Journal of Membrane Science, 81 (1993) 121-126 Elsevier Science Publishers B.V., Amsterdam

121

Permeation of neodymium and praseodymium through supported liquid membranes containing di- (Z-ethylhexyl)phosphoric acid as a carrier

Carlos Moreno, Aleci HrdliEka* and Manuel Valiente** Quimica Analitica, Universitat Authoma de Barcelona, 08193 Belluterra, Barcelona (Spain)

(Received September 17,1992; accepted in revised form February 1,1993)

Abstract

Transport of Pr (III) and Nd (III) cations between nitrate feed and 0.1 mol/l nitric acid strip solutions through a supported liquid membrane has been studied as a function of both the stirring rate and the chemical composition of the system. The liquid membrane consistedof di- (2-ethylhexyl)phosphoric acid (DEHPA) in kerosene immobilized on a porous di-fluoropolyvinylene laminar support. Reuse of the membrane has been investigated. However, good results were obtained only by reimpregnating of the support. Metal permeability increases sharply with carrier concentration up to 0.1 mol/l DEHPA and thereafter becomes almost independent of carrier concentration. The dependence of permeability coef- ficient, P, vs. feed pH, measured at low carrier concentration, exhibit maxima at a pH of ca. 3. The values for the membrane separation factor, (Y = PNd/PPr, of 1.25 and 1.5 have been determined at pH 2.7 and 2.5, respectively.

Keywords: supported liquid membranes; di- (2-ethylhexyl) phosphoric acid; neodymium; praseodymium; nitrate media; fundamental study

Introduction

Liquid membranes (LMs) have been devel- oped and studied for the separation of toxic or valuable metals including rare earth elements (REE). The technique, based on a liquid-liq- uid distribution processes, has been imple- mented by using extracting agents as carriers for facilitated transport.

For the solvent extraction of REE, reagents of different nature have been employed, i.e. sol-

*On leave from the Department of Analytical Chemistry, Masaryk University, Kotlz.Gska 2, 61137 Brno, Czech Republic. To whom correspondence should be addressed.

vating extractants like tri-n-butylphosphate (which has been applied to recover REE from nitrate media [ 1 ] ), cation-exchange extrac- tants including organophosphoric [ 21, phos- phonic and phosphinic acids [ 31 or carboxylic acids [ 41, and anion-exchange extractants as e.g. quaternary ammonium salts [5]. Among these extracting reagents, organophosphorus ones have been widely studied for REE extrac- tion, especially di- (2-ethylhexyl)phosphoric acid (DEHPA) [ 6-81. The well-known extrac- tion reaction of lanthanides with DEHPA (HA) proceeds according eqn. (1) :

Ln3++3(HA),*LnA3(HA)3+3H+ (1)

0376-7388/93/$06.00 0 1993 Elsevier Science Publishers B.V. All rights reserved.

122 C. Moreno et al./J. Membrane Sci. 81(1993) 121-126

where bars denote the species present in the or- ganic phase.

Because of the different level of dimerization of dialkylphosphoric acids in different sol- vents, extraction constants of the individual elements are strongly dependent of the solvent character. The stronger interaction of the rel- atively more polar monomer with the solvent favors the dissociation of the dimeric species. Thus, the highest values for both dimerization and extraction constants are found in nonpolar solvents, as e.g. tetrachlormethane, toluene, and kerosene. On the other hand, in our previous works dealing with LMs, kerosene has been shown to be an excellent solvent from the point of view of LM stability. Currently, the use of different dialkylphosphoric acids for mem- brane transport of REE is under study.

Hirashima et al. [9] investigated the effect of experimental conditions on the transport rate of Pr (III) cations through supported liquid membranes (SLM) containing DEHPA dis- solved in Shellsol 71, a 100% paraffin solvent. This LM, immobilized on a polypropylene sup- port (thickness 25 ym, nominal porosity 38%, pore size 0.02 pm) was placed between HCl so- lutions. The ,highest values for the transport rate (max. ca. lo4 mol-sec-1-m-2 evaluated from 8-hr measurements) were found at 1 mol/ 1 DEHPA in LM and 3 mol/l and 0.1 mol/l HCl [ 0.25 mol/l for Sm (III) ] in the strip and feed solution. The concentration of Pr (III) cations in the feed solution was relatively high, mostly at 0.04 mol/l. A different ionic strength has been used in aqueous phases and it was not maintained constant. The transport rate was found to be practically independent of the stir- ring rate of the strip and feed solutions.

The retardation effect of acetate ions on the permeation of Eu(II1) cations through SLM and DEHPA in kerosene has been studied re- cently [lo]. Permeation of SC, Y, La, Sm, Gd, and Er from nitric acid media through liquid

emulsion membranes containing DEHPA has been also reported [ 111.

