1422
Ind. Eng. Chem. Res. 2001, 40, 1422-1433
Extraction of Cadmium from Phosphoric Acid Using Resins Impregnated with Organophosphorus Extractants L. Hinojosa Reyes,† I. Saucedo Medina,† R. Navarro Mendoza,*,† J. Revilla Va´ zquez,‡ M. Avila Rodrı´guez,† and E. Guibal§ Universidad de Guanajuato, Instituto de Investigaciones Cientı´ficas, Cerro de la Venada S/N, Pueblito de Rocha, Guanajuato, Gto., C.P. 36040, Me´ xico, Centro de Investigacio´ n y Desarrollo Tecnolo´ gico S.A. de C.V., Avenida de los Sauces 3-A, Parque industrial Lerma, Edo. Me´ xico, C.P. 52000, Me´ xico, and Ecole des Mines d'Ale` s, Laboratoire Ge´ nie de l'Environnement Industriel, 6 avenue de Clavie` res, F-30319 Ale` s Cedex, France
In the first part of this work, Amberlite XAD-7, impregnated with Cyanex 301, is selected among several supports and organophosphorus extractants for its high affinity for cadmium in phosphoric acid solutions. Initially, the work focuses on the study of cadmium extraction and stripping, in both batch and column systems, using synthetic solutions. This extraction system can be efficiently used for the recovery of cadmium from concentrated phosphoric acid solutions (up to 12 M). Cadmium can be removed from loaded resins using hydrochloric acid (5 M), and the resin can be reused without additional treatment. The second part of the study focuses on the recovery of cadmium from industrial phosphoric acid solutions. The presence of iron(III) and copper(II) in large concentrations has a significantly negative effect on extraction properties. A pretreatment consisting in the reduction of ferric and cupric ions using Na2S2O4 or iron powder was used. The extraction efficiency was thereby increased, but it remained lower than that obtained with synthetic solutions. Recycling was also strongly limited: the presence of other competitor ions can explain this decrease in extraction efficiency. Introduction Several processes exist for the production of phosphoric acid:1 (i) a dry process for which the phosphorus element is burned in an oven under controlled conditions and (ii) a wet process that includes a chemical reaction of sulfuric acid with phosphorus minerals. Although the dry process results in the production of pure phosphoric acid, because of its high cost, the process is reserved for the production of low volumes of pure material. For large-scale production of phosphoric acid, the wet process is preferred, but a purification step is necessary, as phosphorus minerals usually contain organic and many mineral impurities (such as Mg, F, Si, Al, Ca, Mg, U, Fe, Zn, and Cd). Phosphoric acid can be used in food and agriculture industries, but it should be treated before use in the production of food products because of the strong toxicity of cadmium (cancerous and mutagenic effects). Although cadmium can easily be precipitated from solutions through neutralization processes, this technique can not be used for the elimination of cadmium from phosphoric acid if the acidity of the solution is to be maintained. Various processes have been investigated for the purification of phosphoric acid2,3 or for cadmium extraction from acidic solutions;4,5 one of the most promising is liquid/liquid extraction using selected solvent/extractant systems.2-5 However, solvent extraction is frequently limited to industrial applications because of several economical, * To whom correspondence should be addressed. E-Mail:
[email protected]. Fax: +52 473 275 55. † Universidad de Guanajuato. ‡ Centro de Investigacio ´ n y Desarrollo Tecnolo´gico S.A. de C.V. § Ecole des Mines d’Ale ` s.
technical, and environmental constraints, including extractant dispersion, loss of solvent, and pollution impact.6 To improve the applicability of solvent extraction systems, several impregnation techniques have been developed, including resin impregnation,7,8 the use of liquid membrane systems,9 and more recently, hollow fiber systems.10,11 The impregnation of resins with extractants makes it possible to convert the support into a chelating system12-14 but with less expensive materials. This study presents the investigation of cadmium recovery using extractant-impregnated resins (EIRs). Two different commercial resins have been investigated (Amberlite XAD-2 and XAD-7) because of their high macroporosities, hydrophobicities, and efficiencies for solvent impregnation.15-17 A large number of extractant/ solvent systems have been studied for metal ion extraction.18 Organophosphorus extractants have been selected, including commercial products such as Cyanex 272, Cyanex 301, Cyanex 302, and PC-88A. Preliminary work in the current study focuses on the selection of the optimum extractant system through the optimization of liquid/liquid extraction and impregnated resin stability. The optimization of the extraction process is easily obtained in the liquid/liquid system and can be readily extrapolated to impregnated systems. Amberlite XAD-2 and XAD-7 impregnated with Cyanex 301 are compared for cadmium recovery in both batch and fixedbed column systems using synthetic phosphoric acid solutions. The stripping is also investigated using hydrochloric acid solutions (among several eluents). Finally, selected optimum conditions are used for the treatment of an industrial phosphoric acid solution. In some cases, a prereduction of the interfering metals is
10.1021/ie0005349 CCC: $20.00 © 2001 American Chemical Society Published on Web 02/03/2001
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1423
required, and this study presents a first attempt to optimize this pretreatment. Material and Methods Materials. Amberlite XAD-2 and XAD-7 were purchased from Sigma-Aldrich (St. Louis, MO). Amberlite XAD-2 is commercially characterized as a nonpolar macroporous styrene/divinyl benzene polymer, supplied as spherical particles (20-60 mesh) with a mean pore size of approximately 90 Å and a superficial area of about 330 m2 g-1. XAD-7 is characterized as a moderately polar polymer of acrylic ester, with a particle size and a mean pore size similar to those of XAD-2 and a superficial area of about 450 m2 g-1.19 Cyanex 272, Cyanex 301, and Cyanex 302 were kindly donated by the Cytec Company (West Patterson, NJ), and PC-88A by Dailachi Chemical Industry Co. (Japan). Cyanex 272 is a phosphinic acid group extractant [di-(2,4,4-trimethylpentyl) phosphinic acid], Cyanex 301 and Cyanex 302 are alkylthiophosphinic extractants [bis(2,4,4-trimethylpentyl) dithiophosphinic acid and bis(2,4,4-trimethylpentyl) thiophosphinic acid, respectively], and PC-88A is a phosphonic acid group extractant (ester 2-ethylhexyl phosphonic acid). Cadmium solutions were prepared by dilution of Titrisol Merck standards, and other analytical-grade reagents were supplied by Fisher (H3PO4), Merck [Oxalic acid, Cu(NO3)2], Analitika de Me´xico, S.A. (HNO3), Quimir S.A. de C.V. (industrial phosphoric acid), General Products Company S.A. (Na2S2O4, A grade, 85%), and J. T. Baker (ethanol, EDTA, HCl, and Fe and Zn powder). An atomic absorption spectrophotometer (AAS, Perkin-Elmer 3110) was used to determine the concentration of cadmium (λmax ) 228.8 nm, slit width ) 0.7), iron (λmax ) 248.3 nm, slit width ) 0.2), and copper (λmax ) 324.8 nm, slit width ) 0.7). Impregnation Procedure. Two different techniques were used for the impregnation of the resins: the equilibrium impregnation of the resin (Method I) and the total stripping impregnation method (Method II). Method I included the following preconditioning of the resin: (a) the resin was mixed with a 4 M HCl solution in a 1:1 (v/v) mixture of methanol and water to remove organic impurities and the monomer; (b) a further rinsing step was performed with water to remove HCl; and finally, (c) the resin was dried at 40 °C overnight and sieved to recover the 20-30 mesh fraction that was used for further experiments. The impregnation of the resin was performed using the technique described by Cortina et al.20 and modified by Navarro Mendoza et al. as follows:8 Resin (100 g) was mixed with 1 L of a 75% (v/v) water/ethanol solution containing the extractant (0.025 M). The mixture was maintained in agitation for 24 h and then centrifuged to remove the excess extractant in the capillary network. The resin was then put into contact with HCl (4 M) for 24 h in order to decrease the hydrophobicity of the resin, and subsequently, it was rinsed several times. The resin was stored in water. Method II also included preconditioning of the resin as follows: the resin was put into contact with acetone for 12 h, filtered, and rinsed successively with nitric acid 0.1 M and acetone, before being dried in a vacuum. The total solvent stripping procedures were adapted from Yoshito et al.21 and consist of the following steps: Resin (80 g) was put into contact with 240 mL of acetone for 12 h. Next, 240 mL of acetone containing the extractant (0.26 M) was added to the
