F. J. ALGUACIL*, M. ALONSO, F. A. LOPEZ, A. LOPEZ-DELGADO AND I.
PADILLA
Centro Nacional de Investigaciones Metalurgicas (CSIC), Avda. Gregorio del Amo 8,
Ciudad Universitaria, 28040 Madrid, Spain. *Corresponding author e-mail: fjalgua@cenim.csic.es
______________________________________________________________________
The use of dispersión-free solvent extraction, through microporous hydrophobic membrane has been investigated. The hollow fiber contactor, with surface area of 1.4 m 2 was used to extract Cr(VI) (0.005-0.12 g/L) from aqueous acidic media. Several parameters such as extractant concentration, feed acidity and metal concentration in the initial aqueous solution were investigated. Results revealed that 15 % v/v Cyanex 923 in
Exxsol D-100 as organic phase and feed in the 2.5 pH range, gave optimum extraction
(exceeding 95%) of Cr(VI) and it was possible to strip using 10 g/L hydrazine sulphate.
In this step, Cr(VI) is immediately reduced to the less hazardous Cr(III) state. Results also showed that under the various experimental conditions, chromium (VI) extraction was rate-controlled by the interfacial reaction on the membrane surface.
Introduction
Though solvent extraction and other separations technology are of a wide potential use, some drawbacks presented by them, e.g. unsuitability of solvent extraction to treat dilute solutions, lack of selectivity of ion-exchange or adsorption processes, both common situations when liquid effluents are considered, have led to liquid membranes
(LMs) technologies to be the ideal choice to use when the treatment of such dilute and multielemental solutions is under way.
Thus, membrane-based solvent extraction (MBSX) techniques had shown their potentials and applications for the separation/removal of metals, and other solutes, from a range of diverse matrices.
Several MBSX applications and considerations about their usage and performance can be found elsewhere (1-4) . In dispersion-free solvent extraction (DFSX) using a hollow fiber contactor, the aqueous and organic phases flow (often) through the tube and shell

2 sides of the fibers and come into contact at the pore mouth without dispersion, which minimizes the possibility of emulsion, third phase and/or even crud formation during the process. In DFSX operation using a unique hollow fiber module, only one separation (extraction or stripping) is carried out. To obtain both steps together, an integrated process is used with two hollow fiber modules, one for extraction and one for stripping.
Being chromium a wide use metal, potential exposure to chromium compounds occurs in many environments and their produced effluents. Chromium (VI) compounds are considered to be carcinogenic (5) , thus, being their removal from chromium (VI)bearing effluents a prior target.
In view of the above, a comprehensive program for chromium (III) or (VI) removal from various aqueous media by DFSX with a hollow fiber contactor was developed.
Experiments were conducted for Cr(VI) extraction from aqueous acidic solutions using
Cyanex 923 in Exxsol D100 as organic solution and the investigation includes the effect of various hydrodynamic and chemical parameters which should influenced the process.
Also, reducing stripping of Cr(VI), to the less hazardous Cr(III) oxidation state, with hydrazine sulphate, had been investigated.
Experimental Section
A stock solution of Cr(VI) was prepared from K
2
Cr
2
O
7
AR grade (Merck). Cyanex 923, which is composed by four phosphine oxides (average molecular weight 348) was obtained from CYTEC, Canada; the organic diluent used in DFSX studies was Exxsol
D100, which is a commercially available aliphatic diluent from ExxonMobil Chem.

3
Iberia, Spain. Both Cyanex 923 and Exxsol D100 were used as received. All other chemicals used in the investigation were of AR grade.
The hollow fiber module was manufactured by Hoechst Celanese (now Membrana,
USA): Liqui-Cel 8x28 cm 5PCG-259 module and 5PCS-1002 Liqui-Cel Laboratory
LLE; details of module and hollow fiber membrane specifications are given in Table 1.
A schematic view of the membrane-based Cr(VI) extraction process (DFSX) with
Cyanex 923/Exxsol D100 using a hollow fiber contactor in recirculation mode is shown in Figure 1. The aqueous and organic phases were contacted counter-currently in a hollow fiber module for extraction or stripping runs in recirculating mode. In the extraction module, the feed aqueous phase flows through the tube side of the fibers, while the organic phase circulates through the shell side, wetting the wall of the hydrophobic fibers. The pressure in the feed phase was held 0.2 bar higher than the pressure in the organic phase in order to maintain the interface at the pores of the fiber.
In the stripping run, Cr(VI)-loaded organic extractant flowed through the shell side, whereas hydrazine sulphate solutions flowed through the tube side counter-currently in recirculation mode.
The organic phase is a 400 mL solution of Cyanex 923 in Exxsol D100. 4 L of aqueous feed solution with the desired concentration of Cr(VI) was prepared by taking a suitable aliquot from the stock solution. The desired feed pH was reached by adding
H
2
SO
4
or NaOH solution. 400 mL of hydrazine sulphate solution was used as the stripping phase when stripping of Cr(VI) was carried out in the hollow fiber module.
Small aliquots of the phases were taken at selected times for the analysis of metal concentration by standard atomic absorption spectrometry (Perkin Elmer 1100B

