MEMBRANE ASSISTED LIQUID-LIQUID EXTRACTION OF CERIUM
A thesis submitted for the degree of Doctor of Philosophy
to
The University of New South Wales School of Chemical Engineering and Industrial Chemistry Faculty of Engineering Sydney, Australia
by
Karin Helene Soldenhoff B Sc (Hons) University of Witwatersrand M Sc University of Cape Town
February 2000 CERTIFICATE OF ORIGINALITY
I hereby declare that this submission is my own work and to the best of my knowledge it contains no material previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.
(Signed) .. ABSTRACT
Membrane assisted liquid-liquid extraction of cerium was investigated, with emphasis placed on the study of the reaction chemistry and the kinetics of non-dispersive solvent extraction and stripping with microporous membranes. A bulk liquid membrane process was developed for the purification of cerium(IV) from sulfate solutions containing other rare earth elements. The cerium process was studied in both a flat sheet contained liquid membrane configuration and with hollow fibre contactors.
Di-2-ethylhexyl phosphoric acid (DEHPA) was identified as a suitable extractant for cerium(IV) from sulfuric acid solution, with due consideration of factors such as extraction ability, resistance to degradation, solvent selectivity and potential for sulfate transfer into a strip solution. A detailed study of the extraction of cerium(IV) with DEHPA defined the extraction reaction chemistry.
The Ce/DEHPA/sulfate system was also investigated with a flat sheet bulk liquid membrane configuration, using both sulfuric and hydrochloric acid as receiver solutions. These tests identified that hydrophobic membranes provide better mass transfer for extraction and hydrophilic membranes are better for stripping. The presence of an impurity, mono 2-ethylhexyl phosphoric acid (MEHPA), was found to have a dramatic accelerating effect on the rate of the chemical extraction reaction. This was attributed to its higher interfacial activity and population compared to DEHPA, and the fact that MEHPA was also found to be an active carrier for cerium(IV).
The mass transfer rate of membrane assisted extraction and stripping of cerium, using hydrophobic and hydrophilic microporous membranes, respectively, was investigated using a modified Lewis-type cell. It was quantitatively demonstrated that the extraction process was mainly controlled by membrane diffusion and the stripping process was controlled by the chemical reaction rate, with membrane diffusion becoming important at low distribution coefficients.
Finally, two hollow fibre contactors were operated continuously for 65 hours, highlighting the positive aspects of this technology. The process proved to be easy to control, with very low entrainment levels measured for solvent in the feed solution. High cerium(IV) extraction and good selectivity over other rare earths was achieved.
Note: The main body of the thesis has been changed to correct spelling errors. Where appropriate, footnotes incorporating comments from reviewers have also been included. AKNOWLEDGEMENTS
I would like to thank my supervisor Prof. Tony Fane for an introduction into the wonderful world of membranes and his guidance and comments. A special thanks goes to my co-supervisor Dr. Robert Ring for his encouragement throughout the period of this study.
The support and help of many colleagues and friends at the Australian Nuclear
Science & Technology Organisation is much appreciated. I would like to thank
Des Levins for supporting my part-time studies and useful comments on reaction kinetics. Stuart Macnaugton and Jenny Mcculloch for helping with the construction of the hollow fibre rig and the tedious task of fibre potting.
Lyle Poppitt and Bruce Breadner for always coming up with a practical solution for equipment fabrication. Thanks to Beate Wildner for the diagrams, collation and printing of the manuscript. A special thanks to Deborah Wilkins and
Marleine Shamieh for the many shared cups of coffee and for always being so helpful in the laboratory.
I would also like to acknowledge the contribution of Catherine Chan and
Kate MacFarlane, who carried out some of the experimental work presented in chapter 5, during their undergraduate summer vacations.
Finally a special thanks to my long suffering husband for the many extra hours of child care and the unfailing support.
