Separation Fe(III)-Mn(II) via supported liquid membrane technology in the treatment of spent alkaline batteries

: The transport of iron(III) from Fe(III)-Mn(II)-HCl mixed solutions through a flat-sheet supported liquid membrane is investigated, being the carrier phase of Cyanex 923 (commercially available phosphine oxide extractant) dissolved in Solvesso 100 (commercially available diluent), as a function of hydrodynamic conditions, concentration of manganese and HCl in the feed phase, and carrier concentration in the membrane phase. A transport model is derived that describes the transport mechanism, consisting of diffusion through a feed aqueous diffusion layer, a fast interfa-cial chemical reaction, and diffusion of the Fe(III)-Cyanex 923 complex across the membrane phase. The membrane diffusional resistance (Δ m ) and feed diffusional resistance (Δ f ) are calculated from the model, and their values are 145 s/cm and 361 s/cm, respectively. It is apparent that the transport of iron(III) is mainly controlled by diffusion through the aqueous feed boundary layer, being the thickness of this layer calculated as 2.9x10 -3 cm. Since Mn(II) is not transported through the membrane phase, the present system allows to the purification of this manganese-bearing solutions.


Introduction
Recycling is an issue of increasing necessity in 21 st century, this is caused, among others, by the decrease and depletion of certain raw materials, and also for the more stringent environmental pressures about the discharge of contaminants.
One example of materials to be recycled are alkaline batteries, also they contained valuable metals that can be recovered and reuse, gaining a further profit in the recycle process. CENIM-CSIC developed investigations about the use of ammonium carbonate processing in the recycle of zinc-bearing materials such as electric arc furnace (EAF) flue dusts and Waelz oxides [1][2][3], since zinc is an important component of alkaline batteries, this type of processing was applied on the recycle of these spent alkaline batteries [4,5]. In the process, the mass yielded after breaking of the battery carcase is subjected to a leaching process in ammoniacal ammonium carbonate solutions, being zinc recovered from the leachate, whereas the solid residue contained iron and manganese (VII) compounds. This residue was treated via leaching in HCl medium, yielding a solution containing Fe(III) and Mn(II), since in the operational conditions, manganese(VII) oxidizes chloride ions and, thus, it is reduced to Mn(II). From the solution, Fe(III) and Mn(II) can be separate using liquid-liquid extraction technology, however, the use of liquid membranes, specifically supported liquid membranes technologies, can be also of interest due to the economical and operational advantages that these technologies possessed.
In fact, supported liquid membranes technologies has been proposed for the treatment of strategic and toxic metals, having some of their uses recently reviewed [6,7]  Before scaling up the flat-sheet supported liquid membrane to a more dynamic membrane methodology, such hollow fiber modules, a model of the liquid membrane system is convenient in order to design this more efficient process.
The present investigation presents a kinetic modelling of the active transport of Fe(III), from Fe(III)-Mn(II)-HCl solutions, using Cyanex 923-Solvesso 100 solutions immobilized on a microporous hydrophobic support. Diffusional resistances due to the membrane and the feed phase are calculated from the model, and the influence of hydrodynamic and chemical conditions are established in order to yield and efficient flatsheet liquid membrane system.

Reagents and solutions
The extractant used in the investigation was Cyanex 923 (commercially available phosphine oxide), which was dissolved in Solvesso 100 (Exxon Chem Iberia, aromatic diluent) in order to adequate the range of extractant concentrations to the transport experiments. Also, the composition of the aqueous feed solutions was scaled down to the metal concentrations range convenient to this type of membrane experimentation. All the chemicals used in the experimentation were of AR grade.
The solid support used in the present work was Millipore Durapore GVHP4700 (microporous polyvinylidene fluoride film) of 75% porosity, 1.67 tortuosity and 12.5x10 -3 cm thickness and 0.22 μm effective pore size.

