Environ. Sci. Technol. 2004, 38, 886-891
Arsenic(V) Removal with Polymer Inclusion Membranes from Sulfuric Acid Media Using DBBP as Carrier MA. DE LOURDES BALLINAS,† E D U A R D O R O D R IÄ G U E Z D E S A N M I G U E L , † MA. TERESA DE JESU Ä S R O D R IÄ G U E Z , † O R L A N D O S I L V A , ‡ M A R IÄ A M U N ˜ OZ,§ AND J O S E F I N A D E G Y V E S * ,† Departamento de Quı´mica Analı´tica. Facultad de Quı´mica, UNAM, Ciudad Universitaria, 04510 Me´xico, D.F. Me´xico, Unitat de Quı´mica Analı´tica, Facultat de Ciencies, Universitat Auto´noma de Barcelona, 08193 Barcelona, Spain, and Departamento de Ingenierı´a Quı´mica, ETSEQ-Universitat Rovira i Virgili, Avenue Paisos Catalans 26, Campus Sescel.lades, 43007 Tarragona, Spain
Polymer inclusion membranes (PIMs) based on cellulose triacetate (CTA) and dibutyl butyl phosphonate (DBBP) were tested for arsenic(V) separation from H2SO4 for its recovery from copper electrolytes. Solvent extraction experiments allowed the determination of the As(V)DBBP and H2SO4-DBBP complexes formed in the organic phase. Application of a transient model to membrane transport experiments in solutions containing only arsenic or H2SO4 indicated that it occurred under a kinetically controlled regime by formation of H3AsO4[DBBP]2 and H2SO4[DBBP] species, respectively. When arsenic and H2SO4 are simultaneously present, the existence of a third species, H3AsO4[DBBP][H2SO4], explains well the fact that As(V) flux decreases and that H2SO4 flux increases. In both cases, a limiting 50% recovery value was obtained. However, active arsenic transport (>50%) is achieved if the H2SO4 concentration gradient is assured (e.g., using a triple-cell configuration). In this way, high arsenic recovery factors (90% in 800 min) were obtained with initial concentrations of 5000 mg/L As(V) and 220 g/L H2SO4. In all membrane systems tested, good As(V) selectivity over copper (up to 30 000 mg/L) was attained.
1. Introduction Separation techniques for arsenic and its compounds are important from an environmental protection point of view. For this reason, several processes based on solvent extraction, precipitation, adsorption, ion exchange, and membrane methods are applied for arsenic removalfrom different sources (1-5). In copper processing, As(V) is a common impurity since it is one of the most frequently found elements in copper sulfide concentrates. Its occurrence varies from very low [As (%) 2] levels (6). In electrorefining copper closed-circuit processes, the electrolyte is bled-off for impurity control (As mainly) to avoid contamination of the refined cathode copper (7) and to * Corresponding author phone: +525-556223792; fax: +525556223723; e-mail:
[email protected]. † Ciudad Universitaria. ‡ ETSEQ-Universitat Rovira i Virgili. § Universitat Auto ´ noma de Barcelona. 886
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004
prevent the production of the noxious gas H3As when the copper content in the electrolyte is less than 0.5 g/L. Solvent extraction processes, where tri-n-butyl phosphate (TBP) is the main extracting reagent, are employed in some copper refineries for arsenic removal (1). Recently, membrane-based separation processes have gained considerable importance (8) since they are regarded as potential environmental friendly technologies (9). The possibility of using polymeric liquid membranes with increasing stability and selectivity for the separation of valuable or toxic metal species from acidic media makes this separation process even more attractive (10, 11). Since the pioneer work of Vofsi and Jagur-Grodzinski (12), who investigated solvent-polymeric membranes based on poly(vinyl chloride) (PVC) and organophosphorus compounds, other authors (13) have proposed liquid membranes using CTA, plasticizers, and extracting reagents as the membrane matrix. Low carrier and solvent consumption, uphill concentration availability, continuous operating mode with scaling-up possibilities (mainly in the hollow fiber configuration), and simultaneous implementation of extraction and back-extraction steps are further advantages of these liquid membrane systems. Although several reagents have been proposed for the solvent extraction of As(V) from sulfuric acid media, organophosphorus reagents have been thoroughly studied due to their high affinity for this species (14, 15). Recently, the increasing solvating properties of DBBP over TBP have encouraged research work with this reagent (16). However, to the best of our knowledge, no study has reported on the application of a liquid membrane system for As(V) separation. In this work, we have studied As(V) removal from sulfuric acid media associated with copper electrolyte bleeds using a PIM containing DBBP as the carrier. Arsenic and sulfuric acid speciation in solvent extraction and membrane experiments allowed the mathematical description and optimization of the membrane process. Relevant thermodynamic and kinetic data were evaluated. The influence of parameters such as membrane composition and arsenic and sulfuric acid concentrations on permeability is discussed.
