Ion Exchange Recovery of Metals from Waste Water of Complex Sulfide Hydrometallurgy

Transcription

1 Hydrometallurgy, 23 (1990) Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands Ion Exchange Recovery of Metals from Waste Water of Complex Sulfide Hydrometallurgy w 29oq4,PnF FEDERICO MIJANGOS and MARIO DIAZ' Department of Chemical Engineering, Faculty of Sciences, University of Pais Vasco, Apdo 644, Bilbao (Spain) (Received April 30,1988; revised and accepted February 6,1989). ABSTRACT Mijangos, F. and Diaz, M., Ion exchange recovery of metals from waste water of complex sulfide hydrometallurgy. Hydrometallurgy. 23: The treatment by ion exchange ol' waste water from comples sulfide hydrometallurgy has Iwen studied. Several commercial resins have been tested for a range of ph in solution. The equilibrium results with an iminodiacetic resin for the multicomponent system have been fitted with basic equations. The simple methods proposed could be useful in industrial processing and design. INTRODUCTION The processing of complex sulfides is highly dependent on metal recovery. Sulfuric acid production from these ores is a function of the price of the cinders obtained from roasting. Nowadays copper, iron, gold, silver and zinc are being recovered from Iberian pyrite cinders. In order to make the process more economical and to reduce the contaminant effect of effluents, it is necessary to recover other valuable and contaminant metals. The problem of the low values of concentration involved in treatment of waste waters from hydrometallurgy is further complicated by high waste water flow rates usually leaving these industries [ 1 1. The use of ion exchange in the extraction of products in low concentrations can help to solve this problem if the interference of other cations found in higher degrees of concentration is avoided. It is also useful for achieving a good selectivity for specific metals in order to recover those of economic interest. Unfortunately, problems frequently arise in relation to the appearance of precipitates, mainly when considerable ph changes are produced, or due to the presence of solids entrained 'Present address: Department of Chemical Engineering, University of Oviedo, (Spain ) X/90/$ Elsevier Science Publishers B.V.

2 .. k 366 F. MIJANCOS AND M. DIA%,<-- from the previous pyro- or hydrometallurgical operations, or to the complexity due to the high ionic sttengths. The scope of the use of ion exchange in hydrometallurgy has been widely discussed [2-41. Metals like copper, zinc, cobalt or nickel have been recovered by ion exchange from industrial streams of different compositions with various technical solutions. The waste water of pyrite hydrometallurgy contains the four metals mentioned above. Cobalt is the most valuable metal and its recovery from the solution appears to be a complicated problem. In this work the above-mentioned complex solution, after recovery of iron, copper, zinc and noble metals, has been taken as a means of studying the application of ion exchange for retention of metals. Firstly a selection of different commercial resins has been made and tested with the industrial solution, in the ph range which is of practical interest, i.e. lower than 4.0. With the resin that gave the best results other experiments have then been carried out to obtain the equilibrium isotherm for the existing values of ph and metal concentration. Furthermore, due to the importance in modeling the kinetics and in fitting equilibrium data, these experiments give an important indication of process viability. EXPERIMENTAL METHODS Equilibrium data were determined by contacting resin and solution in a stirred batch for 8 h, a time found to be sufficient for achieving equilibrium (see Fig. 2). To study the effect of equilibrium ph, experiments with solutions prepared at different initial ph were undertaken. To avoid precipitations during the experiments, in sample preparations the concentration of some metals was slightly modified. The samples were analyzed by atomic absorption spectrophotometry. Results achieved from analysis of the resin elution and those obtained from mass balance of the solution data, gave good agreement over a wide range of conditions. STREAM CHARACTERISTICS AND RESIN SELECTION The average concentrations of the most relevant metals and other conditions of this type of stream were determined and are shown in Table 1. We have taken samples with this composition from a local industry, and the experiments were done with this base concentration. There was no interest in studying the equilibrium with artificial solutions by modifying each one of the metal concentrations. In the first selection made here, several types of ion exchangers were chosen looking for retention of heavy metals, trying to avoid the interference of high iron concentration. Commercially available macroporous resins of polystyrene-divinylbenzene