slurry, and the mixture was agitated for 24 h. Finally, acetone was evaporated in a vacuum. The concentration of the extractant in the aqueous phase was determined by alkaline potentiometric titration using a Titrino 716 automatic titrator (Methrom, Germany). The extractant concentration in the resin was measured by potentiometric titration after triplicate stripping with ethanol (for 12 h). The amount of water retained in the polymer was measured by weight loss at 50 °C for 24 h. Liquid/Liquid Extraction. For solvent extraction studies, the following procedure was used: Extractant (10 mL, 0.1 M concentration in kerosene or xylene) was put into contact with 10 mL of the aqueous solution (metal ion initial concentration ) 1 mM, with Na2SO4 to adjust the ionic strength at a concentration of 0.5 M) with a reciprocal shaker (model 51502, Cole-Parmer, Vernon Hills, IL) for 1 h (time needed to reach equilibrium). The residual metal concentration in the aqueous phase was determined by AAS. The amount of metal extracted was obtained by a mass balance between the organic and aqueous phases. Solid/Liquid Extraction. Impregnated resin (0.5 g dry weight) was put into contact with 50 mL of a 10 mg L-1 cadmium solution in nitric acid or phosphoric acid (at different concentrations) for 24 h using a reciprocal shaker (150 rpm) at 25 °C; the temperature was maintained with a controlled temperature chamber (Terlab DBO model, Mexico) or a temperature-controlled recirculator (model 1166, VWR Scientific, West Chester, PA). At equilibrium, the residual concentration was determined, and the concentration in the solid phase was obtained using the mass balance equation. Experiments were performed with Amberlite XAD-4 and XAD-7 without extractant impregnation under similar experimental conditions, and they showed that cadmium was not sorbed on the resins. The influence of the phosphoric acid concentration was studied by mixing 50 mg of impregnated sorbent with 50 mL of cadmium solution (10 mg Cd L-1) for 24 h. Samples for the measurement of extraction isotherms were prepared by mixing 500 mg of EIR with 50 mL of cadmium solution with increasing concentrations of cadmium at different concentrations of phosphoric acid (3.2 or 12.7 M) for 24 h. Samples for extraction kinetics measurements were prepared by mixing 2.5 g of EIR with 250 mL of cadmium solution (10 or 45 mg L-1) in phosphric acid solutions (3.2 or 12.7 M). For the study of cadmium stripping from loaded EIRs, 500 mg of EIR was mixed, in two successive steps, with 50 mL of different eluents: (1) HCl, 5 M; (2) Cu(NO3)2, 0.1 M at pH 2; (3) ethanol; (4) EDTA, 0.1 M at pH 8; and (5) HNO3, 3 M. Resins were initially loaded with cadmium (1.4-1.7 mg of Cd g-1). Extraction Kinetics Modeling. Overall extraction kinetics can be controlled by several mechanisms, including external diffusion, intraparticle diffusion, and chemical reaction.15,22 Although some examples of ionexchange processes controlled by reaction rate exist,23 extraction mechanisms are most frequently controlled by film diffusion and/or intraparticle diffusion. Simplified models exist, such as the homogeneous diffusion model (HDM) and the shrinking core model (SCM). These models have been unsuccessfully used for the modeling of experimental data. Thus, extraction kinetics have been modeled under finite solution volume conditions using simplifying hypotheses for the separation of the external and intraparticle diffusion steps. Mass
1424
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001
transfer is modeled by several relationships that take into account the diffusion mechanisms and their related equations and the coupling between the liquid and solid phases, as well as the initial and boundary conditions. The solution generally requires a numerical analysis. However, at the beginning of contact time (between 0 and 15-20 min), the system can be simplified by assuming that the metal concentration at the sorbent surface tends to zero (pure external diffusion) and that the intraparticle mass transfer resistance is negligible. Thus, for external diffusion, the solution is given by
ln
[ ]
C(t) A ) -kf t Co V
(1)
The initial external mass transfer coefficient, kf (m min-1), can be obtained from the slope of a plot of ln[C(t)/Co] versus t (min), where C(t) is the solution concentration; Co, the initial sorbate concentration (mg L-1); A, the sorbent exchange surface area (m2); and V, the volume of the solution (L).24 This solution for an external diffusion model cannot be extended beyond the initial minutes of contact. However, this equation, which is mathematically equivalent to a first-order kinetic equation [d(C(t))/dt ) kC(t)], has been used here for the modeling of a larger range of contact times. Crank proposed a model whereby diffusion is controlled only by intraparticle mass transfer for a wellstirred solution of limited volume (V), assuming the solute concentration to be constantly uniform (initially Co) and the sorbent sphere to be free of solute.25 Under these conditions, the total amount of solute Mt (mg g-1) in a spherical particle after time t, expressed as a fraction of the corresponding quantity after infinite time (M∞, mg g-1), is given by
Mt M∞
)1-
2
∞
6R(R + 1) exp(-Dqn t/d2p)
n)1
9 + 9R + q2nR2
∑
(2)
where D is the intraparticle diffusion coefficient (m2 min-1). The fractional approach to equilibrium, FATE, can be used to estimate the intraparticle diffusion coefficient D, when the external diffusion coefficient is neglected. R is the effective volume ratio, expressed as a function of the equilibrium partition coefficient (solid/ liquid concentration ratio), and is obtained by the ratio Ceq/(Co - Ceq). qn represents the nonzero solutions of the equations
tan qn )
3qn 3 + Rqn
2
and
M∞ 1 ) VCo 1 + R
(3)
The infinite sum terms are added until the summation does not vary. In this study, eq 2 was used to determine the overall intraparticle diffusivity that best fit experimental data (minimizing the sum of the square of the differences between experimental results and the calculated data). Extraction in Column Systems. For experiments performed on synthetic solutions, the columns (internal diameter ) 7 mm) were filled with Amberlite XAD-7/ Cyanex 301 (1.4 g dry weight, column depth ) 9.8 cm) and fed with a Masterflex 7253-30 peristaltic pump (Cole-Parmer) using phosphoric acid (3.2 M) containing cadmium (initial concentration ) 10 mg L-1). The flow
rate was fixed at 54 mL h-1 (superficial velocity ≈ 1.4 m h-1). Samples were collected every 8 min using a Gilson 203 automatic sampler. Experiments were conducted at 25 °C in a controlled-temperature chamber (Terlab DBO model). The resins were preconditioned with phosphoric acid. Stripping was performed with HCl (5 M) using the same equipment (flow rate ) 78 mL h-1, superficial velocity ) 2 m h-1), after a rinsing step (using 50 mL of 3.2 M phosphoric acid). Samples were collected every 2 min. In the case of industrial phosphoric acid, the same experimental procedure was used, and the experimental conditions were as follows: cadmium initial concentration ) 11.2 mg L-1, phosphoric acid concentration ) 3.2 M. The solutions were pretreated with Na2S2O4. The column was filled with 1.2 g of resin, and the solution flow rate was fixed at 48 mL h-1 (column depth ) 8.4 cm; superficial velocity ) 0.6 m h-1); the stripping was performed with HCl (5 M) at the same flow rate. Results and Discussion Selection of the Extractant System. Liquid/ Liquid Extraction. Preliminary experiments (not shown) were focused on the study of the influence of the equilibrium pH on cadmium extraction in sulfate solutions for several extractant systems. These preliminary results were used for the selection of the best extraction systems. Although the experimental conditions were not exactly the same for all of these experimental runs, it appeared clear that cadmium extraction was almost identical for Cyanex 272 (in xylene) and for PC-88A (in kerosene), i.e., extraction was negligible up to pH 3, and above pH 3, the