This work has been undertaken to character- ize the permeation of praseodymium (III) and neodymium (III) cations through SLMs with DEHPA, by a systematic study of both the hy- drodynamic conditions and the chemical com- position of the system.

Experimental

Reagents and solutions

Stock solutions of 2.5 mmol/l Pr(II1) or Nd( III) in 0.1 mol/l HNO, were prepared by dissolving their pure or analytical grade ni- trates (Fluka, Switzerland) and standardized chelatometrically.

Di(2-ethylhexyl)phosphoric acid, general purp, reag., (BDH Chemicals, UK), Xylenol Orange (Panreac, Spain ) and cetylpyridinium bromide monohydrate, pure (Aldrich, USA) were used as received. Stock solutions of DEHPA were prepared in kerosene (Petronor, Spain) previously purified by successive wash- ing with 0.5 mol/l NaOH, water, and 0.5 mol/l HCl followed by final washing with water until the chloride test was negative. Ammonia was distilled isothermally and water was redistilled in a quartz still. All other reagents were of an- alytical grade purity and purchased from Pan- reac, Spain.

Feed solutions containing Pr or Nd at initial concentrations of 4-5 &ml were prepared from their stocks and acidity was adjusted with di- luted nitric acid or sodium hydroxide. Ionic strength was kept constant at 0.1 mol/l ( HN03, NaNO,). As a strip solution, 0.1 mol/l nitric acid was employed.

Procedure and apparatus

Durapore GVHP 047 00 (Millipore, USA), a laminar microporous di-fluoropolyvinylene film of 125 ,am thickness was used as a solid support. Nominal porosity of the support was 75% and

C. Moreno et al./J. Membrane Sci. 81 (1993) 121-126

effective pore size was 0.22 pm. In membrane experiments, the effective surface area of the supported liquid membrane (SLM) was 8.51 cm’. This value was determined as the product of the nominal porosity coefficient (0.75) and the membrane surface area being in contact with the solutions.

The SLM was prepared by impregnating the support with a solution of DEHPA in kerosene and a 1-min treatment in an ultrasonic bath (model 1200, Branson Ultrasonics, USA). Subsequently, the membrane was gently dried between filter paper and placed in the trans- port cell (described elsewhere [ 121) between the two compartments for feed and strip solu- tions, 200 ml each. In membrane experiments, the two solutions were stirred at the same speed, varying from 350 to 1700 rpm. For a new ex- periment, the membrane was gradually rinsed with water, dried between filter paper, purified in kerosene, dried again, and reimpregnated with DEHPA solution. If not used, the mem- brane was stored in kerosene. Significant dif- ferences in permeability were found when dif- ferent sheets of the described membrane support were used. For this reason, the proce- dure took into account reuse of the membrane support.

The variations in concentration of the lan- thanide cations in the feed solution with time were monitored by an on-line FIA spectropho- tometric method [ 131. Membrane experiments were performed at ambient temperature of 23+1”C.

Acidity was measured by means of a model 506 pH-meter (Crison, Spain) equipped with a U455-S7 glass-Ag/AgCl combined electrode (Ingold, Switzerland) using standard buffers of pH 4.0 and 7.0 (Crison, Spain) for calibration.

Feed solution acidity was monitored during the permeation experiments. Because the total amount of metal transported was relatively small, the pH values lower than 3.1 were nearly constant. At higher pH values, lanthanide

123

permeability decreased and the observed pH changes were smaller than 0.1 pH unit.

Results and discussion

Mass transport of lanthanide cations has been expressed in terms of permeability coef- ficient P (cm/min). The values were deter- mined from sets of primary data in the form of lanthanide concentration vs. time by using the following linear relationship [ 141:

Q -ln[M],=VPt-ln[M]f,o

f

where Q is the effective membrane area, Vf is the volume of feed solution, and [Ml,, and [M ] f are the feed metal concentrations at time zero and time t, respectively.

First membrane stability was investigated. Permeability coefficients were determined in 50-min measurements by reuse of the same membrane without reimpregnation (feed pH 3.0, SLM containing 25 mmol/l DEHPA in kerosene, stirring rate 1300 rpm). A linear, 30% decrease in the P values was observed during six successive runs in which the initial metal concentration in the feed solution was kept constant while the strip solution was not changed. If a fresh strip solution was applied in each experiment, the P values decreased sharply from the 2nd to the 3rd experiment and only 40% of the original P value was obtained in the 4th run. The observed decrease in permeability may be due to a washing out of the carrier so- lution from the SLM.

However, successive experiments with reim- pregnation of the membrane gave values for the permeability coefficients of acceptable repeat- ability. On the other hand, a comparison of the P values obtained from data collected during 30 min or during the entire 50-min period reveals that these are practically identical. Thus, to avoid the decrease in membrane transport ca- pability, in the following measurements the

124

lanthanide concentration was monitored only during a 30 min period. At these conditions, the differences between the individual P values de- termined in six successive measurements were smaller than If: 2.8% rel.

