding the dissolution of Cu and Co from sulfide sources were studied while investigating the effect of parameters such as H2SO4 concentration, the addition of NaCl, particles size and temperature. The atomic absorption spectroscopy (AAS) was used the determine the amount of dissolved copper, cobalt, and iron. The residues generated from the leaching were also characterized for the chemistry, mineralogy, and mineral distribution while the outcomes were compared to the as-received sample. The results revealed that an increase of temperature from 65 to 85℃ had a positive effect on the recovery of copper and cobalt while under 55 ℃ no large influence was observed. Also, the acid concentration had a significant role in dissolving copper and cobalt from its matrix. The recovery of copper, cobalt and iron was essentially dependent of sulfuric acid concentration in the pH range from 0.8 to 2. Leaching experiments were ran with and without sodium chloride, a range of NaCl concentrations from 1 to 4 M was added in the leaching solution; it was observed that copper and cobalt recovery was enhanced while sodium chloride was added in the leaching solution but started decreasing at high temperature (85 ℃) when the sodium concentration was about 4 M. This recovery decline might be explained by the fact sodium chloride crystallize at temperatures above 80℃. Cobalt recovery was half that copper; the main reason of that observation is that, in the presence of H2S, cobalt can precipitate from an aqueous solution as cobalt sulfide with different stoichiometric ratios and iron precipitated in form of natro-jarosite. Copper and cobalt minerals were exposure at all particle size distribution by this fact a slight increase of copper and cobalt leaching rates was observed by varying the particle size distribution. Chemical control, and diffusion control models gave the best fitting to the copper, cobalt, and iron data. The activation energies were calculated in the temperature range IV from 45 to 85℃. The use of the Arrhenius equation gives the values of the activation energy of 10.92, 18.83 and 25.25 kJ/mol respectively for copper, cobalt, and iron dissolution, for the chemical reaction control and 31.023, 37.28 and 50.32 kJ/mol respectively for copper, cobalt, and iron dissolution for the diffusion reaction control. The measured maximum recovery of copper, cobalt and iron within 120 minutes were nearly 24.71%, 15.65% and 15.64% respectively. Moreover, to compare experimental output data from leaching experiments and the predicted value, an artificial neural network (ANN) was developed. The copper, cobalt, and iron recoveries as desired responses; sulfuric acid concentration (M), sodium chloride concentration (M), temperature (℃) and particle size distribution (μm) and leaching time (minutes) were the inputs variables and the stirring speed as the bias. Feed-forward back-propagation artificial neural network algorithm was used to develop, train, and predict the model. A total of 204 sets of data were used in the development and training of the model. To reach the network with a good agreement and highest generalizability and to reduce the error between the measured and predicted, the neural networks with different number hidden layers (one up to ten number of hidden layers) was searched. As a result, three architectures (arrangements) were developed; {5-1-3}, {5-5-3} and {5-10-3} architectures could give the most accurate prediction for copper, cobalt, and iron leaching rate. The regression analysis of the {5-1-3-3} arrangement gave the correlation coefficient of 0.98, 0.96, and 0.98 and the mean square error of 0.87, 1.37, 0.69 for the training, validation, and testing set, respectively for copper, cobalt, and iron recoveries. In the {5-5-3-3} architecture, it found a good correlation coefficient of 0.995, 0.99, and 0.99 and the mean square error of 0.224, 0,35, and 0.26 for the training, validation, and testing set, respectively. The correlation coefficient was about 0.997, 0.997, and 0.997 and the mean square error values were 0.111, 0,148, and 0.106 for the training, validation, and testing set in {5-10-3-3} architecture. The results demonstrated that ANN has a good potential to predict copper, cobalt, and iron recoveries. The increase in the number of hidden layers was found to increase the performance of the ANN model. The predicted copper, cobalt, and iron recoveries of {5-10-3-3} architecture was used for the comparison with the experimental values. The predicted maximum recovery of copper, cobalt and iron within 120 minutes were nearly 24.794%, 15.576% and 16.204% respectively.