extraction efficiency sharply increased. For Cyanex 302 (in xylene), the same shape extraction curve was obtained; but the extraction began to be significant for lower pH. Specifically, above pH 0.5, extraction significantly increased, and total extraction was achieved at pH 3, whereas it was necessary to increase the pH to 5 to achieve a complete recovery with Cyanex 272 and PC-88A. The case of Cyanex 301 (in xylene) was completely different, as regardless of the pH, the extraction was complete throughout the pH range investigated. The extractants can be ranked in the following order regarding cadmium removal: PC88A < Cyanex 272 , Cyanex 302 , Cyanex 301. This ranking was established according to the pH50, which represents the pH at 50% extraction, with these values being ∼4, 3.8, 1.8, and Cyanex 272 (6.37) > Cyanex 302 (5.63) . Cyanex 301 (2.1). The extractants are similar in structure, with the differences being related to the atoms that are electron donors. PC-88A and Cyanex 272 are similar, being characterized by the presence of two oxygen donors, whereas in Cyanex 302, the electron donors are constituted by S and O, and in Cyanex 301, each of the electron donors is S. These differences can explain the differences in acid/base properties, which, in turn, influence extraction behavior. For weak acid extractants, such as Cyanex 272 and PC88-A, it is necessary to increase the pH to reach a significant removal efficiency. Additionally, the different affinities can be explained by the Pearson theory of stability of metal complexes: hard acids react preferentially with hard bases, whereas soft acids react with soft bases.27 Cadmium can be classified as a weak acid (softness )
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1425
+0.58), with a large size, a low charge, and a high polarizability.28 As a consequence, cadmium will react preferentially with weak bases. In the case of the selected extractants, the electron donor atom controls the softness of the base: the extractants with O groups are harder than those with S groups. It appears that, in acidic solutions, such as those encountered in high concentration phosphoric acid solutions, it would be better to use the softest base, Cyanex 301, for the extraction of cadmium. Experimental results are in good agreement with theoretical behavior. As a result of this preliminary study, PC-88A and Cyanex 272 were removed from the panel of extractants for further experiments. Optimization of the Impregnation of Amberlite XAD-2 and XAD-7. Both Method I and Method II were tested with Amberlite XAD-2 and XAD-7 using Cyanex 301 as the impregnant. The amount of extractant immobilized in the resin depends on both the support and the impregnation method. With Amberlite XAD-2, the impregnation with Method II resulted in a very hydrophobic material whose hydration was very slow, and this modified support did not appear to be efficient for Cd uptake. Better results were obtained with Method I: the impregnation of the resin led to a Cyanex 301 concentration of 0.24 mol kg-1 in the polymer. The amount of water retained in the polymer was about 42.8%. For Amberlite XAD-7, the best impregnation was obtained with Method II, and in this case, the equilibrium concentration in the solid was approximately 0.43 mol kg-1. Preliminary experiments performed with Amberlite XAD-7, impregnated using Method I, showed that the extraction kinetics were very slow, and Method II was preferred for the preparation of the sorbent. Amberlite XAD-2 is a nonpolar resin; after being dried, it is difficult to rehydrate, which might influence the slow sorption of cadmium. In contrast, in the case of Amberlite XAD-7, the rehydration of the resin was fast, and as a consequence, the sorption kinetics were enhanced. Method II appears to be better, perhaps because of the greater amount of extractant that has been adsorbed in the resin, as increasing the extractant loading increases the maximum sorption capacity of the EIR, which, in turn, decreases the limitation due to extractant saturation within the resin. Cadmium Extraction on EIR in Batch Systems. Choice of the Extractant Impregnated Resin System. The behavior of the extractant can be modified after impregnation, so it was necessary to determine whether the order found in the liquid/liquid extraction is maintained after immobilization of the extractant on the resin. In a preliminary study, resins were impregnated with Cyanex 301, Cyanex 302, and Cyanex 272 in either nitric acid or phosphoric acid solutions, and the EIR resins were tested for cadmium extraction. Cyanex 272 exhibited a trend similar to that observed in a liquid/ liquid system in that the removal efficiency curve was slightly shifted (by about 1 pH unit) to highest pH and the pH50 was shifted from ca. 4 to ca. 5, values that are comparable to the pH50 values for cadmium extraction by DEHPA-TOPO-impregnated Amberlite XAD-229 and Lewatit TP807′84 resins.30 However, for acid concentrations above 0.1 mM, cadmium was not significantly extracted from the solution. More complete studies have been continued with the other extractants. Figure 1 illustrates the influence of nitric acid concentration on the cadmium extraction with resins impregnated with
Figure 1. Cd extraction using Amberlite XAD-2 (open symbols) and XAD-7 (solid symbols) resins impregnated with Cyanex 301 (0/9, phosphoric acid; 4, nitric acid) and Cyanex 302 (O, nitric acid). [Cd]o ) 10 mg L-1, EIR dosage ) 10 g L-1.
Cyanex 301 and Cyanex 302. The impregnation did not modify the efficiency ranking, although the efficiency of Cyanex 302 increased after impregnation (pH50 ) 0). For Cyanex 301, cadmium extraction was complete when the nitric acid concentration was below 1 M and decreased above that concentration, but the extraction efficiency exceeded 80% even for 4 M nitric acid. Cyanex 301 appears to be the most efficient extractant for the recovery of cadmium on impregnated Amberlite XAD-2 resin, especially for strong acidic media, such as those encountered in the production of phosphoric acid using the wet process. Kabay et al. report that cadmium does not form complexes with phosphoric acid.19 Thus, the extraction is related to the formation of complexes between cadmium and the impregnated extractants immobilized on the resin according to the reaction:30-31
Cd2+ + HLr T CdL2,r+ H+ Increasing the concentration of the acid displaces the reaction to the left side, according to (a) the acidic characteristics of the extractant (pKa) following the trends observed with liquid/liquid systems and (b) the interaction of the extractant with the resin (influence of hydrophobicity). At high concentration the extraction performance is significantly lower for nitric acid (strong acid) than for phosphoric acid (Figure 1). Influence of Phosphoric Acid Concentration. Figure 1 also presents the influence of increasing phosphoric acid concentration on cadmium extraction for both resins impregnated with Cyanex 301. Whereas the extraction efficiency was independent of acid concentration for impregnated Amberlite XAD-7, a significant decrease in cadmium removal was observed above a phosphoric acid concentration of 4-5 M for Amberlite XAD-2. For the treatment of industrial phosphoric acid, for which the acid concentration is about 12.7 M, it will be necessary to dilute the solution when the Amberlite XAD-2/Cyanex 301 systems are used or to select Amberlite XAD-7 as the support for Cyanex 301. The differences between the two systems can be explained by the facts that (i) when immobilized on Amberlite XAD-7, the Cyanex 301 concentration is about 80% greater than in the case of Amberlite XAD-2 and (ii) the polarity of the support can change the accessibility of the extraction sites to cadmium. Kabay et al. also investigated the influence of the phosphoric acid concentration on cadmium sorption using several chelating
1426
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 Table 1. Extraction Isotherm Modeling Using the Langmuir and the Freundlich Equations (ref. Figure 2) Langmuir
Freundlich
qm b MSRb (mg g-1) (L mg-1) XAD-2 (3.2 M) XAD-7 (12.7 M)
1.56 11.52
24.1 1.3
kFa
n
MSRb
0.126 1.20 11.6 0.151 1.83 6.24 6.62 1.51
a k (g-1 L1/n mg1-1/n). b MSR ) mean square deviation calcuF lated using
[
n
∑[(X
Figure 2. Extraction isotherm for Amberlite resins impregnated with Cyanex 301. For XAD-2, [H3PO4] ) 3.2 M; for XAD-7, [H3PO4] ) 12.7 M. Solid lines, Langmuir model; dashed line, Freundlich model.