2. Experimental Section 2.1. Reagents and Apparatus. The extractants DBBP (CH3(CH2)3P(O)[O(CH2)3CH3]2, 95%, F ) 0.948 g/mL) and TBP ([O(CH2)3CH3]P(O), 98%, F ) 0.979) were kindly supplied by Rhodia, and Aliquat 336 (CH3N[(CH2)7CH3]3Cl, F ) 0.884) was an Aldrich reagent. All were used without further purification. Organic solutions for SX experiments consisted of undiluted DBBP or DBBP diluted in kerosene (low aromatics content, PEMEX). CTA (Aldrich), Na2HAsO4‚7H2O, and CuSO4‚5H2O were analytical grade (Aldrich). A Jobin Yvon JY-18 ICP emission spectrometer was used for elemental analysis (P, As, Cu, Bi, Ni). A Burrel 75 mechanical shaker was used for liquid-liquid extraction. The membrane (effective area 6.8 × 10-4 m2) was clamped in a cell as described elsewhere (17), each compartment having a volume of 75 mL. Aliquots were analyzed in different periods of time, following As(V) concentration profiles and, when required, Cu(II) concentration profiles. For arsenic concentrations lower than 5 mg/L in aqueous phase, analysis was performed by ETAAS (Perkin-Elmer 3100 AA spectrometer with HGA-600). Sulfuric acid profiles were also obtained by titrating the final feed and stripping solutions with standarized NaOH and measuring pH as a function of time in diluted samples. To verify sulfate presence, the barium 10.1021/es030422j CCC: $27.50
2004 American Chemical Society Published on Web 12/10/2003
precipitation method was used (18). For all experiments, the temperature was kept constant at 20 ( 2 °C. Membrane characterization was performed by SEM (Hitachi S570) techniques using Au for coating. An energydispersive X-ray spectrometer (EDX Link Isis 200) was used to analyze the extractant distribution along the membranes. 2.2. Solvent Extraction Procedure. Aqueous solutions of As(V) in sulfuric acid media were contacted with DBBP as received or diluted in kerosene. Arsenic concentration ranged from 500 to 3000 mg/L in sulfuric acid (20-240 g/L). Stirring was carried out at constant rate. Volumes of 5 mL of each phase were used in most of the experiences. Previous experiments indicated that equilibrium conditions were attained in less than 15 min, independent of As, H2SO4, or DBBP concentrations. Extraction was also studied in the presence of copper and other elements usually present in copper electrolytes (Ni, Bi). All studies were carried out at least in duplicate. Standard deviation observed was always less than 5%. 2.3. Membrane Preparation and Transport Experiments. PIMs were synthesized following the procedure of Sugiura et al. (13). Different amounts of DBBP were weighed and diluted with CHCl3. This solution was mixed with a fixed volume of 2% w/v CTA-CHCl3 solution and poured in a Petri dish (9 cm diameter). The solvent was allowed to evaporate overnight. A transparent film was obtained that was peeled off and used in transport experiments. The same procedure was followed for TBP- and Aliquat 336-containing membranes. Feed solutions were prepared by varying As(V) and H2SO4 content. Stripping solutions consisted of 2 M LiCl. No osmotic flux was observed under these conditions. Aliquots for elemental analysis of 250 µL each were taken every 60 or 90 min, and permeability (P) was evaluated using the equation obtained from the first-order expression:
P)
J dC L )C dt C
(1)
where L ) V/A, V is the volume (m3) of aqueous feed solution, A is the effective membrane area (m2), t is time (min), C is the concentration (mol m-3) of analyte (As, Cu, or H2SO4), in feed phase, and J refers to molar flux. Direct integration of eq 1 gives the following relation, assuming that P remains constant:
Ct P ln ) - t Co L
(2)
where Ct is concentration at time t and Co is the initial analyte concentration. Besides permeability, selectivity over copper and other electrolyte concomitants was also evaluated. Durability was tested for DBBP-CTA membranes, as runs performed until membrane permeability dropped 20% its initial value. Phosphorus was also analyzed in aqueous phase as a measured of carrier loss. The membrane was washed with deionized water (2 h) for its continuous use. The best performing DBBP-CTA membrane was used for As(V) uphill transport. Experiments were carried out in a triple cell configuration with compartments of 200 mL each and membrane areas of 8.5 × 10-4 m2.