Note: This PhD thesis is submitted with the permission of the Australian
Nuclear Science and Technology Organisation. Table of Contents
MEMBRANE ASSISTED LIQUID-LIQUID EXTRACTION OF CERIUM
1. INTRODUCTION
1.1 Background 1 1.2 Solvent extraction versus liquid membranes 5 1.3 Scope of work and outline of this thesis 6 1.4 References 7
2. EXPERIMENTAL TECHNIQUES
2.1 Introduction 8 2.2 Chemical Reagents 8 2.3 General analytical procedures 9 2.3.1 Determination of cerium(IV) concentration 9 2.3.2 Determination of metal ions by inductive coupled plasma 11 2.3.3 Determination of aqueous acidity by titration 12 2.3.4 Determination of concentrations of di-2-ethylhexyl phosphoric acid and mono 2-ethylhexyl phosphoric acid 12 2.3.5 Purification of di-2-ethylhexyl phosphoric acid 12 2.4 Experimental procedures for Chapter 3 13 2.4.1 Equilibrium tests 13 2.4.2 Solvent stability tests 13 2.5 Experimental procedures for Chapter 4 14 2.5.1 Membrane characterisation 14 2.5.2 Three compartment flat sheet membrane permeation cell 15 2.5.3 Permeation experiments 15 2.5.4 lnterfacial tension measurements 15 2.6 Experimental procedures for Chapter 5 17 2.6.1 Analytical procedures for the determination of concentrations of iodine, acetone and toluene 17 2.6.2 Two compartment flat sheet membrane permeation cell 19 2.6.3 Permeation experiments 21 2. 7 Experimental procedures for Chapter 6 23 2.7.1 Hollow fibre contactor experimental apparatus 23 2.7.2 Measurement of entrained organic in the raffinate 25 2.8 References 25 Table of Contents
3. SOLVENT SELECTION AND CHEMISTRY
3.1 Introduction 26 3.2 Background 26 3.2.1 Acidic extractants 26 3.2.2 Solvating extractants 28 3.2.3 Basic extractants 29 3.2.4 Solvent extraction processes for cerium(IV) 29 3.2.5 Summary of background literature 30 3.3 Results and discussion 31 3.3.1 Solvent selection 31 3.3.1.1 Effect of aqueous media and acidity 32 3.3.1.2 Solvent selectivity 38 3.3.1.3 Solvent stability 39 3.3.2 Chemistry of extraction of cerium with TOPO and Cyanex 923 40 3.3.3 Chemistry of extraction of cerium with DEHPA 42 3.3.3.1 Acidity range of extraction 43 3.3.3.2 Effect of DEHPA concentration 43 3.3.3.3 Effect of solvent loading 43
3.3.3.4 Extraction reaction in the acidity range 0.5 to 5 M H2S04 46 3.3.3.5 Extraction reaction in the acidity range greater than 5 M
H2S04 51 3.4 General discussion 55 3.5 Conclusions 56 3.6 Nomenclature 57 3.7 References 58
4. TRANSPORT OF CERIUM WITH A FLAT SHEET BULK LIQUID MEMBRANE 4.1 Introduction 66 4.2 Background 66 4.2.1 Transport mechanisms in liquid membranes 66 4.2.2 Types of liquid membrane systems 67 4.2.2.1 Emulsion liquid membranes 69 4.2.2.2 Supported liquid membranes 71 4.2.2.3 Bulk liquid membranes 72 4.2.3 The role of interfacial tension 75 4.3 Results and discussion 76 4.3.1 Membrane characterisation and selection 77 4.3.2 Permeation experiments 81 4.3.2.1 Effect of stirrer speed 84 4.3.2.2 Effect of membrane type 86
ii Table of Contents
4.3.2.3 Effect of mono-2ethyl hexyl phosphoric acid 91 4.3.2.4 Effect of receiver solution composition 94 4.3.2.5 Effect of cerium concentration in the feed 95 4.3.3 lnterfacial measurements 95 4.4 General Discussion 107 4.5 Conclusions 108 4.6 Nomenclature 109 4. 7 References 110
5. THE ROLE OF REACTION KINETICS IN MEMBRANE ASSISTED SOLVENT EXTRACTION
5.1 Introduction 116 5.2 Background 116 5.2.1 lnterfacial Zones 117 5.2.2 Extraction Regimes 118 5.2.3 Experimental Techniques 119 5.2.3.1 Stirred vessels with constant interfacial area 119 5.2.3.2 Highly agitated vessels 121 5.2.3.3 Moving drops 121 5.2.3.4 Rotating diffusion cell 122 5.2.3.5 Hollow fiber membrane extractor 122 5.2.4 Kinetic studies with di-2-ethylhexyl phosphoric acid 123 5.3 Results and discussion 129 5.3.1 Choice of experimental technique 129 5.3.2 Hydrodynamic characteristics of membrane permeation cell 129 5.3.3 Calibration of permeation cell with microporous hydrophobic Millipore GVHP membrane 138 5.3.4 Extraction of cerium with a membrane permeation cell 142 5.3.4.1 Determination of chemical extraction rate equation 145 5.3.4.2 Transport mechanism of membrane liquid-liquid extraction process 149 5.3.5 Calibration of permeation cell with microporous hydrophilic Millipore WLP membrane 153 5.3.6 Stripping of cerium with a membrane permeation cell 157 5.3.6.1 Determination of chemical reverse rate equation 158 5.3.6.2 Transport mechanism of membrane liquid-liquid back- extraction process 162 5.4 Conclusions 164 5.5 Nomenclature 166 5.6 Reference List 167
iii Table of Contents
6. PURIFICATION OF CERIUM WITH HOLLOW FIBRE CONTACTORS
6.1 Introduction 175 6.2 Background 175 6.2.1 Liquid-liquid extraction with hollow fibre membrane contactors 177 6.2.2 Hollow fibre contained liquid membrane technique 178 6.2.3 Hollow fibre supported liquid membrane 179 6.3 Results and discussion 180 6.3.1 Experimental set-up and mode of operation 180 6.3.2 Effect of flowrate on pressure drop 181 6.3.3 Effect of interfacial tension on critical pressure 183 6.3.4 Effect of tube side flowrate on extraction flux 186 6.3.5 Preliminary batch permeation test 187 6.3.6 Continuous run 198 6.3.6.1 Concentration profiles 190 6.3.6.2 Flux change with time 193 6.3.6.3 Purification of cerium with respect to other rare earth elements 195 6.3.6.4 Transfer of sulfate 196 6.3.6.5 Solvent loss to raffinate 198 6.4 General discussion 200 6.5 Conclusions 204 6.6 Nomenclature 205 6.7 References 206