resins containing phosphonic or diphosphonic groups and a strong acid ion-exchanger resin [Dowex 50W(X8)].12 They observed a strong decrease in cadmium removal as the phosphoric acid concentration increased from 0.1 to 1 M, while for higher acid concentrations, the sorption capacity was maintained at a very low level.12 They concluded that there was a competitor effect due to the proton but that acid speciation did not interfere with extraction. They also compared the influence of the phosphoric acid concentration on cadmium extraction using several Amberlite resins (XAD-2, XAD-4, XAD-7, and XAD-8) impregnated with Cyanex 302. They observed a decrease in cadmium extraction with increasing phosphoric acid concentrations; however, the decrease in extraction was greater with the most hydrophobic resins (XAD-2 and XAD-4).19 This is consistent with the present results. Extraction Isotherm. The phosphoric acid concentration was fixed at 3.2 and 12.7 M for the experiments performed with Amberlite XAD-2 and XAD-7, respectively, which are the maximum acid concentrations for which cadmium sorption remains efficient. Figure 2 presents the extraction isotherms obtained under selected experimental conditions. Despite the greatest possible acidity selected for the Amberlite XAD-7/ Cyanex 301 system, the maximum extraction capacity was significantly greater (about 6-7 times) than that obtained with the XAD-2/Cyanex 301 system. The initial slope of the curve, which marks the affinity of the sorbent for the solute, is comparable for the two systems. Therefore, large differences in maximum extraction capacity cannot be attributed solely to the differences in the extractant content in the support or to the acid concentration; rather, the structure of the sorbent might participate in the restriction or enhancement of extraction performance levels. The sorption capacities of the sorbents studied by Kabay et al. were ranked in the following order: Dowex 50W(X8) > Diphonix > Actinide-CU . Diaion-CRP200 . RSPO. The values ranged between 5 and 50 mg g-1 at a 0.1 M phosphoric acid concentration, whereas at a 3 M phosphoric acid concentration, the sorption capacities decreased from 8 to 0 mg g-1.12 These two experiments cannot be directly compared, as the experimental conditions differed. However, the maximum extraction capacity was on the same order of magnitude for Amberlite XAD-7 as for Dowex 50W(X8), although the concentration of phosphoric acid was tripled. This confirms the
]
1/2
exp,i
- Xcalc,i)2]
i)1
n
interest in Cyanex 301-impregnated resins for Cd extraction. Kabay et al. also reported greater efficiency in cadmium extraction for Amberlite XAD-7 impregnated with Cyanex 302 than for Amberlite XAD-2.19 Depending on the concentration of phosphoric acid (in the range of 0.1-0.5 M), they found a cadmium sorption capacity ranging between 27 and 34 mg of Cd g-1. Figure 2 shows that Langmuir’s mathematical equation accurately describes the experimental data for cadmium extraction on Amberlite XAD-2 impregnated with Cyanex 301, whereas for Amberlite XAD-7, this equation introduces many discrepancies in the simulation of sorption capacities at low residual cadmium concentrations (Table 1). The maximum sorption capacity tends to 11.6 mg g-1. The Freundlich model fits the experimental data better at a low residual concentration; however, it is not suitable for describing the asymptotic trend observed at high concentrations. Extraction Kinetics. Figure 3 shows the extraction kinetics results for selected experimental conditions. This figure clearly demonstrates that cadmium extraction was significantly faster for Amberlite XAD-7 than for Amberlite XAD-2 impregnated with Cyanex 301 (top curves). The commercial description of Amberlite resins indicates that the pore size is identical for the two supports and that the superficial area is greater for Amberlite XAD-7 (450 m2 g-1) than for XAD-2 (330 m2 g-1). The most significant difference is due to the polymer structure and the polarity of these resins. This difference might influence their hydrophilic/hydrophobic properties, which, in turn, limits the accessibility of internal sites. It is difficult to really compare experimental data because of the excess of sorbent versus the amount of cadmium actually present in the solution. In the case of the Amberlite XAD-2/Cyanex 301 system, at equilibrium, the extraction capacity was about 1 mg g-1 (about 62% of the maximum extraction capacity, but close to the concentration observed in sorption isotherms for a low residual cadmium concentration). For the Amberlite XAD-7/Cyanex 301 system, a 1 mg g-1 extraction capacity represents only 9% of the maximum capacity. Thus, there is a significant difference in the saturation level of the impregnated resins. Figure 3 also shows the influence of the H3PO4 concentration on the extraction kinetics for Amberlite XAD-7 impregnated with Cyanex 301 (right curves). An increase in the acid concentration involved a slower extraction, as it took 4 h for complete extraction of cadmium, whereas 45 min were sufficient when the concentration of H3PO4 was 3.2 M. This difference can be explained by the increase in the viscosity of the acid solutions with concentration. The bottom curves in Figure 3 compare the extraction
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1427
Figure 3. Extraction kinetics and mass transfer modeling using external diffusion and intraparticle diffusion models. Finite volume solution, solid lines, external diffusion and first-order kinetic models; dashed lines, intraparticle diffusion model. EIR dosage ) 10 g L-1. Table 2. Modeling of Extraction Kinetics Using Eqs 1-3 first-order kinetic equation support
[H3PO4] (M)
[Cd]o (mg L-1)
XAD-2 XAD-7 XAD-7 XAD-7
3.2 3.2 12.7 12.7
10 10 10 45
slope × 103
ordinate intercept
-4.36 -92.2 -15.4 -2.67
-0.177 0.024 -0.061 -0.174
kinetics for different initial concentrations of Cd(II) for Amberlite XAD-7 impregnated with Cyanex 301 in phosphoric acid solutions (12.7 M). An increase in the metal concentration resulted in an increase in the time necessary to achieve the complete extraction of cadmium. With a 45 mg L-1 concentration of cadmium (the actual concentration in industrial phosphoric acid solutions), the contact time required to reach equilibrium might be greater than 24 h. This value is comparable to that obtained with Amberlite XAD-2 with a 10 mg L-1 cadmium solution ([H3PO4] ) 3.2 M). Under these conditions, the extraction capacity was about 4-4.5 mg g-1 (about 35-40% of the maximum sorption capacity). The extraction kinetics are significantly controlled by the saturation level of the extractant immobilized in the resin. This finding confirms the previous conclusions on the selection of the impregnation method. In Figure 3, the extraction kinetics have been modeled with the intraparticle diffusion equation (eq 2) and with the first-order kinetic equation (eq 1) (Table 2). Under selected experimental conditions, the residual concentration at equilibrium tends to zero; only nonzero points have been used for modeling. It appears that the intraparticle diffusion model fits the experimental data
intraparticle diffusion MSR
D × 1012 (m2 s-1)
MSR
0.051 0.020 0.055 0.211
0.84 3.79 0.72 0.18
0.042 0.106 0.060 0.033
well when the extraction kinetics are slow (unfavorable conditions corresponding to a high Cd concentration with Amberlite XAD-2), whereas for fast kinetics (Amberlite XAD-7 and a low Cd concentration), the firstorder kinetic equation gives a better fit (Table 2). The cadmium molecular diffusivity in water is 7.19 × 10-10 m2 s-1.28 This is 2 or 3 orders of magnitude greater than the diffusion coefficient found in the impregnated resins for which the intraparticle diffusion coefficient varied in the range (0.26-5.4) × 10-12 m2 s-1. For comparable experimental conditions ([Cd]o ) 10 mg L-1, [H3PO4] ) 3.2 M), the Cd intraparticle diffusivity in Amberlite XAD-7 is about 5 times greater than that in Amberlite XAD-2. Using D2EHPA [di(2-ethylhexyl)-phosphoric acid] impregnated Amberlite resins, Juang and Lin obtained Cu diffusivities in the range (1.1-2.59) × 10-12 m2 s-1 with XAD-2 and (1.42-2.19) × 10-12 m2 s-1 with XAD4.15 For Zn extraction, the coefficients were comparable with XAD-2 [(1.6-3.12) × 10-12 m2 s-1] but strongly decreased with XAD-4 [(1.33-2.65) × 10-13 m2 s-1].15 The values were similar for each support despite the greater nominal pore diameter of XAD-2 (9 nm instead of 5 nm for XAD-4) in the case of Cu extraction. They