3. Results and Discussion 3.1. Solvent Extraction Experiments. According to reported work on extraction by organophosphorus extractants (19), the extraction of As(V) by DBBP may be described by
jH3AsO4 + kH2SO4 + pDBBP + hH2O w [(H3AsO4)j(H2SO4)k(DBBP)p(H2O)ho] + (h - ho)H2O (3)
where the bar denotes an organic phase species. In the following, equilibrium is simplified as
jB + kC + pA + hH w BCkApHho + (h - ho)H
(4)
Using the definition of the extraction constant and taking logarithms into account, this expression transforms to
h] log DAs ) log K′ + k log[C] + p log[A
(5)
where DAs is the As(V) distribution coefficient and K′ is the conditional extraction constant. Experimental log DAs ) f (log[DBBP]H2SO4) linear relations show fractional slope values, indicating that a complex equilibrium is present. To account for H2SO4 co-extraction, log DH2SO4 ) f (log[DBBP]As) was obtained. The results were treated with the LETAGROP-DISTR solvent extraction program (20) to find the stoichiometric coefficients values of eq 4 and the corresponding extraction constant by least-squares minimization. Under the experimental conditions used in this work, two arsenic species are present in the organic phase: j ) 1, k ) 0, p ) 2 and j ) 1, k ) 1, p ) 1. Sulfuric acid is extracted in the presence of arsenic forming two compounds (j ) 0, k ) 1, p ) 1 and j ) 1, k ) 1, p ) 1) (21). Using equilibria data for the organic complexes formed, a predominance diagram (22) was constructed. H3AsO4[DBBP]2 predominates for HSO4- < 1.4 M. For higher acid media, H2SO4[DBBP] is present in the same proportion as H3AsO4[DBBP]2. Predominance of H2SO4[DBBP] and H3AsO4[DBBP]2 over H3AsO4[DBBP][H2SO4] species is always evidenced. Experiments performed using 30 000 mg/L Cu(II), 3000 mg/L As(V), and 220 mg/L H2SO4 showed that less than 5% of Cu(II) was extracted even at high DBBP concentration. 3.2. Polymeric Membranes. It is known that some organophosphorus extractants polymerize with CTA, forming a plastic network (23). In our case, polymerization between DBBP and CTA occurred spontaneously in all DBBP/CTA ratios studied. For a membrane composed of 0.22 CTA (mg/cm2) and 2.53 DBBP (mg/cm2), a thickness of 8 ( 2 µm was measured by SEM. It was corroborated that it increases with CTA content. Membrane homogenity was determined by EDX-SEM analysis. Results indicated the same composition in terms of %P (from a phosphorus/oxygen relation) for several points tested in the sample. In the PIM system, DBBP maintains affinity for the same extractable elements, as in liquid organic state. Blank membranes (formed exclusively of CTA) showed no evidence of arsenic or sulfuric acid transport after a 20-h test period. Arsenic and sulfuric acid were transported in a wide concentration range (As [0-5000 mg/L] and H2SO4 [0-220 g/L]). In CTA-DBBP membranes, carrier losses were studied by means of %P in aqueous feed and stripping phases. Under the experimental conditions employed, no evidence of P was found in the lapse of time studied for each membrane. It was observed that the number of cycles a membrane can be used depends on the thickness of the membrane. At least 5 cycles were performed under several solution compositions until permeability dropped to 20% of its initial value using a 8 ( 2 µm membrane. However, 8 cycles were performed with thicker membranes (15 ( 2 µm) without appreciable variation in stability, although the initial permeability diminishes. This result indicates that a compromise between stability and permeability must be accomplished in order to determine its possibilities in long-term removal experiments. In Figure 1, the stabilities for different membrane carriers is compared. The advantage of DBBP over TBP in PIMs is shown. VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
