Towards a more environmentally friendly process for gold: models on gold adsorption onto activated carbon from ammoniacal thiosulfate solutions Patricio Navarroa, Cristian Vargasa, Manuel Alonsob, Francisco Jose Alguacilb* a Departamento de Ingenieria Metalurgica, Facultad de Ingenieria, Universidad de Santiago de Chile, Avda. L.B. O´Higgins 3363, Casilla 10233, Santiago, Chile. b Centro Nacional de Investigaciones Metalurgicas (CSIC), Avda. Gregorio del Amo 8, Ciudad Universitaria, 28040 Madrid, Spain. E-mail: fjalgua@cenim.csic.es * Corresponding author 2 Abstract In the context of gold extraction, leaching is the dissolution of the metal or even a mineral in a liquid phase. Thus, it is of a primary concern the dissolution of gold in an aqueous solution, operation which requires both a complexant and an oxidant to achieve acceptable leaching rates and yields. A limited number of ligands form complexes of sufficient stability and at a suitable rate for their use in gold leaching operations. Cyanide is mostly used because of its relatively low cost and great efficiency for gold dissolution. The main disadvantage associated with its use is the extremely toxic character of the reagent. Thus, several other ligands had been considered for gold extraction from ores, regarding to environmental pressures and even restrictions to the use of cyanide, potential of having faster gold leaching kinetics than cyanide, possibility of their use in acidic media, which is suitable in the treatment of refractory ores, and a greater degree of selectivity than cyanide for gold over other metals. Nevertheless, some of them presented some disadvantages which limited their practical use. Ammoniacal thiosulfate leaching has a considerable potential as an effective and less hazardous procedure (than cyanide) for gold extraction from auriferous ores. Once gold is dissolved, separation of the precious metal from the solution can be achieved using different procedures (i.e. adsorption onto activated carbon). In the present investigation, up to fivel models were used to correlate the adsorption of gold onto the carbon, being the Fleming´s model the one which best fit the experimental results on gold adsorption. Keywords: Gold; Adsorption; Carbon; Modelling; Ammoniacal thiosulfate 3 1. Introduction A relative interest has attracted to the use of alternatives to the conventional cyanidation process to extract gold from gold ores. This interest in the use of other alternatives to cyanide for the dissolution of this precious metal arises from concerns relative to the toxicity of cyanide and the lack of effectiveness of cyanidation process to leach, in an effective form, carbonaceous or complex ores. Several alternatives to cyanide have been considered [1], and thiosulfate leaching emerges as one of the more attractive [2-4], because thiosulfate is considered a non-toxic reagent and can leach gold faster than cyanide. On the other hand, the main disadvantage of thiosulfate processing is the reagent consumption and the lack of a suitable gold recovery method. The dissolution of gold can be described by the following reaction: 4Au 8S 2 O 32 O 2 2H 2 O 4Au (S 2 O 3 ) 32 4OH (1) However, the presence of ammonia in the system is needed in order to prevent passivation of the precious metal and to stabilise copper (II), of which presence is also required to increase the rate of gold dissolution. The gold-ammine complex is converted to thiosulfate complex, since measurements at pH values greater than 9.0 have shown that gold rest potentials change with thiosulfate concentration rather than with ammonia concentration according to following equation: Au ( NH 3 ) 2 2S 2 O 32 Au (S 2 O 3 ) 32 2 NH 3 (2) 4 it was also described that under certain conditions gold can be present in solution as the ammine complex. Most of the literature devoted to the gold-thiosulfate processing has been concerned with the leaching of the precious metal, with limited research effort on the process of recovering gold from solution. The various alternatives proposed for gold recovery from thiosulfate media can be found elsewhere [2,5]. Activated carbon has the advantages that it can be added to the pulp and avoid soluble losses in tailings, but apparently, it has a low affinity for the gold-thiosulfate complex. From the view of point of engineering, before scaling up the technology, a theoretical model of the adsorption operation is needed in order to design an efficient recovery process. In the present investigation, several adsorption models [6-8], proposed for the adsorption of gold from cyanide medium, are tested in order to look for a suitable one to describe the adsorption of gold from a synthetic solution onto activated carbon from ammoniacal thiosulfate solution. 