1428
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001
explained this trend by the effect of the structure of the resins (agglomerate of microspherical gel particles) and by the existence of a dual diffusion mechanism in macropores and microspheres. This dual diffusion reduces the impact of the pore size parameter. In the case of Zn, they explained the greater sorption rate with impregnated Amberlite XAD-2 by the steric hindrance caused by the large size of the tetrahedral structure of the Zn-D2EHPA complex and the comparatively small pores of the XAD-4 support. In this study, the nominal pore size is identical for the two supports, and the differences can be explained by the change in the hydrophobic properties of the resins. Komiyama and Smith compared the intraparticle mass transport of benzaldehyde in two Amberlite supports characterized by different hydrophobic behaviors (XAD-7 < XAD-4) in aqueous and methanol solutions.32 They concluded that surface diffusion is the predominant transport mechanism for the strongly hydrophobic XAD-4 support and that it is of equal importance in pore-volume diffusion for the less hydrophobic Amberlite XAD-7. Here, the weaker hydrophobicity of Amberlite XAD-7 can explain the better accessibility to internal sites and the faster kinetics. This information indicates some trends in the control of sorption by diffusion mechanisms. More information would be obtained by studying the influence of particle size, metal concentration, and mixing rate of the solution. However, the main objective of this work was the design of an extractant-impregnated system for the removal of cadmium from phosphoric acid. Complementary experiments varying these experimental parameters would be needed for the final optimization of the process. Stripping of Cadmium from EIR. Various eluents were investigated for the recovery of cadmium from impregnated resins with two simultaneous objectives: complete recovery of cadmium and recycling of the impregnated resin. Figure 4a shows that, for the Amberlite XAD-2/Cyanex 301 system, the best stripping agent was HCl, as about 97% of the elution was achieved during the first stripping step. The second stripping enabled the recovery of about 1-2% more of the cadmium immobilized on the resin. The overall stripping yields were comparable with the acid solution of Cu(NO3)2 and with ethanol, but the second stripping step represented a fraction of the total stripping that became significant. For EDTA and HNO3, the stripping efficiency progressively decreased in both the first and second stripping steps. Among the best stripping agents, ethanol cannot really be used for the recycling of the resin because of the release of the extractant from the resin. The impregnated resin eluted with the acidic copper nitrate solution also cannot be used because of the strong affinity of copper for Cyanex 301, which will prevent a further extraction of cadmium in subsequent extraction/stripping cycles.18 HCl was selected as the best system, and the recycling of the resin was examined for three cycles. The extraction yield decreased progressively throughout the three cycles, being 98% at the first step, 76% at the second step, and only 57% at the third step. Amberlite XAD-2/Cyanex 301 was not a very efficient combination for the recovery of cadmium from phosphoric acid solutions, despite the great efficiency of HCl for the elution of cadmium in a single extraction/ stripping operation. HCl was used for the optimization of cadmium elution from Amberlite XAD-7/Cyanex 301: the concentration of the stripping agent was varied
Figure 4. Stripping of cadmium from loaded resins impregnated with Cyanex 301: (a) selection of the stripping agent [Amberlite XAD-2, (1) HCl, 5 M; (2) Cu(NO3)2, 0.1 M at pH 2; (3) ethanol; (4) EDTA, 0.1 M at pH 8; and (5) HNO3, 3 M]; (b) influence of hydrochloric acid concentration for Amberlite XAD-7 after loading of resins, [Cd]o ) 10 mg L-1, [H3PO4] ) 12.7 M; (c) stripping kinetics with 5 M HCl after loading of resins, [Cd]o ) 10 mg L-1, [H3PO4] ) 3.2 or 12.7 M. EIR dosage ) 10 g L-1 in two successive steps.
between 3 and 8 M for a two-step stripping procedure. Figure 4b shows that the total stripping yield increased up to the 5 M HCl concentration, while the percentage of cadmium removed during the second stripping step significantly decreased. For the highest HCl concentration, cadmium was almost completely extracted during the first step, and the amount extracted during the second step could be neglected. These results are better
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1429
than those obtained by Kabay et al. in the elution of Amberlite XAD-7/Cyanex 302 loaded with cadmium from phosphoric acid solutions for which the elution yield did not exceed 68%.19 Figure 4c presents the stripping kinetics for the Amberlite XAD-7/Cyanex 301 system after cadmium extraction in H3PO4 solutions (3.2 and 12.7 M). Stripping occurred very rapidly, as equilibrium was reached within the first 2-4 min. Recycling of the resin was investigated for 10 successive extraction/stripping cycles, using a similar procedure (extraction was performed in 3.2 M H3PO4 with a Cd content of ca. 1 mg g-1). The extraction yield remained above 98% throughout these 10 cycles, with no significant decrease in the extraction efficiency (results not shown). These results clearly show that the Amberlite XAD7/Cyanex 301 system is significantly better adapted to cadmium extraction from very acidic solutions than the Amberlite XAD-2/Cyanex 301 system. Various explanations for these differences can be proposed, including (i) a difference in the extractant content, (ii) a difference in the impregnation procedure, and (iii) a difference in the polarity of the resin. The difference in the extractant loading does not appear sufficient to explain such large differences in the extraction performances, as the maximum extraction capacity was multiplied by 7 for an increase in the extractant concentration in the resin that was only doubled. The difference in the impregnation method results in a difference in the hydrophilic/ hydrophobic properties of the resin, a characteristic that might more specifically influence the extraction kinetics. Furthermore, Amberlite XAD-7 is characterized as a moderately polar polymer because of the presence of oxygen, which gives the support a structure that is significantly more hydrophilic than that of Amberlite XAD-2. As a consequence, the hydration of the polymer matrix and the diffusion of metal ions in the porous network were enhanced, and the extraction efficiency was improved. Alexandratos et al. explain the enhancement of the sorption properties of ion-exchange resins by the grafting sulfonic acid as resulting from the improvement of ionic accessibility to internal ligands.14,33 Amberlite XAD-7/Cyanex 301 was selected for further experiments in continuous systems and for experimentation on industrial phosphoric acid. Cadmium Extraction on EIR in Column Systems with Synthetic Solutions. Figure 5 presents the breakthrough curve for cadmium extraction. The amount of cadmium removed at saturation of the column was about 15.5 mg, corresponding to an extraction capacity of 11.1 mg g-1. This value is higher than that obtained in extraction isotherms in batch systems; however, in the present case, the phosphoric concentration was lower (3.2 M instead of 12.7 M). Breakthrough occurred at 350 mL [93 BV (bed volumes)], while exhaustion was achieved at approximately 1500-2000 mL (400-530 BV). The solid line on the breakthrough curve represents the empirical modeling of the decay curve of the relative concentration using the simplified equation
C(t) 1 ) Co 1 + exp[- Rb(V - Vo)]
(4)
where Rb (mL-1) and Vo (mL-1) represent the parameters of the equation. The parameters of eq 4 can be correlated to the parameters determined using the Bohart and Adams equation.34,35 The parameters were found using nonlinear regression analysis (Mathematica
Figure 5. Cadmium extraction with Amberlite XAD-7/Cyanex 301 (breakthrough curve, top) and stripping (elution profile, bottom) in dynamic systems. Synthetic solution: [Cd]o ) 10 mg L-1, [H3PO4] ) 3.2 M, extraction flow rate ) 54 mL h-1, stripping flow rate ) 78 mL h-1, mass of resin ) 1.4 g, stripping with HCl (5 M). Industrial phosphoric acid: [Cd]o ) 11.25 mg L-1, [H3PO4] ) 3.2 M, extraction and stripping flow rates ) 48 mL h-1, mass of resin ) 1.2 g, stripping with HCl (5 M)).