4 spectrophotometer, England), and the overall mass transfer coefficient K was computed using the next relationship (6) : ln
 
  t
0
 
A
V

K
 t (1) where [Cr] t
and [Cr]
0
are the chromium concentration in the feed phase at a time t and time zero, respectively; A is the membrane effective area, V is the volume of feed solution and t the time.
Results and Discussion
The mechanism for extracting chromium (VI) with Cyanex 923 in the hollow fiber module is schematically view in Figure 2. The next assumptions were made: i) the system is at steady state, ii) the pore size and wetting characteristics are uniform throughout the membrane, iii) mass transfer is described by simple film-type mass transfer coefficients, iv) no solute transport occurs through the non-porous part of the membrane, v) the solutions are virtually insoluble in each other.
In DFSX, the system consists of an aqueous phase containing Cr(VI), mainly as
HCrO
4
, flowing through the tube side of microporous hollow fiber membranes whose pores are filled with the organic phase (Cyanex 923 dissolved in Exxsol D100), and this phase was flowing counter-currently through the shell side (Fig. 1). The extraction reaction takes place at the inside wall of the membrane where the phase interface is located (Fig. 2). To determine the rate-controlling step it was considered that extraction involves various sequential steps: the effective rate of the interfacial reaction at surface, mass transfer from the bulk aqueous feed to the inside surface of the hollow fiber membrane, diffusion across the porous membrane and mass transfer into the organic solution surrounding the fiber.

5
Considering a resistance in the series model, the reciprocal of the overall mass transfer coefficient is the total resistance to mass transfer and can be described as the sum of the mass transfer resistances inside the fibers (feed phase), across the fiber wall (membrane) and outside the fibers (organic solution). Therefore, the expression for the overall mass transfer coefficient, K, can be expressed as:
1
K

1 k i

1 k aq
 d i d lm
1
D
Cr k m
 d d o i
1
D
Cr k org
(2) where k i
is the effective rate of interfacial reaction at the surface; k aq
, k m
and k org
are the mass transfer coefficients in the feed, membrane and organic phases, respectively, and d lm
is the logarithmic mean fiber diameter. On the other hand, D
Cr
is the chromium distribution coefficient:
D

Cr
 
  org aq
(3) defined as the ratio of the total analytical chromium concentrations in the organic phase and aqueous phase. D
Cr
appears in the membrane resistance, because in the present system the organic phase wets the membrane but water does not.
Influence of the Flow Rate in the Feed and Organic Phases.
As seen in Table 2, the overall mass transfer coefficient increased with the feed flow rate and its value was maximum at 25 cm
3
/min. The decrease in the K value at higher flows can be due to lower residence time of the solution in the module, which caused incomplete loading of the carrier.
Table 3 shows the effect of changing the shell side (organic phase) flow rate keeping the aqueous flow constant for 30 % v/v Cyanex 923/Exxol D100. Module extraction efficiency decreased below 50 cm
3
/min of organic flow, and also for higher flows, being the reason for this decrease probably the same as above.