Liuid-liquid extraction experiments
Iron (III) liquid-liquid extraction experiments were performed by mixing, in thermostatted separatory funnels, equal volumes of organic solutions of Cyanex 923 (from 1.9x10 -2 to 0.13 M) in Solvesso 100 with 0.01 g/L iron(III) in 4 M HCl solutions for 10 min at 20º C. Previous experiments showed that equilibrium was reached within 5 min of contact between both phases. After the quick phase disengagement (less than 30 s), iron(III) was analyzed in the aqueous solutions by atomic absorption spectrometry, reproducibility within ±3%, and the metal concentration in the equilibrated organic solutions was estimated by the corresponding mass balance.

SLM experiments
Transport experiments were carried out in a two compartments cell which consisted of a feed solution half-cell (200 cm 3 ) separated from the receiving solution half-cell (200 cm 3 ) by the solid support having an effective membrane area of 11.3 cm 2 . The feed and the receiving solutions were mechanically stirred, at 20º C, to avoid concentration polarization conditions at the support interfaces and in the bulk of both solutions.
The supported liquid membrane was prepared by impregnation of the solid support, with the corresponding carrier solution, by immersion for 24 h and then left to drip for 20 seconds before being placed in the cell. Metals transport were determined by monitoring Fe(III) and Mn(II) concentrations in the feed and receiving phases, as a function of time, by atomic absorption spectrometry. Metal concentrations in the solutions were found to be reproducible within ±3%. The iron overall mass transfer coefficient (KFe) was computed using the next equation: where A was the support area, V is the volume of the feed solution, [Fe]f,t and [Fe]f,0 were the iron concentrations in the feed solution at an elapsed time and time zero, respectively, and t was the elapsed time.
The percentage of iron transported to the receiving solution was calculated by: where [Fe]r,t was the iron concentration in the receiving solution at an elapsed time. Manganese(II) transport was evaluated as above, and under the present experimental conditions, basically manganese(II) was not transported (KMn <1x10 -6 cm/s) across the membrane.

.Iron(III) liquid-liquid extraction equilibrium
The extraction of iron(III) by the phosphine oxide was based on a solvation reaction In order to validate eq.(3) and calculate both the extraction constant and the stoichiometric factor n values, the experimental data were treated by a tailored computer program which minimized the U function, defined as: The results derived from the experimental data and treated by the program, indicated that the extraction of iron (III) by Cyanex 923 corresponded to the reaction showed in eq.(3), with log KHCl 4.86 and U 0.093. The stoichiometric factor n was calculated as 4. Figure 1 showed a probable transport scheme for Fe(III) with Cyanex 923 (represented as L) dissolved in Solvesso 100 through a supported liquid membrane. The driving force for iron(III) transport was the difference in acidity between the feed and receiving phases, thus, iron(III) transport was coupled to the acid co-transport from the feed to the receiving phase. In order to yield effective iron(III) transport through the supported liquid membrane, it was of importance establish the influence of the hydrodynamic conditions, thus and firstly, it was investigated the influence of the stirring speed, applied to the feed phase, on the overall mass transfer coefficient. The transport of iron(III) across the supported liquid membrane was dominated by diffusional resistances which can be of two types: i) one associated to the feed phase boundary layer, and ii) other associated to the membrane support. It is relatively usual that the magnitude of the first competed with the value of the support resistance [19].