software). In this case, Rb was found to be 0.01 mL-1 and Vo to be 660 mL. Vo can be approximated by the volume at half-exhaustion of the sorbent, and in the case of a perfectly symmetrical breakthrough curve, it tends to Vo ) mq(Co)/Co, where m is the mass of sorbent and q(Co) is the sorption capacity for a residual metal concentration of Co (the inlet concentration). Rb can be correlated to the slope of the breakthrough curve at the half-exhaustion point. This empirical equation fits the experimental data well (MSR ) 0.027). Figure 5 also presents the stripping curve. It seems that complete desorption was achieved after four fractions were collected (14 mL); the mean concentration of the eluted fractions was ca. 2 g L-1. The concentration factor was ca. 200 between the stripped solution and the initial phosphoric acid. Cadmium Extraction from Industrial Phosphoric Acid Using Amberlite XAD-7/Cyanex 301 in Batch Systems. The composition of the industrial phosphoric acid is presented in Table 3. The main objective of this work was the extraction of cadmium from phosphoric acid; however, because of the presence of copper and iron, known to be extractable using organophosphorus extractants,18,36 it appeared necessary to verify their removal using EIR, to determine their competitor effect, and to establish whether chemical pretreatment of the solutions can reduce this inter-
1430
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001
Table 3. Chemical Composition of Industrial Phosphoric Acida element/ compound
concentration (mg L-1)
element/ compound
concentration (mg L-1)
P2O5 SO42FAl Ca Fe Mg Cr Ni
53.91% (w/w) 0.72% (w/w) 1295 2580 301 3900 7510 486 73.8
Zn V Cu Pb Mn As Cd Sn Co
470 643 69.3 2.3 48.2 32 45 23 1
a
Analytical data from Quimir S.A. de C.V.
fering effect. Figure 6 shows the extraction efficiency as a function of the EIR dosage for cadmium, copper, and iron. Previous work on cadmium removal using an ionic flotation technique has shown that these metals interfere the most.37 Complete extraction of cadmium and copper occurred when the mass of the resin exceeded a dosage of 80 g L-1. This is consistent with results from Koopman et al. who investigated extraction of several metals through liquid/liquid extraction and ion exchange (on selected resins) and found that the distribution coefficient was about 50 times greater for copper than for cadmium using Cyanex 301.18 With a dosage of 10 g L-1 (the experimental conditions selected for synthetic solutions), the extraction efficiency did not exceed a few percent for cadmium. For iron, the amount of metal extracted was negligible compared with the initial concentration. The stripping of metals immobilized on the impregnated resin was performed with HCl (5 M); cadmium and iron were removed, while copper remained in the support. This stability of copper limited the possibility of recycling the sorbent and could cause a poisoning of the resin. To improve the extraction properties, phosphoric acid was pretreated through a chemical reduction of interfering metal ions. Selection of Reducing Agents for the Pretreatment of Phosphoric Acid. Under these experimental conditions, iron is present as Fe(III), which proves to be a very strong oxidizing agent for the extractant and which might strongly decrease the extraction efficiency. Reducing processes have been proposed for the pretreatment of phosphoric acid in the recovery of cadmium with ion flotation techniques.37 Because of the large excess of iron (3.9 g L-1) compared with copper (69.3 mg L-1), the calculation of the reducing agent dosage was made as a function of moles of Fe(III) equivalent to be reduced in Fe(II). Several reducing agents were selected for their redox potential to be lower than those of both iron (Fe3+/Fe2+ ) 0.771 V) and copper (Cu2+/Cu ) 0.337 V). The reducing agent first reacts with the most oxidizing ions Fe3+); when ferric ions are completely reduced to ferrous ions, the reducing agent interacts with cupric ions to form Cu. Based on these criteria as well as their availability at low cost, oxalic acid (CO2/H2C2O4 ) -0.490 V), iron powder (Fe2+/Fe ) -0.440 V), Na2S2O4 (HSO3-/S2O42- ) -0.013 V), and hydrogen (H+/H2(g) ) 0 V) were selected. For hydrogen production, zinc powder was directly added to a HCl solution. Oxalic acid and Na2S2O4 were tested under both solid and solution conditions (7.5 and 15% w/v in water, respectively) because of the difficulty of dissolving oxalic acid in phosphoric acid and the formation of a precipitate with solid Na2S2O4 (the precipitate was less abundant when the salt was dissolved in water). The
Figure 6. Influence of resin dosage on Fe, Cd, and Cu extraction from an industrial phosphoric acid solution using XAD-7 impregnated with Cyanex 301. [H3PO4] ) 12.7 M, [Fe]o ) 3900 mg L-1, [Cd]o ) 45 mg L-1, [Cu]o ) 69.3 mg L-1.
Figure 7. Copper removal by pretreatment with reducing agents (reagent addition determined as a function of the ratio of equivalents of reducing agent to equivalents of oxidizing ferric ions). [H3PO4] ) 12.7 M, [Fe]o ) 3900 mg L-1, [Cd]o ) 45 mg L-1, [Cu]o ) 69.3 mg L-1, Na2S2O4 was added as a solid or as a 15% w/w aqueous solution.
reduction properties were too low to justify an industrial application of this process with oxalic acid and hydrogen (results not shown). Better results were obtained with both iron powder and Na2S2O4. Figure 7 shows that, to remove more than 97% of the copper, a ratio between the reducing agent and the equivalents of ferric ion greater than 8 and 6 for iron powder and Na2S2O4, respectively, is required. Na2S2O4 seems to be a better choice for preventing an increase in the iron concentration in the solution. Cadmium Extraction and Stripping Kinetics after Pretreatment of Industrial Phosphoric Acid Solution. Figure 8 (top) shows the sorption kinetics obtained in phosphoric acid solutions after the prereduction of copper and ferric ions with either iron powder or Na2S2O4 (in solution). With dilute phosphoric acid, when the solutions were preconditioned with Na2S2O4, the extraction kinetics were significantly faster than with iron powder. In particular, whereas it took 4 h to achieve about 95% of the total extraction with Na2S2O4, with iron powder, the same level of extraction was reached within 10 h. With concentrated H3PO4, the results were very different. With Na2S2O4pretreated solutions, a fast decrease in the cadmium concentration was observed within the first 6-8 h, followed by an increase in the residual concentration,
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1431
Figure 8. Cadmium extraction (top) and stripping (bottom) kinetics (HCl, 5 M) for Amberlite XAD-7 resin impregnated with Cyanex 301 from industrial phosphoric acid solutions pretreated with either Na2S2O4 (solid symbols) or iron powder (open symbols). The legend gives the concentrations of phosphoric acid solutions and consequently the initial concentrations of cadmium during the extraction step. [H3PO4] ) 3.2 M, [Cd]o) 11.3 mg L-1 and [H3PO4] ) 12.7 M, [Cd]o) 45 mg L-1.