887
Differential equations describing the transient concentrations of species involved in arsenic membrane extraction are
∂2 C A ∂CA ) DA 2 ∂t ∂x ∂CBA2 ∂t
∂2CBA2
) DBA2
∂2x
dCB kfB kbB ) - (CA)2x)δ(CB)x)δ + (C ) dt L L BA2 x)δ
FIGURE 1. Comparison of the performance (permeability and stability) of different carriers in PIMs. TBP, 2.04 mg/cm2; DBBP, 2.04 mg/cm2; TBP and Aliquat 336, 1.75 and 0.24 g/cm2, respectively. Feed: 3000 g/L As(V), 2 M H2SO4.
3.2.1. Arsenic and Sulfuric Acid Transport for DBBPCTA Membranes: Modeling for an Independent Extraction Process. In PIMs, the carrier is contained in a continuous phase and is mobile within as plasticizers in polymers (24). Membrane extraction can be viewed as an interfacial reaction in which the extractant (DBBP) diffuses to the membrane interface where it complexes with a species susceptible to be extracted [As(V) or H2SO4] present in the aqueous feed side. The complex is then transferred to the bulk membrane phase and finally is released to the aqueous stripping solution. The transport rate can be controlled by the diffusion of species in the membrane (diffusion-controlled regime) or by interfacial reactions among aqueous compounds and carrier (kinetic-controlled regime). To determine the influence of the diffusion and chemical kinetics in the arsenic extraction process on one hand and in sulfuric acid extraction on the other, modeling of transport was performed using a transient approach (25-28). This framework is required in order to account for non-steadystate conditions (i.e., the time-dependent As(V) concentration gradient as the main force of the process). The main assumptions of the model used were as follows: (i) Extraction rate is influenced by both the rate of the chemical reactions occurring at the feed interface and by the Fickian diffusion of species within the membrane. (ii) Rapid mass transport is occurring in the bulk of the aqueous phases due to stirring, and there is rapid diffusion of As(V) or H2SO4 across the stagnant diffusion layers (separating the membrane and the bulk aqueous phase). (iii) Mass transfer at the stripping-side interface does not play a significant role due to fast chemical reactions taking place and the aqueous chemical conditions that favor As(V) or H2SO4 release. As previously stated, arsenic extraction in sulfuric aciddepleted media can be described mainly by this equilibrium:
H3AsO4 + 2DBBP S H3AsO4[DBBP]2
(6)
simplified in the following as
B + 2A S BA2
(7)
where the extraction reaction rate is characterized by the respective forward (kf) and backward (kb) kinetic rate constants. 888
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004
(8)
(9)
(10)
where DA and DBA2 are the diffusion coefficients for carrier and complex, respectively; kbB and kfB are related to the equilibrium constant previously obtained by KexB ) kbB/kfB; L is the geometrical length (L ) V/A); δ is the membrane thickness; x is the axial distance (x ) 0 corresponds to the strip membrane interface and x ) δ corresponds to the feed membrane interface); t is time; and the subscripts A and B refer to the different compounds. The model is simplified further assuming that DBA2 ≈ DA so the mass balance expression CA + 2CBA2 ) C0A is valid, and the complex concentration is determined provided C0A is known. In this way, the process is described by eqs 8 and 10 where the initial and boundary conditions are
CA(0,x) ) C0A
( ) ( ) ∂CA ∂x
∂CA ∂x
x)0
)
x)δ
)0
-L dCB 2DA dt
CB(0) ) C0B