3 RECOVERY OF METALS FROM WASTE WATER 367, TABLE 1 Characteristics of the industrial waste water effluent Composition (g I-') Sodium 37.0 Iron 14.0 Potassium 3.0 Zinc 0.8 Calcium 0.45 Manganese Nickel L Copper Cobalt Cadmium Lead Sulphate Clilc ride % were selected (Table 2). Three types of chelating ligands, iminodiacetic, aminophosphonic and amidoxime, were tested. Anionic resins were tried due to the available chloride medium, and the series of ligands was completed with two cationic, carboxylic and sulfonic ligands. Twelve different resin types were selected with these groups from different commercial firms. For the three chelating ligands (regardless of the resin used) copper has the largest distribution coefficient (Table 3). Zinc was comparatively favoured with the aminophosphonic resin for which the retention for copper was smaller than with the other two. The distribution coefficient with amidoxime ligand for copper was more than 300 times that of the other metals, and so this ligand is specific for copper in this solution. For cobalt and nickel the highest distribution coefficient was obtained with the iminodiacetic group, giving also the highest quantity of global valuable metals under consideration. No resin showed important retention of manganese, and its analysis was not made in the following experiments. Distribution coefficients with quaternary amines (anionic resins) are too low to be used for extraction of the selected metals. The chloride concentration is not high enough to form anionic complexes. With sulfonic and carboxylic groups, the retention was very low due to the interference of iron. In the case of the carboxylic resins the highest value of the distribution coefficient corresponded to cobalt. For this reason further experimentation was carried out with this type of resin by changing the ph and the counterion, NHZ and H+

4 '\ / TABLE 2 Commercial resin properties Group Sulfonic Carboxylic Quater. Amidoxine Iminodiacetic Aminophos. Amine Resin L SI00 L SP112 D C20 D C26 L CMP80 D C464 L MP500 D ES346 LTP207 D ES466 A IRC718 D ES467 Bead size distribution (mm) Ionicform Na Na Na Na H H CI chelate free Na Na Na Na Density (kg m-3) Total capacity o - (eq I-') Moisture content? K E > n z t=

5 ~~ ~ ~ ~~ RECOVERY OF METALS FROM WASTE WATER 369, L -. TABLE 3 Distribution coefficients for different ligands with the effluent (I solution kg-' dry resin) Ligand PH co Ni cu Zn Mn lminodiacetic Aminop hosphonic < Amidoxime < <1 Sulphon ic Carboxylic < Quater. Amine < <1 TABLE 4 Distribution coefficients (I solution kg-i dry resin) with a carboxylic resin Counterioir PH CO Ni cu Zn M II H < 1 <1 < I H ;I N H: <1 4 <I 3 N H: instead of Na+ [5]. The approximate results shown in Table 4 are not promising enough to continue with this group. Later experiments were always done with the iminodiacetic type resin. SI<LKC'I'ION OF WORK1 NC CONDI'I'IONS In the selection of the commercial resins we considered that we needed to work in the acidic range of ph to avoid the precipitation of basic species. The- c/c, PH Fig. 1. Residual concentration of metals as a function of ph.

6 ..'. 370 F. MIJANGOS pind M. DIAZ oretical equations of solubility do not give good predictions for complex solutions especially those with very high ionic strength. For this reason and to find out the ph conditions where the mass of precipitate would remain manageable for fixed or stirred equipment, an experimental study was made. Digestion was done in a cylindrical stirred tank of for one hour, after the addition of NaOH 1 M. The precipitate, which was separated by centrifugation, appears at a ph value higher than 3.3, and is formed mainly of iron (ferric hydroxide ). The kinetics of Fe(I1) oxidation at a lower ph, even in the presence of precipitation is slow for these samples (oxidation reaction could be catalyzed, e.g. by copper ). Thus, in the clear liquid after separation, precipitation could continue at the same ph values, but the reaction is terminated by the protons produced in the reaction itself. Figure 1 shows copper and lead precipitation curves which bend sharply at ph = 5, while the slope is lower for the other metal curves, With this residence time in the fixed bed the ph of the solution should be lower than 3.3 to avoid having a clogging problem. In continuous or discontinuous agitated equipment, the solution ph could be higher, but some other aspects must be considered. EXPERIMENTAL RESULTS WITH AN IMINODIACETIC TYPE RESIN The kinetics of the loading were investigated in batch experiments to find the time required to achieve equilibrium and to look for possible kinetic differences of possible use in metal separation. In Fig. 2, the experimental results with liquid/solid ratio (1 solution/kg dry resin) of 438, are shown. Nickel and copper which are more strongly retained, also require a longer time to achieve equilibrium. The first order kinetic constants were estimated, the smallest being for nickel. In order to obtain an idea of the equipment size, units in Fig. 2 (mol/kg dry resin) can be converted using the factor 334 kg dry resin m-'3 of bed in Naform. -. time. mrn Fig. 2. Kinetics of the simultaneous loading of metals in a stirred tank.