7. OVERALL CONCLUSIONS 212
APPENDICES
Appendix 1 Calculation of equilibrium hydrogen ion concentration in a sulfuric acid solution
Appendix 2 Calculation of hydrogen ion activity as a function of sulfuric acid concentration
Appendix 3 Options for the recovery of cerium by solvent extraction
Appendix4 Examples of calculations of interfacial concentrations
Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials
iv Chapter 1
Introduction
Summary This chapter briefly describes the application of cerium chemicals in various industries and presents an overview of the main processing routes for the purification of cerium from its mineral concentrates. Solvent extraction and liquid membrane technology are discussed in the context of cerium solution purification. A scope and outline of this thesis is given in the last section. Chapter 1 Introduction
1.1 Background
Cerium is part of the group of elements collectively known as the rare earth metals (RE). The rare earth elements consist of the fifteen lanthanides, ranging from lanthanum through to lutetium. Yttrium and scandium are often included in this group because of their presence with lanthanides in mineral deposits.
Cerium is used in many applications where its chemical form and purity vary (Kilbourn 1992). Cerium's major use, in terms of tonnage, is in the mixed lanthanide form such as mischmetal", rare earth silicide or as a ferro-alloy. These materials are used as additives for steel and cast irons and contain cerium and other rare earth metals in the same proportion as found in the mineral concentrates from where they originate. The exact cerium to total rare earth metal ratio depends on the composition of the original material but is around 50%.
Broadly, other areas of application for cerium products include:
i) glass and ceramics; ii) catalysis and chemicals, and iii) phosphor/luminescence
In the glass and ceramics industries, cerium is used as a polishing powder and in glass decolorisation. Cerium's special chemical properties also enhance the resistance of glass to ultra-violet radiation and to bombardment by high energy electrons, experienced in television screens. Cerium can also be used to provide pigments with light fastness and to prevent clear polymers from darkening in sunlight. In ceramics, applications range from dental ceramics that mimic natural tooth appearance to cerium oxide-doped zirconia used in aircraft engine parts.
A widespread use of rare earth metals is in cracking catalysts to produce low octane fuel from heavy crude oil. The demand in this application is in decline because of the increased production of high octane fuel. A growing market for
• Mischmetal is made up of a mixture of lanthanides in the metallic form and contains approximately 50% cerium. It is produced by electrolytic reduction of fused mixed-lanthanide chloride.
1 Chapter 1 Introduction cerium, however, is as an active ingredient in catalytic converters, in combination with platinum group metals, for vehicle emission control.
Finally cerium is an essential component in new generation fluorescent lighting, and it is also used in specialised cathode ray tube and high-intensity lighting applications.
Whilst some products such as polishing powders for glass require medium purity cerium products (96% Ce02), uses in the high technology industries such as the speciality glasses, phosphors and the electronics industries require greater than 99.9% Ce02•
The most common raw materials used for the production of rare earth metals are bastnasite, a fluorocarbonate mineral, monazite, a phosphate mineral and xenotime, also a phosphate mineral (Vijayan 1989). The first step in processing is to separate the rare earth elements from their respective fluoride or phosphate matrix. This requires either high temperatures or very alkaline or acidic conditions.
Two processes are used commercially to remove fluoride from bastnasite mineral. Molycorp in the USA roasts the concentrate to remove the carbonate component as carbon dioxide and the fluoride component as hydrogen fluoride. The rare earth metals, with the exception of cerium, are leached with hydrochloric acid (Koch, 1987; Zhang Guocheng, 1986). Cerium becomes refractory during the roasting process and remains in the residue during acid leaching.
Fluoride is removed from the Baotou ore in China, by differential flotation (Zhang 1982). The bastnasite flotation concentrate is then digested in sulfuric acid at elevated temperature, driving off any remaining fluoride, as well as carbon dioxide.
The phosphate matrix in monazite and xenotime minerals is resistant to mild chemical attack but can be destroyed in strong, hot acids and alkalis. Figures 1.1A and 1.1 B present a summary of these two processing options. In the alkali route, the concentrate is treated with caustic soda, which converts the rare earth phosphates into rare earth hydroxides. Once in this form the rare earths are amenable to mild acid leaching. Hydrochloric or nitric acid are used for this purpose.