2. Experimental The gold thiosulfate complex used in this investigation was obtained from Alfaaesar, a stock gold solution was prepared by dissolution of the reagent in distilled water. Working gold solutions were prepared by dissolving a known quantity of the stock solution into the corresponding ammoniacal solution. All other chemicals were of reagent grade. 5 The activated coconut-shell carbon, obtained from Anglo Chilena SA, was washed with deionized water, dried at 60º C and kept in a dessicator under vacuum. Table 1 shows the main physical characteristics of the activated carbon. Gold adsorption tests were carried out in a 500 mL glass reactor equipped with a stirring device. The reactor was placed in a water bath. Experiments were run at 20º C and at 500 min-1 with 250 mL of the gold solution and the experiments were initiated by placing the weighed amount of the activated carbon into the reactor. The experimental conditions used to correlate the various model´s fit to the experimental data were summarized as shown in Table 2, at these conditions, gold adsorption performance can be considered as representative of the present system. Samples were taken periodically for gold concentration analysis by atomic absorption spectrometry using a Perkin Elmer 1100B spectrophotometer. The effectiveness of the adsorption process was estimated from the percentage of gold adsorption as the following: %Au ads Au 0 Au t 100 Au 0 3. Results 3.1. The model fitting The five models applied to fit the experimental data are described as the following: First order model (3) 6 The first order model can be represented by the equation (4): r k 1 Au t k 2 Au c, t (4) By linearization of the equation (4), a plot of r/[Au]t versus [Au]c,t/[Au]t can be built-up and a straight line should be obtained with slope k2 and intercept k1. The values of k1 and k2 obtained from the present experimental data are given together with the other models in Table 3. It should be noted here that data obtained from the 35 mg /L gold solution do not fit properly to the model (r2 < 0.9). In this model k1 can be related to a rate constant of the adsorption process, whereas k2 to the desorption process. The Dixon model In this model the corresponding equation is given by the following expression of: r k 1 Au t Au c,m Au c, t k 2 Au c, t (5) As in the above case, a linearization of the equation and by plotting r/[Au] t versus [Au]t([Au]c,m-[Au]c,t)/[Au]c,t, a straight line with slope k2 and intercept k1 should be obtained. As seen in Table 3, data experimentally obtained with the 35 mg/L gold solution do not fit to the model. The Nicol model The equation (6) can be used to represent the Nicol model: 7 r k 1 KAu t k 1 Au c, t (6) Similarly to all the above, a plot of r/[Au]t against [Au]c,t/[Au]t should results in a straight line with slope k1 and ordinate k1K. According to results shown in Table 3, the experimental data obtained using the 35 mg/L gold solution are not given due to the poor (R2<0.9) fit with this model. The Fleming model The model proposed by Fleming is given by the expression of: Au c,t k1 Au 0 t n (7) The experimental data were fitted to the model by linearization of the above by the plot log [Au]c,t versus log t. From this plot the coefficient n is given by the corresponding slope, whereas from the intercept the value of k1 can be calculated. The experimental data for 35 mg/L also fitted this model (R2>0.9) as seen in Table 3. The La Brooy model In this model, the following equation had been proposed to explain the adsorption of gold onto activated carbon: Auc,t k1 Aut t n (8) 8 Similar to the above model a plot of log ([Au]c,t/[Au]t) versus time should results in a straight line with slope n and ordinate to calculate k 1. Also this model fitted well (R2>0.9) the data obtained for 35 mg/L of gold concentration. 