6
Influence of the pH in the Feed Phase.
The effect of pH on the overall mass transfer coefficient is presented in Table 4. The increase in the feed pH value causes a decrease in the value of the mass transfer coefficient, this is attributable to that in the extraction reaction:
H
 aq

HCrO

4 aq
 nL org

H
2
CrO
4
 nL org
(4) the driving force for Cr(VI) transport is the co-transport of protons, thus, the overall mass transfer coefficient decreased to 3.9x10
-6
cm/s at pH 4.2±0.05, as suggested by the above equation.
Influence of Cr(VI) Concentration.
The effect of the Cr(VI) concentration in the feed phase on the overall mass transfer coefficient is presented in Table 5. The K value was observed to decrease with an increase in the chromium concentration during experiments in counter-current mode (aqueous flow at 25 cm
3
/min). Probably at higher chromium concentrations the organic phase within the membrane pores get saturated with metal complex on increasing the metal concentration in the aqueous solution.
Further, the metal-organic complex diffuses slowly into bulk of organic phase which in nutshell decreased the mass transfer in the organic phase.
Influence of the Ionic Strength of the Feed Phase. The overall mass transfer coefficient of chromium (VI) is also dependent of the ionic strength of the feed solution
(results in Table 6), with the K value plummeted from 4.2x10
-5 to 3.5x10
-5 cm/s when I in the feed solution increased from 0 to 2 (chloride medium). The difference between sulphate and chloride media should be attributable to the presence of less extractable chromium-chloro complexes in the feed phase at these chloride concentrations in the aqueous solution.

7
Influence of Carrier Concentration.
The effect of Cyanex 923/Exxsol D100 concentration on Cr(VI) transport was also investigated. Experimental conditions were established as organic phase with various concentrations of Cyanex 923 in the organic diluent, and feed concentration of 0.03 g/l Cr(VI) at pH 2.5±0.05. Results are shown in
Table 7, the overall mass transfer coefficient was seen to increase with Cyanex 923 concentration of up 15% v/v and the transport rate should be therefore limited by diffusion through the aqueous film on the feed side of membrane in this region.
Influence of Operational Mode.
Table 8 indicates that the counter-current mode was more efficient for efficient mass transfer of Cr(VI) than the co-current mode. However, it may be consider that K was found to be of the same order of magnitude irrespective of the contacting mode.
Estimation of Local Resistances.
The first term in eq. (2) represents the local value of the total resistance. This resistance is in turn the sum of the resistance owing to interfacial reaction and the mass transfer resistances inside the fiber, the membrane and outside the fiber. The mass transfer coefficient inside the fiber can be calculated from the next equation: k aq

1 .
62
D aq d i d i
2 v aq
D aq
L
1
3
(5) where D aq
is the diffusivity of chromium (VI) species in the feed solution (estimated as
10
-5
cm
2
/s (7) ), v aq
is the mean velocity of the feed phase and L is the fiber length. The k aq
value calculated from the above equation was 4.8x10
-4
cm/s for v aq
of 25 cm
3
/min.
The membrane mass transfer coefficient does not depend on the hydrodynamics; it is related to the membrane properties and to the diffusion coefficient of the solute in the

8 organic phase immobilized in the pores. Then, an estimate of the membrane mass transfer coefficient can be determined by the following equation: k m

D org

 
 d o

2 d i


(6)
In the present case, k m
is 3.3x10
-5 cm/s, for an estimate D org
value of 10 -6 cm 2 /s.
For the estimation of the shell side mass transfer coefficient, the next equation was used: k org

D org
0 .
245
2 r shell


2 r shell v org



2
3
 

D org
1
3
(7) where v org
is the flow velocity of the organic phase in the shell side of the contactor and
μ is the viscosity of the organic phase. The value of k org
is estimated as 6x10
-6
cm/s.
From data presented in this work (Table 9), it can be considered that the rate-controlling step was the interfacial reaction on the membrane surface, except at higher pH values in which the resistance in the shell side becomes dominant. In the above Table, the fractional resistance of each step in the overall process (R i
0
, R aq
0
, R m
0
and R org
0
) could be calculated by the following equation, for example, R aq
0 is defined as:
R
0 aq

R aq
R
(8) where R aq
is the mass transfer resistance due to feed, and R is the overall mass transfer resistance.
Stripping Operation.
Stripping was performed separately in a single module by circulating loaded organic solution (Cyanex 923 in Exxsol D100 with Cr(VI)) on the shell side and hydrazine sulphate solution on the tube side. In the stripping process, chromium (VI) is already reduced to the cationic Cr
3+
species, which is not extracted by