2 Chapter 1 Introduction
A) Caustic Route
Concentrate
Caustic NaOH
Na3P04 Caustic recovery RE hydroxides t HN03 or HCI Mild Lime acid leach Ca2(P04h
RE solution for further purification
Residue
B) Acid Route
Concentrate
RE sulfates Double sulfate precipitate
NaOH RE conversion
RE hydroxides .---~---, HCI or HN03 Acid dissolution
RE solution for further purification
Figure 1.1 - Processing options for monazite concentrate
3 Chapter 1 Introduction
Direct sulfuric acid digestion of monazite concentrates is also practiced. Once in solution, the rare earth elements are separated from other impurities by precipitation of the double sulfate salt ((RE)2(S04)3.Na2S04). This rare earth precipitate has to be converted into the hydroxide form before it can be solubilised.
The leach solutions containing the mixed rare earths arising from the processes discussed above are the starting point for purification and separation of the rare earth elements from each other. In these solutions the lanthanides are present in the trivalent oxidation state and due to their similar chemical properties are very difficult to separate from each other. Typically this separation is achieved by many stages of solvent extraction. The tetravalent state of cerium however, is also common, and it is this property that is often exploited to effect its separation from the other rare earth metals.
The processes used industrially for cerium separation and purification can be summarized as follows:
• For production of low to medium purity cerium: i) oxidation or calcination of cerium in the solid concentrate followed by subsequent leaching of the trivalent rare earths; and ii) chemical oxidation of cerium from rare earth liquor.
• For production of high purity cerium: i) separation of all trivalent rare earths from each other by sequential solvent extraction; and ii) separation of cerium(IV) from other rare earths by solvent extraction.
Regardless of the mineral rare earth source or the initial processing route followed, published material indicates that current industrial processes use chloride or nitrate media for solvent extraction. Sulfate media does not seem to be used industrially, presumably due to the lower rare earth solubility and the possible complication arising from formation of double salts. Sulfate media has however the advantage of being cheap and sulfate waste solutions pose less of a disposal problem than nitrate solutions. Notwithstanding the disadvantages of sulfate, the potential economic and environmental benefits of a totally sulfate based route for the purification of cerium is worth exploring. This is especially
4 Chapter 1 Introduction relevant to medium purity cerium products, where the sulfate route could decrease production costs.
This thesis investigates the liquid-liquid extraction and purification of cerium from sulfate media, and explores the use of membrane assisted solvent extraction and bulk liquid membrane technology for this application.
1.2 Solvent extraction versus liquid membranes
Solvent extraction is a mature technology with widespread industrial application due to its many advantages. Some of these advantages include high throughput and ability to cope with feed concentrations ranging from 500 ppm to 100 g L·1. Solvent extraction is also a very selective technique and offers the possibility of producing concentrated solutions by control of phase ratios in both the extraction and stripping circuits. In the separation of trivalent rare earth metals, where many equilibrium stages and a high degree of pH control are required, solvent extraction is unchallenged as the standard industrial separation technology.
The contacting equipment for solvent extraction processes includes the commonly used mixer-settlers, pulsed and packed columns, in-line mixers and centrifugal contactors. In the last decade, research studies on the use of hollow fibre contactors for liquid-liquid extraction have attracted some attention. This non-dispersive technique has the advantage of reducing the solvent inventory by eliminating the need for a settler. Phase separation problems can also be avoided because the two phases are not dispersed in one another. It is also claimed that solvent losses are much reduced, although not many experimental studies are currently available to corroborate this claim. When hollow fibre extraction is coupled with stripping in another hollow fibre contactor, and the solvent circulated between the two under non-equilibrium conditions, the process effectively becomes a type of bulk liquid membrane process. Liquid membrane techniques have the added advantages of being able to achieve even higher concentration factors and lower raffinates than are possible with solvent extraction.
Liquid-liquid extraction with hollow fibre contactors are investigated in this thesis in the context of extraction and purification of cerium from sulfate solutions.
5 Chapter 1 Introduction
1.3 Scope of work and outline of this thesis
The main aims of this PhD study are:
• To develop a solvent system applicable to the purification of cerium by extraction of cerium(IV) from sulfate media;
• To understand the mechanisms controlling mass transfer of cerium in membrane assisted extraction and stripping; and
• To explore the applicability of using hollow fibers in a bulk liquid membrane configuration in such a solvent system.
Following is a summary outline of the thesis and the contents of each chapter.
The first chapter, Chapter 1, introduces the topic of cerium extraction and purification and places it in the context of industrial practice.
Chapter 2 details the experimental techniques used in this thesis. Procedures developed for the analysis of cerium(IV) in solution are presented. Three sets of apparatus are described:
i) a flat sheet bulk membrane permeation cell to measure mass transfer from an aqueous feed to a receiver solution, through a solvent phase; ii) a membrane permeation cell to measure mass transfer kinetics in a two phase system; iii) a rig consisting of two hollow fiber units for extraction and stripping. Literature reviews and corresponding references are included in Chapters 3-6.