which tended finally to the initial concentration (not shown). No explanation has been found for this unexpected trend. With iron-powder-pretreated solutions, the standard decay curve was observed but with very slow extraction kinetics; the level of 95% of the total uptake was reached after about 100 hours. From these curves, it appears that the extraction kinetics are very sensitive to the phosphoric acid and cadmium concentrations and to the pretreatment of the solution with either iron powder or Na2S2O4. A comparison with the kinetic curves obtained with synthetic solutions confirms that the presence of other interfering elements in the solution strongly decreases the extraction rates for both systems. Another interfering phenomenon might be the change in the viscosity of the solution resulting from the presence of other constituents. Figure 8 (bottom) presents the stripping kinetics for the systems used in the previous extraction kinetics curves. For the resins loaded at low phosphoric acid concentration, the stripping kinetics were independent of the prereduction technique and surprisingly slower than those obtained with resins that were loaded in a solution with a high phosphoric acid concentration. This
result contrasts with the conclusions obtained for synthetic solutions for which the concentration did not influence stripping kinetics. Whereas 2-3 min were sufficient to achieve 95% of the total stripping yield with synthetic solutions, the contact time for comparable results increased to 8-10 min with the industrial solution. The Amberlite XAD-7/Cyanex 301 system was tested for recycling in five extraction/stripping cycles under identical experimental conditions. The results (not shown) indicate that it was difficult to recycle the resin. At the second extraction step, the efficiency did not exceed 40%, and for the following cycles, extraction did not exceed a few percent. Although the influence of interfering compounds cannot be neglected in stripping kinetics, the equilibria were comparable, and interfering ions should affect the extraction step more significantly. HCl (5 M) was very efficient for stripping cadmium from loaded resins. The decrease in extraction performance can be explained by a progressive poisoning of the EIR resulting from the sorption of other metals that are not eluted from the resin. Elements other than iron and copper also interfere. For example, arsenic has a special affinity for the resin,36 so that small amounts of resin are sufficient to significantly decrease the arsenic concentration (results not shown). In this preliminary study focusing on cadmium, the stripping of arsenic from the resin using HCl (5 M) has not been investigated; however, we assume that the experimental conditions are not sufficient to remove arsenic from the loaded resin. The increasing concentration of arsenic in the resin and the weak stripping efficiency with HCl could explain the enrichment of the resin with arsenic and its progressive poisoning. Cadmium Extraction from Industrial Phosphoric Acid Using Amberlite XAD-7/Cyanex 301 in Column Systems. Figure 5 superimposes experimental results (breakthrough curve and desorption curve) obtained for synthetic solutions with those obtained for dilute industrial phosphoric acid. The breakthrough curve was similar to that obtained with synthetic solutions and was characterized by a sigmoidal trend. However, there was a great difference between these experimental runs, especially regarding the breakthrough volume, which was strongly reduced for industrial solutions. Whereas the breakthrough volume was about 350 mL for synthetic solutions, it reached only 75 mL for industrial solutions (23 BV for the industrial solution, instead of 93 BV for the synthetic solution). The complete saturation of the resin was observed after approximately 1000 mL had been pumped through the column, and the amount of cadmium immobilized on the resin reached 11.6 mg. The extraction capacity was about 9.7 mg g-1, close to the value obtained with synthetic solutions (11.1 mg g-1). The solid line represents the empirical modeling of the breakthrough curve according to eq 4. The parameters are Rb ) 0.020 mL-1 and Vo ) 180 mL; a larger dispersion around the theoretical curve is observed in the distribution of experimental points (MSR ) 0.080). The bottom curve of Figure 5 shows the stripping curve, which is characterized by its sharpness. A maximum cadmium concentration of 1.4 g L-1 is obtained. After 15 mL, no more cadmium is stripped from the resin, and the average concentration is ca. 194 mg L-1; the amount of cadmium eluted from the resin does not exceed 3.7 mg (stripping yield ) 32%). Within the first 10 mL of the extraction,
1432
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001
95% of the total desorption was achieved (with an average concentration of 284 mg L-1). This result contrasts with stripping operations performed in batch systems for which the stripping is complete during the first extraction/stripping cycle. Conclusion This study shows that Amberlite XAD-7 is very efficient at removing cadmium from dilute phosphoric acid solutions. Even at concentrations as high as 12.7 M, extraction capacities higher than 11 mg g-1 are obtained. The extraction is rapid: a few hours are sufficient to achieve about 90% of the total extraction under unfavorable conditions (i.e., high acid and cadmium concentrations), and less than 1 h is required to completely remove cadmium from 3.2 M phosphoric acid solutions. HCl ensures a complete stripping of cadmium from loaded resin, which can be reused for at least 10 cycles without any detectable decrease in extraction efficiency. In a column system, cadmium can also be efficiently recovered. The presence of large concentrations of competitor ions strongly decreases the efficiency of cadmium removal from industrial phosphoric acid. Although pretreatment with reducing agents such as Na2S2O4 enhances the extraction efficiency, the stripping and recycling of loaded resins is significantly more difficult. It seems that the presence of large amounts of other metals poisons the resin, which would require complementary stripping operations to remove the metals, actually co-extracted. Preliminary work has shown that arsenic might be the element that interferes the most in the process. Another way to improve the extraction, stripping, and recycling performance levels might consist in combining the treatment of industrial phosphoric acid by impregnated resins with other physicochemical processes. Ortiz et al. combine a preliminary extraction of uranium by sorption on an ion-exchange resin prior to cadmium removal using a nondispersive solvent extraction process.11 The combination of different EIRs might also be considered as a promising alternative, as the combination of several impregnated resins enables the separation of metals from multicomponent solutions.38 Acknowledgment The authors thank the University of Guanajuato, the CONACyT (Ref. 0471P), and the CID (Centro de Investigacio´n y Desarrollo Tecnolo´gico, S.A. de C.V.) for their financial support. Literature Cited (1) Austin, T. J. Manual de Procesos Quı´micos en la Industria; McGraw-Hill: Mexico, 1989. (2) Marcilla, A.; Ruiz, F.; Campos, J.; Asencio, M. Purification of Wet Process Phosphoric Acid by Solvent Extraction with 3-Pentanone, Study of Impurities Distribution. Solvent Extr. Ion Exch. 1993, 11 (3) 455. (3) Skororovarov, J. I.; Ruzin, L. I.; Lomonosov, A. V.; Tselitschev, G. K. Solvent Extraction for Cleaning Phosphoric Acid in Fertilizer Production. J. Radioanal. Nucl. Chem. 1998, 229 (1-2) 111. (4) Avila Rodriguez, M.; Cote, G.; Navarro Mendoza, R.; Saucedo Medina, T. I.; Bauer, D. Thermodynamic Study of the Extraction of Indium(III) and Cadmium(II) by Cyanex 301 from Concentrated HCl Media. Solvent Extr. Ion Exch. 1998, 16 (2) 471. (5) Rickelton, W. A. The Extraction of Cadmium from a Mixture of Phosphoric and Hydrochloric Acids. Solvent Extr. Ion Exch. 1999, 17 (6) 1507.