9
Cyanex 923. The overall mass transfer coefficient K for stripping ranged from 6.2x10
-6 to 3.5x10
-6
cm/s. The flow rate of the stripping solution influenced the mass transfer as this variable increased, as shown in Table 10. The values of K were found to be of the same order of magnitude regardless of the contacting mode and the concentration (5-10 g/L) of hydrazine sulphate for stripping experiments.
The rate of the stripping reaction at the surface was not considered because the reaction is instantaneous and fast. The following equation could be used to estimate the overall mass transfer resistance for stripping:
1
K

1 k s
D
´
Cr
 d i d lm
1 k m
 d d o i
1 k org
(9) where k s
, k m
and k org
are the mass transfer coefficients for the tube side, the membrane and the shell side. D
Cr
´
is the chromium distribution coefficient for stripping, defined in a similar way that for extraction. For chromium extraction, the distribution coefficient appears in the membrane and organic terms, but it does not for chromium stripping.
This is due of the wetting situation of the membrane. The membrane is wetted by the organic phase in the case of extraction, having this phase a high solubility for the extracted species. In stripping operation, the membrane is also wetted by the organic solution, but in this case the organic phase has no affinity for the solute.
Equation (9) indicates that the overall resistance for stripping is the sum of local values of the individual resistances. The value of k org
for organic phase, dependent of the hydrodynamic conditions, varied between 2.9x10
-6 and 4.6x10
-6 cm/s, whereas k m was 3.3x10
-5
cm/s, as determined earlier. The value of k s
was in the order of 7x10
-4 cm/s, calculated from eq. (5). Then, on the right hand side of eq. (9), the first term is in the 10
3
s/cm, or less, order; the second term is around 3.4x10
4
s/cm, and the third term is in the 1.7x10
5
-2.8x10
5
s/cm range. The first and second terms are thus negligible. The

10 experimentally estimated overall resistance is around 1.6x10
5
-2.9x10
5
s/cm (as deduced from Table 10), which showed that the resistance due to organic mass transfer is apparently dominant.
Acknowledgements
We thank the CSIC for support.
Literature Cited
(1) Kumar Pabby, A.; Sastre, A. M. In Ion Exchange and Solvent Extraction, Vol. 15 ;
Marcus, Y., Sengupta, A. K., Eds.; Marcel Dekker: New York, 2002; p 331.
(2) Cox, M. In Solvent Extraction and Liquid Membranes (Fundamentals and
Applications in New Materials) ; Aguilar, M., Cortina, J. L., Eds.; CRC Press: Boca
Raton, 2008; p 1.
(3) Kumar Pabby, A.; Sastre, A. M. Ibid ; p 91.
(4) Ortiz, I.; Irabien, J. A. Ibid ; p 201.
(5) Casadevall, M.; Kortenkamp, A. In Heavy Metals in the Environment ; Sarkar, B.,
Ed.; Marcel Dekker: New York, 2002; p 271.
(6) Winston Ho, W. S.; Wang, B. Ind. Eng. Chem. Res.
2002, 41, 381.
(7) Alguacil, F. J.; Alonso, M.; Lopez, F.A.; Lopez-Delgado, A. Chemosphere 2008, 72,
684.

11
______________________________________________________________________
TABLE 1. Module and Fiber Characteristics
Module
Length
Diameter
Nº of fibers
28 cm
8 cm
10000
Fiber material
Effective membrane area polypropylene
1.4 m
2
Fiber
Length (L)
Internal diameter (d i
)
Outer diameter (d o
)
Porosity (ε)
Tortuosity (τ)
15 cm
24x10
30x10
30%
3
-3
-3
cm
cm
_____________________________________________________________________
______________________________________________________________
TABLE 2. Variation of Overall Mass Transfer Coefficient with Feed Flow feed flow K
(cm 3 /min)
10
(cm/s)
3.5x10
-5
25
50
4.2x10
-5
3.9x10
-5
100 3.6x10
-5
Feed phase: 0.03 g/L Cr(VI) at pH 2.5±0.05. Organic phase: 30% v/v Cyanex 923 in Exxsol D100.
Organic flow: 50 cm 3 /min.
______________________________________________________________________