It is known that extraction of REE with DEHPA is a rapid process. Considering that the volume of an LM is very small in comparison with the feed solution volume, the membrane would be saturated soon. However, using 25 mmol/l DEHPA and ca. 30 pmol/l lanthanide solution, a linear relationship between In [M] f and time was observed during a 90-min mea- surement period. This can be explained by si- multaneous transport of metal ions across the membrane and their release into the strip so- lution. Direct evidence of lanthanide transport has been obtained by ICP-AES analysis of some of the strip solutions after finishing the experiment.

The dependence of permeability on stirring speed is shown in Fig. 1. Permeability of Pr and Nd cations increases with stirring rate in a sim- ilar manner for both elements and becomes in- dependent on stirring speed above 1300-1500 rpm. Thus at a sufficient by high stirring speed,

0.1 ‘,*““““‘,‘,,,“” 0 1000

wm 2000

Fig. 1. Effect of stirring rate on permeability coefficients. Feed: Pr at 31.2 or Nd at 31.9pmol/l, pH 3.0, ionic strength 0.1 mol/l (HNO,, NaNOs). Strip: 0.1 mol/l HNOs. SLM: 35 mmol/l DEHPA in kerosene.

C. Moreno et al./J. Membrane Sci. 81 (1993) 121-126

metal transport across the membrane becomes independent of diffusion in the aqueous phases, which is found to be minimum under these con- ditions. As a consequence, the diffusion in the LM is considered to be the rate determining step in the described metal transport process in the mentioned rpm range. According to these re- sults, the systematic study of the different chemical parameters was carried out at 1300 rpm. Our results apparently contrast with those reported for chloride media [ 91, where no vari- ation in permeability was shown. The observed differences are difficult to explain because of the lack of details from Hirashima’s work. Probably, the higher lanthanide concentration used by Hirashima (two or four orders of mag- nitude) may contribute to the constancy of the permeability values.

Figure 2 shows the increase of P with carrier concentration. The membrane permeability increases by 310% for Pr and by 410% for Nd going from 0.01 to 0.1 mol/l DEHPA. At higher carrier concentration the increase is much lower, especially for Pr cations where almost constant P values are reached at c (DEHPA) > ca. 0.25 mol/l. For both ele-

0.0 0.25 0.6 c(DEHPA) (mol/l)

Fig. 2. Permeability coefficients vs. carrier concentration. Concentration of Pr or Nd in feed sol. 31.2 or 30.7 poll ml, respectively. Other conditions as in Fig. 1. Solutions stirred at 1300 (Pr) or 1000 (Nd) rpm. SLM: DEHPA in kerosene.

C. Moreno et al./J. Membrane Sci. 81(1993) 121-126 125

ments, relatively high values of the permeabil- ity coefficients are reached already at low car- rier concentrations, e.g. at 25 mmol/l.

hand, the presence of a waxy-gelled substance in the feed solution has been observed at pH ca. 4 in some experiments.

The results reported by Hirashima et al. on the effect of DEHPA concentration were ob- tained at much higher concentrations. How- ever, they also found a limitation on permea- bility around 1 M DEHPA. An explanation for the obtained limitations may be found in the diffusion of the organic species in the mem- brane phase.

As demonstrated in Fig. 3, the pH is the vari- able. that masters the transfer of metal through the SLM. The dependence of P on pH shows a maximum at pH ca. 3. The described decrease of permeability at pH > 3 can not be explained in terms of hydrolysis of lanthanide cations be- cause this does not start until a fairly high pH. As known, e.g. in Nd( III) aqueous solutions, formation of polynuclear species and precipi- tation of hydroxides proceeds at pH> ca. 5.8 and pH>ca. 6.2, respectively [ 151. Thus, the observed behavior may probably be explained by the formation of species having a diffusion coefficient substantially lower than that of the complexes formed in reaction (1). On the other

The results reported by Hirashima et al. on the acidity effect were obtained at a much higher acid concentration. However, the ob- served differences at pH 1 are remarkable. Such differences may be attributed to a possible dif- ferent extraction mechanism from the chloride media. The dependence of the transport rate on HCl concentration also exhibits maxima and occurrence of a gelled substance.

At conditions described in Fig. 3 the value for the membrane separation factor [ 141, (Y =&I Pr,, determined from the experimental data at pH 2.70 is equal to 1.25, o.r using the values found at pH 2.5 in Fig. 3, (Y= 1.5. The values for the separation factor are comparable or slightly higher than those mentioned in pre- vious work [ 91. These differences in behavior of Pr (III) and Nd(II1) cations are not large enough to separate mutually the two elements on an SLM in a single-stage process. However, the utilization of this chemical system in a technique as e.g. centrifugal partition chroma- tography [ 161 would be promising.