Irond(III) transport through the supported liquid membrane
In the present work, stirring of the feed solution was carried out from 500 to 1500 min -1 (Table 1). It can be seen that the overall mass transfer value increasing with the increase of the stirring speed up to 1250 min -1 , and then remained constant. Thus, at 1250 min -1 , the thickness of the aqueous feed boundary layer reached a minimum, and iron(III) transport maximized. Also, from the results showed in this Table 1, it can be observed that iron(III) recovery in the receiving phase was practically quantitative; the transport of Mn(II) can be considered as negligible (see Section 2.2.2.), thus, a t a first instance, the present system (carrier phase and support) can be used to separate Fe(III) from Mn(II). It can be also worth to note here that the pH of the receiving phase became more acidic (from pH 5 to pH 1.5-2) as the time was elapsed, this result can be attribute to the cotransport of HCl from the feed to the receiving phase (see Section 3.3.).
3.3.2. Influence of the stirring speed applied on the receiving phase on iron(III) transport The influence of the stirring speed, on iron(III) transport, applied to the receiving phase was also investigated using the same experimental conditions showed in Table 1, but using a stirring speed in the feed phase of 1250 min -1 and varying the stirring speed in the receiving phase from 500 to 1000 min -1 The results showed that this variation had not influence on iron(III) transport and also in the recovery of the element in this phase. This behaviour was attributable to that if the stirrer in the receiving phase half-cell was very close to the membrane support, the thickness of the boundary layer was considered to be minimized and the resistance in this side can be neglected [20].

Influence of the HCl concentration in the feed phase on iron(III) transport
The influence of the variation of the acid concentration in the feed phase on iron(III) transport was investigated using the same experimental conditions showed in Table 1, but with acid concentrations ranging from 1 to 8 M. In these series of experiments the stirring speeds applied on the feed and receiving phases were of 1250 and 500 min -1 , respectively. The results from these series of experiments were summarized in Table 2. It can be seen that iron(III) transport increased when the HCl concentration increased in the 1-4 M range, and then decreased at HCl concentrations of 6 and 8 M; this decrease can be explained due to the competitive transport of HCl across the supported liquid membrane impregnated with Cyanex 923 dissolved in Solvesso 100, it was shown [21], that Cyanex 923 extracted mineral acids from aqueous solutions. In the case of manganese(II), negligible transport of this element was detected along the experiments. Previously, it was detected that the solid support impregnated only with Solvesso 100 did not transport neither iron(III) or manganese(II). Thus, the presence of the extractant or carrier was essential to achieve metal transport from the feed to the receiving solution. This influence was investigated using a feed phase of 0.01 g/L Fe(III) and 0.17 g/L Mn(II) in 4 M HCl medium, whereas the receiving phase was of distilled water; in these series of experiments, the extractant concentration was varied from 1.3 to 40% v/v (0.03-1 M) dissolved in Solvesso 100. The results of these experiments were summarized in Table 3. Results showed in the above Table 3 indicated that there was an increase of iron(III) permeation with the increase from 1.3 to 10% v/v Cyanex 923 concentration in the carrier solution, and then a decrease in iron(III) transport resulted when higher Cyanex 923 concentrations (20-40% v/v) were used. This decrease, at a first instance, can be attributed to the increase of the carrier phase viscosity, which resulted in lower values of the overall mass transfer coefficient. In the case of manganese(II), negligible transport was observed, even using the most concentrated Cyanex 923 solution. At 10% v/v Cyanex 923 in Solvesso 100 concentration, a maximum in iron(III) transport was attained, this maximum or limiting mass transfer coefficient (Klim) was explained by the assumption that in this experimental condition, the diffusion in the membrane (Δm) was negligible in comparison with feed diffusion (Δf) , and the transport process was entirely controlled by the diffusion in the film of the feed solution, thus: In the above equation, Df = 10 -5 cm 2 /s represented the average diffusion coefficient in the aqueous feed phase, and df to the thickness of the feed phase boundary layer. Accordingly, in the present system df was calculated as 2.9x10 -3 cm, being this value the minimum thickness of the boundary layer under the present experimental conditions. 3.3.5. Influence of manganese(II) concentration in the feed phase on iron(III) transport The possible variation of manganese(II) concentration in the feed phase on iron(III) transport was also investigated. The carrier phase was of 10% v/v Cyanex 923 in Solvesso 100 impreganting Durapore GVHP4700 support, whereas the receiving solution was water. The aqueous feed phase contained 0.01 g/L Fe(III) in 4 M HCl medium and varying Mn(II) concentrations from nil to 0.17 g/L. The results indicated that the presence of Mn(II) in the solution had not detrimental effect on iron(III) transport (KFe= 3.2±0.2 cm/s), thus, crowding or population effect [22], due to the presence of Mn(II) in the solution, was negligible on iron(III) permeation under these experimental conditions.