12
______________________________________________________________________
TABLE 3. Influence of Organic Phase Flow on the Overall Mass Transfer
Coefficient organic flow
(cm 3 /min)
25
50
100
K
(cm/s)
3.5x10
-5
3.9x10
-5
3.6x10
-5
3.2x10
-5
200
Feed phase: 50 cm 3 /min. Other variables as in Table 2.
______________________________________________________________________
______________________________________________________________________
TABLE 4. Influence of Feed pH on the Overall Mass Transfer Coefficient pH±0.05
K
2.5
3.5
(cm/s)
4.2x10
-5
1.6x10
-5
3.9x10
-6
4.2
Feed phase: 0.03 g/L Cr(VI) at various pH. Organic phase: 30% v/v Cyanex 923 in Exxsol D100. Feed flow: 25 cm 3 /min. Organic flow: 50 cm 3 /min.
______________________________________________________________________

13
______________________________________________________________________
TABLE 5. Overall Mass Transfer Coefficients at Various Initial Cr(VI)
Concentrations in the Feed Phase
Cr(VI) K
(g/L)
0.005
0.03
0.12
(cm/s)
4.5x10
-5
4.2x10
-5
2.9x10
-5
Feed phase: Cr(VI) solutions at pH 2.5±0.05. Organic phase: 30% v/v Cyanex 923 in Exxsol D100. Flows as in Table 4.
______________________________________________________________________
______________________________________________________________________
TABLE 6. The Influence of Feed Ionic Strength on the Overall Mass Transfer
Coefficient
I K
0 (no salt addition)
(cm/s)
4.2x10
-5
3.5x10
-5 1 (chloride medium)
2 (chloride medium) 3.5x10
-5
3 (sulphate medium) 3.9x10
-5
Feed phase: 0.03 g/L Cr(VI) at pH 2.5±0.05 and added salt (LiCl for chloride medium and Li
2
SO
4
for sulphate medium). Organic phase: 30% v/v Cyanex 923 in Exxsol D100. Flows as in Table 4.
______________________________________________________________________

14
______________________________________________________________________
TABLE 7. Influence of Extractant Concentration on the Overall Mass Transfer
Coefficient
[Cyanex 923] K
(% v/v)
3.75
7.5
15
30
60
(cm/s)
8.9x10
-6
1.4x10
-5
4.2x10
-5
4.2x10
-5
4.2x10
-5
Flows as in Table 4.
______________________________________________________________________
______________________________________________________________________
TABLE 8. Effect of Operational Mode on Cr(VI) Transport operational mode K counter-current
(cm/s)
3.5x10
-5
2.9x10
-5 co-current
Feed phase: 0.03 g/L Cr(VI) at pH 2.5±0.05 and 2M LiCl. Organic phase: 30 % v/v Cyanex 923 in Exxsol
D100. Flows as in Table 4.
______________________________________________________________________

15
______________________________________________________________________
TABLE 9. Fractional Resistance of Mass Transfer of Cr(VI) in Hydrophobic
Hollow Fiber Membrane Extraction
Cr(VI) Cyanex 923 pH±0.05
R aq
0 R m
0 R org
0 R i
0
(g/L)
0.03
(% v/v)
3.75
7.5
15
2.5
2.5
(%)
2.0
2.9
8.7
(%)
8.8
1.8
0.7
(%)
35.5
10.9
4.3
(%)
53.7
84.4
86.3
0.005
0.03
30 9.1
8.7
0.09
0.09
0.6
0.5
90.3
90.7
0.12
0.03 30 3.5
4.2
5.9
3.2
0.8
0.06
3.2
17.9
0.4
12.9
72.1
93.4
80.7
9.2
Subscripts aq, m, org and i corresponded to terms: aqueous phase, membrane, organic phase and interfacial, respectively.
______________________________________________________________________
______________________________________________________________________
TABLE 10. Overall Mass Transfer Coefficients for Stripping of Cr(VI)-loaded
Cyanex 923 in Exxsol D100 using the DFSX technique contact mode co-current strip solution flow rate
(cm 3 /min)
50
100 organic flow rate
(cm 3 /min)
50
50
K
(cm/s)
6.2x10
-6
3.5x10
-6
3.6x10
-6 counter-current
Strip phase: 10 g/L hydrazine sulphate.
100 50
______________________________________________________________________

16
FIGURE 1. Schematic view of DSFX plant with one hollow fiber module in the recirculation and counter-current mode. 1. Membrane module. 2. Feed reservoir tank. 3. Organic reservoir tank. In the stripping operation, 2 acts as stripping phase reservoir tank.
FIGURE 2. Schematic concentration profile for DFSX operation. L= carrier. H + acts as counter-ion for Cr(VI) transport.

FIG.1
17

FIG.2
18