(6) Rydberg, J., Musikas, C., Choppin, G. R. Principles and Practices of Solvent Extraction. Marcel Dekker: New York, 1992. (7) Rovira, M.; Cortina, J. L.; Arnaldos, J.; Sastre, A. M. Recovery and Separation of Platinum Group Metals Using Impregnated Resins Containing Alamine 336. Solvent Extr. Ion Exch. 1998, 16 (5) 1279. (8) Navarro Mendoza, R.; Saucedo Medina, T. I.; Vera, A.; Avila Rodriguez, M.; Guibal, E. Study of the Sorption of Cr(III) with XAD-2 Resin Impregnated with Di-(2,4,4-trimethylpentyl)phosphinic acid (Cyanex 272). Solvent Extr. Ion Exch. 2000, 18 (2) 319. (9) de Gyves, J.; Rodriguez de San Miguel, E. Metal Ion Separation by Supported Liquid Membranes. Ind. Eng. Chem. Res. 1999, 38 (6) 2182. (10) Urtiaga, M.; Zamacona, S.; Ortiz, M. I. Analysis of a NDSX Process for the Selective Removal of Cd from Phosphoric Acid. Sep. Sci. Technol. 1999, 34 (16) 3279. (11) Ortiz, I.; Alonso, A. I.; Urtiaga, A. M.; Demircioglu, M.; Kocacik, N.; Kabay, N. An Integrated Process for the Removal of Cd and U from Wet Phosphoric Acid. Ind. Eng. Chem. Res. 1999, 38 (6) 2450. (12) Kabay, N.; Demircioglu, M.; Ekinci, H.; Yuksel, M.; Saglam, M.; Akcay, M.; Streat, M. Removal of Metal Pollutants Cd(II) and Cr(III) from Phosphoric Acid Solutions by Chelating Resins Containing Phosphonic or Diphosphonic Groups. Ind. Eng. Chem. Res. 1998, 37 (6) 2541. (13) Bilba, D.; Bilba, N.; Albu, M. Kinetics of Cadmium Ion Sorption on Ion Exchange and Chelating Resins. Solvent Extr. Ion Exch. 1999, 17 (6) 1557. (14) Alexandratos, S. D.; Trochimczuk, A.; Horwitz, E. P.; Gatrone, R. C. Synthesis and Characterization of a Bifunctional Ion Exchange Resin with Polystyrene-Immobilized Diphosphonic Acid Ligands. J. Appl. Polym. Sci. 1996, 61 (2) 273. (15) Juang, R.-S.; Lin, H.-C. Metal Sorption with ExtractantImpregnated Macroporous Resins. 1. Particle Diffusion Kinetics; 2. Chemical Reaction and Particle Diffusion Kinetics. J. Chem. Technol. Biotechnol. 1995, 62, 132 and 141. (16) Kenji, I.; Fumito, T.; Tooru, K.; Eiichiro, N. Preconcentration of Trace Metals from Seawater with 7-Dodecenyl-8-quinolinol Impregnated Macroporous Resin. Anal. Chem. 1987, 59, 2491. (17) Philip, H.; Dietz, M. L.; Chiarizia, R.; Diamond, H. Separation and Preconcentration of Uranium from Acidic Media by Extraction Chromatography. Anal. Chim. Acta 1992, 266, 25. (18) Koopman, C.; Witkamp, G. J.; Van Rosmalen, G. M. Removal of Heavy Metals and Lanthanides from Industrial Phosphoric Acid Process Liquors. Sep. Sci. Technol. 1999, 34 (15) 2997. (19) Kabay, N.; Demircioglu, M.; Ekinci, H.; Yu¨ksel, M.; Saglam, M.; Streat, M. Extraction of Cd(II) and Cu(II) from Phosphoric Acid Solutions by Solvent-Impregnated Resins (SIR) Containing Cyanex 302. React. Funct. Polym. 1998, 38, 219. (20) Cortina, J. L.; Miralles, N.; Sastre, A.; Aguilar, M.; Profumo, A.; Pesavento, M. Solvent Impregnated Resins Containing Cyanex 272. Preparation and Application to the Extraction and Separation of Divalent Metals. React. Polym. 1992, 18, 67. (21) Yoshito, W.; Serigne, A. N.; Hideyuki, M.; Toshirou, Y.; Kenichi, A. Extraction of Arsenic(III) with Macroporous Resin Impregnated with Bis(2-ethylhexyl)ammonium Bis(2-ethylhexyl)dithiocarbamate. Anal. Sci. 1998, 14, 299. (22) Juang, R.-S.; Ju, C.-Y. Kinetics of Sorption of Cu(II)Ethylenediaminetetraacetic Acid Chelated Anions on CrossLinked, Polyaminated Chitosan Beads. Ind. Eng. Chem. Res. 1998, 37 (8) 3463. (23) Helfferich, F. Ion Exchange; Dover Publications: Mineola, NY, 1995. (24) Guibal, E.; Larkin, A.; Vincent, T.; Tobin, J. Chitosan Sorbents for Platinum Sorption from Dilute Solutions. Ind. Eng. Chem. Res. 1999, 38 (10), 4011. (25) Crank, J. The Mathematics of Diffusion; Oxford University Press: Oxford, U.K., 1995. (26) Sole, K. C.; Hiskey, J. B. Solvent Extraction Characteristics of Thiosubstituted Organophosphoric Acid Extractants. Hydrometallurgy 1992, 30, 345. (27) Pearson, R. G. Hard and Soft Acids and Bases. J. Am. Chem. Soc. 1963, 85, 3533. (28) Marcus, Y. Ion Properties; Marcel Dekker: New York, 1997. (29) Cortina, J. L.; Miralles, N.; Sastre, A. M.; Aguilar, M. Solid-Liquid Extraction Studies of Zn(II), Cu(II) and Cd(II) from Chloride Media with Impregnated Resins Containing Mixtures of
Ind. Eng. Chem. Res., Vol. 40, No. 5, 2001 1433 Organophosphorous Compounds Immobilized on to Amberlite XAD2. Hydrometallurgy 1995, 37, 301. (30) Cortina, J. L.; Miralles, N.; Aguilar, M.; Sastre, A. M. Distribution Studies of Zn(II), Cu(II) and Cd(II) with Levextrel Resins Containing Di(2,4,4-trimethylpentyl)phosphinic Acid (Lewatit TP807′84). Hydrometallurgy 1996, 40, 195. (31) Cortina, J. L.; Miralles, N.; Sastre, A. M.; Aguilar, A.; Profumo, A.; Pesavento, M. Solvent-Impregnated Resins Containing Di-(2,4,4-trimethylpentyl)phosphinic Acid. II. Study of the Distribution Equilibria of Zn(II), Cu(II) and Cd(II). React. Polym. 1993, 21, 103. (32) Komiyama, H.; Smith, J. M. Intraparticle Mass Transport in Liquid-Filled Pores. AIChE J. 1974, 20 (4), 728. (33) Alexandratos, S. D.; Shelley, C.; Horwitz, E. P.; Chiarizia, R. A Mechanism for Enhancing Ionic Accessibility into Selective Ion Exchange Resins. Solvent Extr. Ion Exch. 1998, 16 (4), 951. (34) Ruiz, M.; Sastre, A. M.; Zikan, M. C.; Guibal, E. Palladium Sorption on Glutaraldehyde Cross-linked Chitosan in Dynamic Systems. J. Appl. Polym. Sci. 2001, in press. (35) Ko, D. C. K.; Porter, J. F.; McKay, G. Correlation-Based Approach to the Optimization of Fixed-Bed Sorption Units. Ind. Eng. Chem. Res. 1999, 38 (12), 4868.
(36) Facon, S.; Avila Rodriguez, M.; Cote, G.; Bauer, D. General Properties of Bis(2,4,4-trimethylpenthyl)dithiophosphonic Acid (Cyanex 301) in Acid Liquid-Liquid Extraction Systems. In Solvent Extraction in the Process Industries; Proceedings of International Solvent Extraction Conference ISEC’93; Logdail, D.H., Slater, M. J., Eds.; Elsevier: Amsterdam, The Netherlands, 1993; Vol. 1, pp 557-564. (37) Jdid, E.; Blazy, P.; Bessiere, J.; Durand, R. Removal of Cadmium Contained in Industrial Phosphoric Acid Using the Ionic Flotation Technique. In Trace Metal Removal from Aqueous Solution; Thompson, R., Ed.; Borax Holding Ltd.: Chessington, U.K., 1986; pp 109-136. (38) Gonzalez Mun˜oz, M. P.; Saucedo Medina, T. I.; Navarro Mendoza, R.; Avila Rodriguez, M.; Guibal, E. Selective Separation of Cd(II), Ni(II) and Fe(III) Using Extractant-Impregnated Resins. Manuscript in preparation.
Received for review May 30, 2000 Revised manuscript received December 11, 2000 Accepted December 19, 2000 IE0005349