Estimation of diffusional parameters and evaluation of mass transfer resistances
The permeation rate of iron(III) was estimated by the rates associated to the diffusion of Fe(III)-bearing species through the feed phase diffusion layer and the diffusion of the Fe(III)-Cyanex 923 species through the membrane; Fick´s first diffusion law was applied to the diffusion layer in the feed and the membrane sides [23]. The fluxes in each of the above phases can be represented by: where [Fe]f,0 was the initial iron concentration in the feed phase, [Fe]f,i was the iron concentration at the feed phase/membrane interface, and [HFeCl4L4 -]f,i and [HFeCl4L4]r,I were the concentrations of the respective species at the feed phase/membrane interface and at the membrane/receiving phase interface, respectively.
Usually, the concentration of the metal-organic ligand species (HFeCl4L4 in the present investigation) in the membrane phase at the receiving phase side was negligible when compared with the corresponding one at the feed phase side, thus, eq.(9) was rewritten as: It can be considered that the chemical reaction (eq.(3)) was fast if compared to the diffusion rate, thus, local equilibrium at the feed phase-membrane interface was attained and concentrations in this interface was related by eq.(4). At steady state, J=Jf=Jm, being J expressed as:   where Δf and Δm were the transport resistances due to diffusion by the feed phase boundary layer and the membrane, respectively. This last equation combined in one expression the equilibrium and diffusional parameters involved in the iron(III) transport though a flat-sheet supported liquid membrane containing Cyanex 923 in Solvesso 100 as carrier phase.
The resistances to the mass transfer were evaluated by the use of eq.(12), from this equation the next expression can be derived: was estimated as 3.2x10 -4 cm 2 /s; this value was greater than the value of the diffusion coefficient in the membrane phase (eq.(14)), this was attributable to the diffusional resistance caused by the membrane thickness, separating the feed and the receiving phases, In eq.(15), τ was the membrane tortuosity (1.67) and ε was the support porosity (75).
Assuming that the extractant concentration in the membrane phase was constant, an apparent diffusion coefficient for iron (III)-Cyanex 923 dissolved in Solvesso 100 species was estimated as: where J was calculated accordingly to eq.(11) and an initial iron(III) concentration in the feed phase of 0.01 g/L; considering an extractant concentration of 10% v/v Cyanex 923 in Solvesso 100, this apparent coefficient van be calculated as 3x10 -8 cm 2 /s. Since the overall mass transfer resistance was the result of the sum of the different resistances participating in the permeation process, eq.(13) can be rewritten as: where Rf and Rm were the resistances in the feed and the membrane phases, respectively. Table 4 showed the contribution of these resistances to the iron(III) transport.  Furthermore, the fractional resistance of each step (%Rf 0 and %Rm 0 ) to the transport process was calculated under various experimental conditions, as it can be seen from the results presented in Table 4, in almost all the conditions, diffusion by the aqueous feed boundary layer was the controlling step to the iron(III) transport process.

Conclusions
The use of Cyanex 923, dissolved in Solvesso 100, coupled to a solid supported liquid membrane (flat-sheet mode) technology, allowed to the selective separation of Fe(III) over Mn(II) from solutions derived in the treatment of spent alkaline batteries. A kinetic model for Fe(III) transport is developed and finding that Fe(III) transport is mainly controlled by diffusion of the FeCl4 -anion across the feed aqueous layer, having a contribution, under certain experimental conditions, the diffusion by the membrane of the neutral complex formed by the above iron(III) species and Cyanex 923.