7 rl. RECOVERY OF MM ALS FROM WASTE WATER 37 1 Although no acute differences were observed, from the results of Fig. 2, the residence time of the resin phase in continuous equipment can be selected, depending on the metal separation or loading purpose. For equilibrium determinations a contact time higher than 8 h was used in all the experiments. Equilibrium experiments were carried out by changing the liquid/solid ratio and the ph at equilibrium. The maximum degree of saturation for the concentration level shown in Table 1, increases from 0.33 to 1.08 mol kg- with liquid/ solid ratio from 43 to kg-, the iron load being excluded from the balance. Isotherm expressions to be considered must be a compromise giving accuracy and simplicity and in that case we look also for an easy application in the cases of metal separation and of whole metal retention. For this latter purpose we have considered the system as binary. The counterion concentration is calculated by summing up the individual metal concentrations of all the most heavily loaded. The free chelating group in the resin was calculated from the 0 L 8 I L ZC,. mo//mj Fig. 3. Total retention of metals versus the sum of concentrations in the solution. ie l PH Fig. 4. The global distribution coefficient as a function of the solution ph.

8 E P. MIJANCOS AND hl. DIN difference between the total capacity and the actual load of the metal under consideration. The sodium concentration in the solution is taken as constant. The experimental results are shown in Fig. 3. Figure 4 shows how the global distribution coefficient, D, =Cq,/CC,, increases with the ph up to a maximum at around ph=4.2. Similar variation was found with synthetic solutions (61. From the mass balance extended over the considered counterions a linearized expression is obtained by using the global distribution coefficient (Eq. 1 ): _-- Here L/S represents the liquid/resin ratio, q, the amount of metal i loaded, C, the solution concentration of metal i and the suffix "0" denotes the initial values. As is shown in Fig. 5, from the straight slopes, this global coefficient remains constant in the resulting ph range from 3 to 4.5 and it is smaller for ph values lower than two. For a given solution 0, has a particular value depending on the metals considered. We find these simple expressions useful for treatment of the industrial behavior of resin-solution contactors and for dealing with very complex solutions, although the influence of variables like salinity or ph on the maximum available capacity must be additionally considered. In a more theoretical way, equations such as the Gibbs-Donnan Eq. 2 have been proposed to describe the equilibrium according to membrane diffusion considerations. An even more complex problem appears when the ion-pair association, distribution of active groups, complex formation, hydrolysis of' the acid-basic functional groups and the electrolyte sorption need to be considered when working with a multicomponent system. I L \ LOO LIS. L/ kg dry resin Fig. 5. Graphical determination of the global distribution coefl'icient.

9 L RECOVERY OF METALS FROM WASTE WATER 313 This expression gives the selectivity coefficient as a function of the physical and chemical characteristics of the ion exchanger and enclosed solution, yi being the activity coefficient in the bulk solution and yi in the resin. xi is the swelling pressure (atm), zi means electrochemical valencies and ui is the partial volume for each species (1 mol- ' ). For treatment of complex streams using chelating resins in particular, the introduction of a simplified expression of the equilibrium constant type might be necessary. The ion-pair association between the chelating group and metal ions can be considered to have the most important effect, i.e. chelation and hydrolysis, the physical ones being negligible. The corresponding conditional equilibrium constant K, is a function of the ligand and characteristics of the ion exchanger and of the liquid composition. Here Qu,, is an apparent experimental capacity (mol kg- ' dry resin), n is the observed stoichiometry. C, is the equilibrium concentration for counterion i (mol 1-I) and 9, is the amount retained on the resin (mol kg-i dry resin ). Several effects must be considered to achieve a good correlation of the experimental results using these kinds of expressions. First, the hydrolysis of the functional group must be evaluated. This can be done in terms of the total apparent capacity as a function of the solution ph. Although we have a high chloride concentration in solution, this is not enough [7] to promote stability and sorption of chloro-complexes with the metals (see Table 3). Iminodiacetic type ion exchangers have also, besides the group (R-N(-CH,- COOH),, the ligands glycine (R-NH-CH2-COOH) and amine [8,9]. The resin for which the experimental work has been carried out has an oxygen/nitrogen ratio of This means that % of ligands are iminodiacetic, deduced from the mass balance extended over nitrogen and oxygen. If a value of n is supposed (Eq. 4), the conditional equilibrium constant can be calculated plotting Z9, vs. D,"". z9, =-!!E-; Q n Dll" nk, where D, is the distribution coefficient for the species i calculated as 9,/C,. The correlation coefficient increases with the value of n until n=2. From the ordinate D, =0, Qap can be obtained, which is the same for the metals measured, in the range of ph for each value of n, but increasing from 0.94 to 2.65 for n= 1-2. The load of iron ranges between 0.8 and 0.98 mol kg- for different L/ S ratios and ph between 3.5 and 4.3. If we consider the experimental value for (4 1