Chapter 3 deals with the selection of a solvent system for the liquid - liquid extraction of Ce4+ from acidic sulfate solutions. This chapter also presents a comprehensive original study of the chemistry of extraction of Ce4+ with di-2- ethylhexyl phosphoric acid, the extractant selected for further study.
The mass transport of cerium is studied in a batch, flat-sheet contained liquid membrane configuration in Chapter 4. Parameters such as membrane properties, interfacial tension and chemical composition of the solutions are investigated.
6 Chapter 1 Introduction
The kinetics of membrane extraction and back-extraction of the cerium(IV) sulfuric acid-di(2-ethyhexyl)phosphoric acid-n heptane system is studied in Chapter 5. The relative resistances to mass transfer attributed to aqueous, organic and membrane diffusion and chemical processes are quantitatively determined.
The use of hollow fibers for the purification of cerium is demonstrated in Chapter 6. The long-term performance of the process with respect to solvent losses is also evaluated.
A discussion with concluding remarks is presented in Chapter 7.
1.4 References
1. Kilbourn B. T. 1992. Cerium - A guide to its role in chemical technology. Published by Molycorp, Inc, White Plains, NY, U.S.A.
2. Vijayan S., Melnyk A.J., Singh R.D. and Nuttall K. 1989. Rare earths: Their mining, processing, and growing industrial usage. Mining Engineering: 13- 18.
3. Zhang B. Z., Lu K.Y., King K.C., Wei W.C. and Wang W.C. 1982. Rare earth industry in China. Hydrometallurgy 9: 205-10.
7 Chapter 2
Experimental techniques
Summary This chapter presents all experimental procedures used in the studies carried out for this thesis. Details of chemical reagents, membrane specifications, analytical procedures and experimental protocols for solvent extraction and membrane permeation tests are provided. Three separate experimental apparatus are described, including a modified Lewis cell, a flat sheet bulk liquid membrane permeation cell and a two module hollow fibre contactor rig. Chapter 2 Experimental techniques
2.1 Introduction
This thesis is based on experimental studies that encompass various aspects of membrane assisted solvent extraction and bulk liquid membrane technology. In order to focus on different areas of these extraction techniques, three separate experimental apparatus were designed and developed. These include two different types of batch permeation cells and a rig consisting of two hollow fibre contactors and accessory equipment.
This chapter describes all experimental and analytical procedures used for the experimental studies and supplements information given in subsequent chapters.
2.2 Chemical Reagents
Unless specified in the text, all chemical reagents were used without further purification.
Analytical grade reagents
Di-2-ethylhexyl phosphoric acid commonly known as DEHPA (>98%) was supplied by both Merck and BDH. A mix of DEHPA and MEHPA (mono 2- ethylhexyl phosphoric acid), containing approximately 50% of each was supplied by Merck. Each batch was checked for purity by the procedure outlined in Section 2.3.4. Tributyl phosphate (TBP), 98.5%, was purchased from Ajax Chemicals, and trioctyl phosphine oxide (TOPO) was obtained from
Aldrich Chemicals. Ce(S04) 2.4H 20 (98%) was obtained from BDH and
CeCl3.7H 20 (98.5%) from Merck. Thorium sulfate solutions were made up by dissolving Th(N03) 4 .5H 20 in sulfuric acid. Analytical grade n-heptane, supplied by Ajax Chemicals, was used as a diluent in all cases except for the 0.47 M solution of TOPO, which was made up in Shellsol 2046. All other reagents such as acids and bases were of analytical grade.
All water used for the tests was treated with a reverse osmosis Milli-RO water purification system followed by a Milli-Q Ultrapure water system containing ion exchange resins and carbon filters, both supplied by Millipore.
8 Chapter2 Experimental techniques
Commercial grade reagents
Following is a list of commercial grade reagents used and their respective suppliers.
Dibutylbutyl phosphonate (DBBP), di 2-ethylhexyl 2-ethylhexyl phosphonate (BEHEHP) and mono 2-ethylhexyl 2-ethylhexyl phosphonate (lonquest 801) were supplied by Albright & Wilson, USA. Bis-2,4,4-trimethylpentyl phosphinic acid (Cyanex 272), Cyanex 923 and Cyanex 925 were supplied by Cytec, Canada. Alamine 336 was supplied by Henkel. Shellsol 2046, a kerosene based diluent was supplied by Shell, Australia.
2.3 General analytical procedures
2.3. 1 Determination of cerium(IV) concentration
The concentration of cerium(IV) was determined by two methods: uv-vis spectrometry and titration.
Determination of cerium(IV) by titration
The concentration of cerium(IV) in solution was determined by a back titration method. To an aliquot of the sample to be analysed (1-5 ml), a known amount (1 O ml) of ferrous sulfate (0.003 - 0.02 M) was added and the solution titrated against standardised eerie sulfate (0.003 - 0.02 M), using ferroin as an indicator. The ferrous ion reduces cerium(IV) to cerium(III). Any excess ferrous ion remaining is titrated with standard cerium(IV) solution. To achieve low detection limits (ppm range), care needs to be taken to choose reagent concentrations in proportion to the amount of cerium(IV) in the sample.