(11) (12)
(13) (14)
Meanwhile for sulfuric acid extraction in the absence of arsenic, the main extraction equilibrium is
H2SO4 + DBBP S H2SO4[DBBP]
(15)
simplified as
C+A h S CA
(16)
The transient model applied in this case corresponds to a set of equations analogous to the arsenic set previously described. Both arsenic and sulfuric transport equation sets were solved numerically in an independent way by the implicit-finite difference method using a Matlab program in order to calculate the concentration profiles of the transported species. Parameters used for arsenic and sulfuric acid transport modeling are listed in Table 1. DA and DC and the corresponding kbB and kbC were systematically varied for each As(V) and H2SO4 concentration profile data (CB/CB0 or CC/ CC0 as a function of time) with the aim of fitting the model equations to the latter experimental data. It was noticed that for experimental conditions (given by KexB, KexC, δ, L, C0A, C0B, and C0C values), the adjustment is only dependent on the respective kb, indicating kinetic control for both sulfuric and arsenic extraction processes. This result is explained considering that a high carrier-concentrated thin membrane was used. Under this regime, only a broad estimation of the value of the diffusion coefficient of the analyte in the membrane may be obtained. Time-course profiles of experimental data
TABLE 1. Experimental Parameters and Results for Data Modeling of As(V) and H2SO4 Transports Kex As(V) H2SO4
2.884 0.039
δ (m) 10-6
8× 8 × 10-6
L (m) 0.1103 0.1103
C0A (mol/m3) 11 980 11 980
C0B (mol/m3) 16.2 0
FIGURE 2. Dimensional time-dependent concentration profiles for As(V) and H2SO4 according to the experimental conditions reported in Table 1. Points, experimental values, continuous lines, results of data modeling. (points) and theoretical curves (lines) are compared in Figure 2. Furthermore, when DBBP is in excess in relation to As(V) and H2SO4 concentration, the following relations can be written:
PoB dCB )[C ] dt L B
(17)
dCc PoC )[C ] dt L c
(18)
PoB ) kfBC2A
(19)
PoC ) kfCCA
(20)
and
where
and
where PoB and PoC are the initial theoretical permeabilities
C0C (mol/m3) 0 2000
Kb (m/s) 10-14
1.1 × 3 × 10-10
Po calcd (m/s) 10-6
4.5 × 3.6 × 10-6
Po exptl (m/s) 4.9 × 10-6 3.3 × 10-6
for arsenic and sulfuric acid, respectively. Permeability calculated values are reported in Table 1 and compared with experimental values. As observed, good agreement among initial permeability values and concentration profiles (within 5% error) confirm the validity of the model. 3.2.2. Arsenic Transport in the Presence of Sulfuric Acid in Aqueous Phase by DBBP-CTA Membranes. Arsenic permeability was determined as a function of time for several membrane compositions (variable CTA and DBBP content), using constant aqueous feed and stripping phases. Permeability is affected by CTA more appreciably than by DBBP content, accounting for membrane thickness modification. Arsenic transport was measured as a function of its concentration in feed phase. In all experiments sulfuric acidarsenic mass relation is conserved. It is observed that arsenic permeability does not change significantly when its concentration varies from 5 to 3000 mg/L for a given membrane. To test membrane selectivity, several experiments were carried out in the presence of sulfuric acid and copper following their respective concentration profiles. Results obtained in an experiment using 30 000 mg/L Cu(II), 5000 mg/L As(V), 5000 mg/L Ni(II), 2000 mg/L Bi(III), and 220 g/L sulfuric acid as feed phase showed no significative Cu, Ni, and Bi transport in the period of time studied (for 1200 min