0.3

3 E

2 0.2

s

PI 0.1

0.0

1 3 PH

5

Fig. 3. Permeability coefficients vs. pH of feed solution. Initial lanthanide concentration in feed solution 29.1 mol/ ml, stirring rate 1300 rpm, SLM: 25 mmol/l DEHPA in kerosene. Other conditions as in Fig. 1.

Acknowledgement

This work was funded by CICYT (Spanish Commission for Research and Development) project No. PTR 89-0206. Dr. A. HrdliEka ac- knowledges the Spanish Ministry of Education and Science for financing his stay at U.A.B. (Quimica Analitica).

References

L. Bednarczyk and S. Siekierski, Extraction of light lanthanide nitrates by tri-n-butyl phosphate, Solvent Extr. Ion Exch., ‘7 (1989) 273. S. Motomizu and H. Freiser, Extraction of tervalent lanthanides with acidic organophosphoric com- pounds, Solvent Extr. Ion Exch., 3 (1985) 637.

126 C. Moreno et al.@. Membrane Sci. 81(1993) 121-126

3 J.S. Preston and A.C. Du Preez, Solvent extraction processes for the separation of the rare-earth metals, in: T. Sekine (Ed.), Solvent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part A, Elsevier, Amsterdam 1992, p. 883.

4 A.C. Du Preez and J.S. Preston, Solvent extraction of rare-earth metals by carboxylic acids, in: T. Sekine (Ed.), Solvent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part A, Elsevier, Amsterdam 1992, p. 1081.

5 B. Gorski, N. Gorski and M. Beer, Extraction of SC, Y, and lanthanoids by quaternary ammonium salts, in: T. Sekine (Ed.), Solvent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part A, Elsevier, Amsterdam 1992, p. 1021.

6 N.V. Thakur, D.V. Jayawant and K.S. Koppiker, Sep- aration of neodymium from its concentrate using 20% saponified di- (2-ethylhexyl)phosphoric acid in kero- sene as an extractant, in: T. Sekine (Ed.), Solvent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part A, Elsevier, Amsterdam 1992, p. 1003.

7 Y. Minagawa and K. Yamaguchi, Relative extraction rates of rare-earth ions from weakly acidic solution of binary mixture of rare-earth chlorides by di- (2-ethyl- hexyl)phosphoric acid/kerosine, Hydrometallurgy, 24 (1990) 333.

8 T.-M. Hung, J.-Sh. Horng and Y.-Ch. Hoh, Stripping rare earth from its loaded Di(2-ethylhexyl)-phos- phate complex by oxalic acid, in: T. Sekine (Ed. ) , Sol- vent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part A, Elsevier, Amsterdam 1992, p. 901.

9 Y. Hirashima, S. Kohri and G. Adachi, Transport of lanthanoids through liquid membranes containing DZEHPA as a carrier, in: T. Sekine (Ed.), Solvent

Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part B, Elsevier, Amsterdam 1992, p. 1499.

10 T.-M. Hung and Ch.-J. Lee, Retardation of the trans- port of Eu (III) through supported di- (2-ethylhexyl) - phosphoric acid-kerosene liquid membrane, in: T. Sekine (Ed.), Solvent Extraction 1990, Proc. Intl. Solvent Extr. Conf. ISEC ‘90, Part B, Elsevier, Am- sterdam 1992, p. 1469.

11 M.R. El-Sourougy, S.M. Khalifa, E.E. Zaki and H.F. Aly, Lanthanide extraction by liquid emulsion mem- brane based on di- (2-ethylhexyl)phosphoric acid, Proc. Intl. Solvent Extr. Conf. ISEC ‘88, Yu.A. Zolo- tov (Ed.), Vol. III, Nauka, Moscow 1988, p. 96.

12 V. Salvadb, A. Masana, M. Hidalgo, M. Valiente and M. Muhammed, Permeation of gold (III) through Cy- anex 471 solid supported liquid membranes, Anal. Letters, 22 (1989) 2613.

13 J. Havel, C. Moreno, A. HrdliEka and M. Valiente, Spectrophotometric determination of rare-earth ele- ments by flow injection analysis based on their reac- tion with Xylenol Orange and cetylpyridinium bro- mide, Talanta, (in press).

14 P.R. Danesi, Separation of metal species by supported liquid membranes, Sep. Sci. Technol., 19 (1984/85) 857.

15 J. Kragten, Atlas of Metal-Ligand Equilibria in Aqueous Solution, E. Horwood, Chichester, 1978, p. 497.

16 T. Araki, Centrifugal countercurrent type chromato- graphy as multistage liquid membrane transport sys- tems, in: T. Araki and H. Tsukube (Eds.), Liquid Membranes: Chemical Applications, CRC Press, Boca Raton, FL, 1990, p. 201.