Determination of cerium (IV) by uv-vis spectrometry
This technique was used to measure changes in cerium(IV) concentrations continuously in both the aqueous and solvent phases. Cerium(IV) in sulfate solution is bright orange and therefore its concentration can be determined by uv-vis spectrometry. The absorption spectra of cerium(IV) has a single peak at
320 nm with an absorption coefficient of 5675 l mo1·1 cm·1• The cerium-DEHPA complex in n-heptane also absorbs in a similar region, but the absorbance coefficient is not as high. The difference in absorption readings of cerium(IV) in the aqueous and solvent phases is shown in Figure 2.1. The instrument used
9 Chapter 2 Experimental techniques
I •Aqueous oOrganiCJ 2.0 1.8 1.6 • 1.4 (I) g 1.2 -eIll 1.0 0 i 0.8 • Figure 2.1 Absorbance versus concentration calibration curves for cerium(IV) Aqueous: 0.55 M H2S04, wavelength = 390 nm Organic: 0.2 M DEHPA in heptane, wavelength = 390 nm --~------=-;:_=.==--;======,~=====---- 1.------rst derivative o pH ~d derivative 20 50 18 11111 200 16 150 14 12 • 100 ::i:: 10 c. 8 ...____... __~----4-4"r- 6 4 0 0 0 2 • , 0 +-11------'--____-+- ____ e-----t----'----t--~=-- 7 8 9 10 NaOH Volume (mL) Figure 2.2 Potentiometric titration for determination of DEHPA and MEHPA concentration. [DEHPA] = 3.1 M, [MEHPA] = 0.032 M 10 Chapter 2 Experimental techniques for these analyses was a Shimadzu UV-1601, capable of taking multiple readings of up to six solution streams, but only at a single set wavelength. Taking into consideration the disparity of absorption coefficients for aqueous cerium and the cerium-DEHPA complex, in addition to the differences in concentration in the two phases, it was not always feasible to simultaneously monitor both the solvent and the aqueous concentrations. The calibrating protocol using the uv-vis technique is detailed below. Calibrating aqueous phase Three aqueous standards were made up from a volucon standard cerium stock solution. The stock solution is not stable as the cerium(IV) is reduced slowly to cerium(III). The rate of change in cerium(IV) concentration was measured as 1.0377E-05 Mor 1.45 ppm/day for a 0.05 M Ce(S04) 2 solution. Once prepared, the stock solution was calibrated against primary standard potassium ferrocyanide and checked once a month against the predicted concentration. The standards were made up daily in the same acid concentration as the feed solution. The stock was dispensed by an automatic titrator. Absorbance readings for three aqueous standards were taken in the same flowthrough cell as used for the kinetic run. Calibrating the organic phase The organic standards were prepared by contacting a known volume of solvent with aqueous standard cerium(IV) solution. The loaded solvent was then split into two portions. Half of the solvent was used to measure the absorbance reading. The remaining portion was used for determination of the cerium concentration by the method outlined in Section 2.3.2. In summary, the solvent phase was calibrated by using inductive coupled plasma readings (ICP OES) as the standards. 2.3.2 Determination of metal ions by inductive coupled plasma Concentrations of Ce (total cerium), Th, La, Nd, Y and S were determined by inductive coupled plasma - optical emission spectrometry (ICP-OES) using a Perkin-Elmer 3000 DV instrument. Aqueous solutions were fed to the instrument after appropriate dilution according to the calibrating range. Solvent samples were first contacted with a stripping solution to back extract the elements into the aqueous phase. The stripping solution varied depending on 11 Chapter 2 Experimental techniques the element to be determined. For cerium analysis, the solvent was contacted with 2 M HCI and 0.5 M H20 2. Other rare earth elements were back extracted with 6 M HCI. An alkali solution of Na2CQ 3 was used to strip the thorium. 2.3.3 Determination of aqueous acidity by titration The amount of free acid in solution was determined by addition of a sample aliquot to a solution containing sodium oxalate to complex all metal ions. The free acid was then titrated against a standard solution of sodium hydroxide. 2.3.4 Determination of concentrations of di-2-ethylhexyl phosphoric acid and mono 2-ethylhexyl phosphoric acid Di 2-ethylhexyl phosphoric acid (DEHPA) often contains small amounts of mono 2-ethylhexyl phosphoric acid (MEHPA) as an impurity. MEHPA is a by product of DEHPA manufacture, but it is also produced by hydrolysis of DEHPA with prolonged storage (Rao 1993). The relative proportions of the two constituents in a given batch of DEHPA were determined by potentiometric titration of a sample aliquot dissolved in 50% ethanol and titrated against standard sodium hydroxide. All supplied batches of DEHPA and DEHPA/MEHPA mixtures were analysed by this method. The impurity MEHPA has an extra hydroxy group which causes a second inflection point on the pH versus base curve. This second inflection point is best detected by calculation of the second derivative of the data. An example of a titration result is given in Figure 2.2. 