10 , 374 F. MIJANCOS AND M. DIAL acid-base capacity, that is 5.37 mol kg-' dry resin, taking the value of 2 as stoichiometric coefficient for the chelating reaction, the maximum available capacity would be 2.68 which agrees perfectly with the result from Eq. 4 applied for chelating experiments (Table 5). A stoichiometry degree smaller than 2 could be justified due to the existence of 1 : 1 and 1 : 2 chelates both for iminodiacetic and glycine ligands [ 10,111. The Ki values change with the value of n. Thus, for 1.8 and 2.0 (graphical calculation for Cu, shown in Fig. 6), we obtain the results of Table 5. These conditional equilibrium constants follow analogous tendencies to the chelate stability constants for the ligands involved in this type of resin, as reported by Eger et al. [lo] in the resin phase and Sillen and Martell [ll] for solutions. The regression coefficient obtained with cobalt retention results is worse because cobalt is present at greater dilution and has less affinity for the resin. TABLE 5 Calculated values of Ki and Q., by Eq. 4 for two different stoichiometries Stoichiometry App. capacity Conditional Equilibrium Constant (mol kg- ' dry resin ) KCO KN, KC,, Kz,, LO Dc,. L 1 kg dry rrsin Fig. 6. Graphical determination of Kc, and fitting Eq. 4 for n = 2.

11 -.,..... RECOVERS OF ME I ALS FROM WASTE WATER CONCLUSION I Y From the functional groups of different commercial resins tested, an iminodiacetic type resin has shown very good selectivity properties which could be applied for the recovery of valuable metals from hydrometallurgical waste waters obtained from complex sulfide treatment. The results achieved with the iminodiacetic resin have been correlated using a global distribution coefficient, which is strongly dependent on the composition and the ph of the solution. A conditional equilibrium constant for each metal and the maximum available capacity has been calculated with this industrial waste water effluent. REFERENCES B I Arbiter, N. and Kellog, H.H., What the future holds for hydrometallurgy. Eng. Min. J., 152( 7 ): Streat, M., Application of ion exchange in hydrometallurgy. NATO AS1 Ser. E, 107: Rosato, L., Harris, G.B. and Stanley, R.W., Separation of nickel from cobalt sulfate medium by ion exchange. J. Hydrometal.. 13: Nachod, F. and Schubert, J., Ion Exchange Technology. Academic Press, New York, N.Y., pp Kolonina, N.P. and Leshch, I.Y., Behaviour of cobalt and nickel during their ion exchange by a carboxyl cation exchanger from ammonium sulfate and ammonium carbonate solutions. Issled. Ins. Gipronikel, 42: Leyden, D.E. and Underwood, A.L., Equilibrium studies with the chelating ion exchange resin Dowex A-1. J. Phys. Chem., 68(8): Birney, D.B., et al., Adsorption of chloro-complexes of the first row transition elements by Dowex A-1. Talanta, ls(6): Hering, R Chelillbildende lonenaustauscher. Akademie Verlag, Berlin, p Loewenschuss. H. and Schmuckler, G., Chelating properties of the chelating ion exchanger Dowex A-I. Talanta, 11 (10): Eger. L., Anspach, W.M. and Marinski, J.A., The Coordination Behavior of co, Ni, cu and Zn in a Chelating Ion Exchange Resin. J. Inorg. Nucl. Chem., 30: Sillen, L.C. and Martell, A.E Stability Constants of Metals-ion Complexes. Chemical Society, London, p. 377,426,583,632.