4. Discussion and Conclusions Five models have been examined in order to explain the adsorption of gold onto activated carbon from ammoniacal thiosulfate medium. Three of the above models (the first order, Dixon and Nicol models) seemed to be only suitable to explain the adsorption of gold at lower metal concentrations, whereas with the Fleming and La Brooy models the fit of the experimental data to the respective model can be broaded to all the range of the gold concentrations tested. However the fit of the experimental data to the Fleming model is slightly better than that obtained with the La Brooy model (Table 3). Thus, this model can best represent the adsorption of gold onto activated carbon within the experimental conditions using in this investigation. Fig. 1 compares the experimental data on gold adsorption (see Table 2 for details) with those obtained using the theoretical data estimated from the present model. It can be seen that reasonable good agreement is obtained between each set of data. On the other hand, from results obtained within this work it can be concluded that the values of kn tends to decrease as the initial gold concentration in the aqueous solution increases. Results of this work also allow to conclude that activated carbon has a low affinity for Au(S2O3)23- species, however the presence of the Au(NH3)2+ in the aqueous phase can not be completely neglected, and this species will help in the adsorption of the precious metal. Moreover, the presence of AuNH3S2O3- ion had been proposed [2], though it has 9 not been confirmed experimentally. A possible mixed gold-ammine-thiosulfate complex has also been mentioned in the literature [9]. As it is stated in the literature [4], there is a need for more thermodynamic data for the Au/NH3/S2O32- system. The effectiveness of the gold adsorption operation onto carbon can be improve by the use of a more effective operational device, i.e. columns, or also by adding a small amount of cyanide to the system [2]. Acknowledgements The authors wish to acknowledge CONICYT (Chile) for project FONDECYT 7040009 and CSIC (Spain) for support to carry out this work. Nomenclature [Au]0 initial gold concentration in the aqueous phase [Au]t gold concentration in the aqueous phase at an elapsed time [Au]c,t mass of gold adsorbed onto the carbon at an elapsed time [Au]c,m maximum mass of gold adsorbed onto the carbon [NH4OH] initial ammonium hydroxide concentration in the aqueous phase [S2O32-] initial thiosulfate concentration in the aqueous phase kn, K constants for the different models n time coefficient for the Fleming and La Brooy models r gold adsorption velocity onto the carbon t time R correlation coefficient 10 References [1] J. Marsden and I. House, The Chemistry of Gold Extraction, Ellis Horwood, New York, 1992. [2] M.G. Aylmore and D.M. Muir, Thiosulfate leaching of gold-a review, Min. Eng., 14 (2001) 135-174 and references therein. [3] P. Navarro, C. Vargas, A. Villarroel and F.J. Alguacil, On the use of ammoniacal/ammonium thiosulphate for gold extraction from a concentrate, Hydrometallurgy 65, (2002) 37-42. [4] E. Molleman and D. Dreisinger, The treatment of copper-gold ores by ammonium thiosulfate leaching, Hydrometallurgy, 66 (2002) 1-21. [5] P. Navarro, R. Alvarez, C. Vargas and F.J. Alguacil, On the use of zinc for gold cementation from ammoniacal-thiosulphate solutions, Min. Eng., 17 (2004) 825-831. [6] J.S. Van Deventer, Kinetic model for the reversible adsorption of gold cyanide on activated carbon, Chem. Eng. Comm., 44 (1986) 257-274. [7] J.D. Le Roux and A.W. Bryson, A comparison of several kinetic models for the adsorption of gold cyanide onto activated carbon, J. South African IMM, 91 (1991) 95103. [8] M.N. Vegter and R.F. Sandenberg, Rate-determining mechanisms of the adsorption of gold di-cyanide onto activated carbon, J. South African IMM, 96 (1996) 109-118. [9] P.L. Breuer and M.I. Jeffrey, An electrochemical study of gold leaching in thiosulfate solutions containing copper and ammonia, Hydrometallurgy, 65 (2002) 145-157. 11 Table 1 Characteristics of the activated carbon. Shape Cylinder Length (cylinder) 5.6 mm Diameter (cylinder) 3 mm Surface area 925 m2/g Micropores 96% Table 2 Experimental conditions used in the investigation. [Au]0 (mg/l) [S2O32-]0 (mg/l) [NH4OH]0 (M) pH±0.05 Carbon added (g) 10 11.2 1 10.5 0.16 25 28.0 1 10.5 0.16 35 39.2 1 10.5 0.16 In this table, the concentrations of gold and thiosulfate responded to these given by the Au(S2O3)23- stoichiometry 12 Table 3 Values of the constants and correlation coefficient of the models tested R2 [Au]0 (mg/l) aFirst 10a k1: 9.0x10-5 s-1 k2: 6.4x10-5 kg/L s 0.911 25a k1: 1.2x10-5 s-1 k2: -3.0x10-6 kg/L s 0.991 10b k1: 1.4x10-6 kg/L s k2: 3.2x10-5 kg/L s 0.999 25b k1: 1.6x10-5 kg/L s k2: 1.3x10-5 kg/L s 0.992 10c k1: 6.4x10-5 kg/L s K: 1.4 L/kg 0.911 25c k1: -3.0x10-5 kg/L s K: -3.4 L/kg 0.991 10d k1: 0.026 L/kg s n: 0.36 0.929 25d k1: 6.5x10-4 L/kg s n: 0.71 0.998 35d k1: 1.7x10-4 L/kg s n: 0.83 0.991 10e k1: 0.012 L/kg s n: 0.51 0.910 25e k1: 3.8x10-4 L/kg s n: 0.79 0.995 35e k1: 1.0x10-4 L/kg s n: 0.91 0.985 order, bDixon, cNicol, dFleming, eLa Brooy 13 Fig 1. Percentage of gold adsorption onto activated carbon. Experimental data (points), predicted values according to the Fleming model (lines). 14 Fig.1