2.3.5 Purification of di-2-ethy/hexyl phosphoric acid Purification of DEHPA was undertaken according to the method of (Partridge 1969), where the DEHPA-Cu salt is precipitated. The following procedure was used. A 1 molar solution of DEHPA in cyclohexane was contacted with a concentrated CuS04 solution and NaOH added to promote extraction according to Equation 2.1. 2. 1 where HR represents the acidic DEHPA extractant 12 Chapter 2 Experimental techniques After loading the solvent with copper, the aqueous phase was discarded and acetone was added to the solvent phase. With acetone addition, the Cu DEHPA salt precipitated and was filtered and washed. The solid was then redissolved in n-heptane and repeatedly contacted with dilute hydrochloric acid until all the copper was removed and the DEHPA re-protonated. The concentration of DEHPA and MEHPA in the purified solvent was then determined by the titration method described in Section 2.3.4. 2.4 Experimental procedures for Chapter 3 2.4.1 Equilibrium tests Equilibrium data were obtained by contacting aqueous and organic phases in separating funnels by the general procedure outlined here. Specific details can be found in the legends for the figures where the data are presented. All experiments were conducted at a controlled temperature of 22 ± 1°C. Aqueous solutions of cerium(IV) sulfate (0.002-0.28 M Ce) and sulfuric acid (0.25-12 M H2S04) were contacted with organic solutions of DEHPA in n-heptane (0.02-1.2 M). The solutions were shaken for at least 30 minutes; preliminary experiments having established that 15 minutes were sufficient to attain equilibrium. The organic solutions were stripped by contact with 4 M HCI/ 0.35 M H 20 2• Cerium concentrations in both the aqueous and solvent phases were determined for each experiment. The procedure carried out for equilibrium tests for cerium(III) (results presented in Section 3.3.1.1, Chapter 3) was slightly different. Only aqueous solutions were analysed and the organic concentrations determined by difference. 2.4.2 Solvent stability tests Cerium(IV) is a strong oxidising agent which could potentially degrade the organic solvent used. Solvent degradation was measured by the following method. An aqueous cerium(IV) solution was contacted with a solvent at unit phase ratio so that the equilibrium concentration of cerium in the organic was around 2-3 g L-1. The two phases were kept in contact with stirring, and equal volume samples were withdrawn from both phases at various time intervals. The solvent sample was stripped with 4 M HCl/0.35 M H20 2 and the aqueous solution analysed by ICP-OES. All aqueous samples were analysed for total cerium by ICP-OES and cerium(IV) by titration. The experiments were run for 13 Chapter 2 Experimental techniques up to 40 days. DEHPA and TOPO solutions were prepared with the reagents as received from the suppliers. Two tests were carried out with Alamine 336 solutions. In the first test the Alamine 336 was made up with heptane and used with no pre-treatment. In the second test, the solvent solution was pre equilibrated with Na2C03, followed by contact with H2S04 . 2.5 Experimental procedures for Chapter 4 2.5. 1 Membrane characterisation The microporous membranes used and their respective suppliers are shown in Table 2.1. TABLE 2.1 Membrane suppliers for flat sheet microporous membranes Name Supplier Pore size Thickness Porosity Wetting Material µm µm % properties Durapore Millipore 0.22 125 75 hydrophobic PVDF GVHP Gelman Gelman 0.2 45-50 hydrophobic PVDF FP-200TM Durapore Millipore 0.2 125 70 hydrophilic PVDF GVWP Durapore Millipore 0.1 125 70 hydrophilic PVDF WLP GHPolypro Gelman 0.2 hydrophilic pp PVDF = Polyvinylidene fluoride PP = Polypropylene The thicknesses of the flat sheet membranes were determined by measuring two and three membranes together with a micrometer. The pore size distribution was measured with a Coulter® porometer. A special pore wetting substance, Porofil™, of known interfacial tension was used to fill the pores of the membranes to be tested. The pressure required to force the liquid out of the pores can then be used to determine the pore size and the distribution by means of the Cantor Equation 2.2. 2.2 14 Chapter 2 Experimental techniques where Pc is the critical displacement pressure, y is the interfacial tension and r is the pore radius. 2.5.2 Three compartment flat sheet membrane permeation cell A three compartment batch permeation cell made of clear polyvinyl chloride (PVC) was constructed. A photograph of the experimental set-up is presented in Figure 2.3. The volumes of the feed, solvent and receiver solutions were 145, 155 and 145 ml, respectively. These solutions were separated by microporous membranes with a surface area of 12.57 cm 2• All three compartments were individually agitated with overhead stirrers with variable controls in the range of 100-1800 rpm. 2.5.3 Permeation experiments The transport of cerium was measured with the batch three compartment permeation cell described in Section 2.5.2. The concentrations of cerium in the feed and receiver solutions were continuously monitored by a uv-vis spectrophotometer. The volume taken up by the uv-vis cell and connecting lines was less than 4 ml. Therefore only less than 3% of the volume of either the feed or receiver solution was outside the permeation cell. The solvent concentration was calculated by difference. In selected experiments, the final samples for all three phases were analysed by ICP and the total accountability for the system was found to be greater than 95%. The experiments were run at room temperature (22 ± 1 °C) for a period varying from 4 to 24 hours. Fresh membranes were used for each run. The solution compositions were as follows: Feed: Ce (200 -3200 ppm) in 0.55 M H2S04 Receiver or strip: 5.5 M H2S04 or 2 M HCI and 0.5 M H20 2 Solvent: 0.2 M DEHPA, MEHPA (0.002 - 0.027 M) in n-heptane 2.5.4 lnterfacial tension measurements The interfacial tension was measured by the DuNuoy ring method using a digital tensiometer, model K10ST, manufactured by Kruss, Germany. This method is based on measurement of the force required to pull a platinum iridium ring of known geometry from the interface of two non-immiscible liquids. 15 Chapter 2 Experimental techniques Figure 2.3 Experimental set-up for the three compartment flat sheet membrane permeation tests 16 Chapter 2 Experimental techniques Measurements of interfacial tension between aqueous solutions (0.55 or 5.5 M H2S04) and solutions of DEHPA or MEHPA (1x10·9 - 1 M) in heptane were taken. In preliminary tests, the solutions to be measured were saturated with each other. These readings were compared with those taken from solutions which had not been in contact. It was found that the two measurements were significantly different (ie 16% difference) only at very low extractant concentrations (10-0-10-s M), where the interfacial tension has the same value as that of the diluent. In the part of the curve where the interfacial tension decreases, the two measurements were the same (ie within 0.5% of each other). Since this part of the curve is the most significant for the calculation of surface excess, all tests were conducted without presaturation of the phases. 2.6 Experimental procedures for Chapter 5 2.6.1 Analytical procedures for the determination of concentrations of iodine, acetone and toluene Iodine concentration determination Iodine solutions were made up from the solid. The solubility of iodine in water is very low at approximately 0.0013 M and dissolution of the solid close to its solubility limit is very slow. Iodine is much more soluble in ethanol with very rapid dissolution occurring. Concentrated stock iodine solutions were therefore made up in ethanol (0.37 M). The iodine concentration in these stock solutions was determined by titration against sodium thiosulfate according to a standard method (Jeffery 1989). The uv-vis absorption spectrum for aqueous iodine has three peaks. None of the three peaks were found to increase linearly with concentration and therefore no attempt was made to continuously monitor the changes in aqueous absorbance. The absorbance spectrum for iodine in heptane however has a single peak, which follow Beer's law. A calibration curve (wavelength = 521 nm) for iodine in heptane is shown in Figure 2.4. It was found that the presence of ethanol decreased absorbance readings and care was taken to keep the amount of ethanol in all calibrating samples and test solutions below 0.08 µL mL·1, where no effect from ethanol could be detected. The absorption 1 1 coefficient measured by this method was 887 L mo1· cm· • 17 Chapter 2 Experimental techniques . 0.70 0.60 • 0.50 ~ y =~ 0.40 .Cl I., 0 0.30 • .Cl"' < 0.20 0.10 • 0.00 •• 0 0.0002 0.0004 0.0006 0.0008 (Iodine]/ M Figure 2.4 Calibration curve for iodine at wavelength = 521 nm ------[- •water +heptane 0.16 0.14 • 0.12 • ~ • y 0.10 • =~ .Cl 0.08 ·• • • I., 0 0.06 .Cl"' • < 0.04 0.02 • 0.00 0 0.005 0.01 0.015 (Acetone)/ M Figure 2.5 Calibration curve for acetone at wavelength = 261 nm [--I--+water a heptane-~ 1.40 ~------, 1.20 • ~ 1.00 y ; 0.80 .Cl • • ; 0.60 .Cl"' • • < 0.40 • 0.20 ,- 0.00 ..___ _,,...----,----,------4 0.000 0.002 0.004 0.006 0.008 , (Toluene] / M l~------Figure 2.6 Calibration curve for toluene at wavelength = 265 nm 18 Chapter 2 Experimental techniques Acetone concentration determination Changes in acetone concentration in the aqueous phase were monitored by uv vis spectrophotometry, by periodic readings of the absorbance at a wavelength of 261 nm. An example of the aqueous calibration curve is shown in Figure 2.5. Toluene concentration determination Changes in toluene concentration in the aqueous phase were similarly monitored by uv-vis spectrophotometry, by periodic readings of the absorbance at a wavelength of 265 nm. An example of the aqueous calibration curve is shown in Figure 2.6. 2. 6.2 Two compartment flat sheet membrane permeation cell This batch permeation cell is essentially a Lewis type cell modified to hold a membrane at the interface of two liquid phases. A schematic diagram is shown in Figure 2.7. The outer walls were made up of two glass cylinders held in place by three high density polyethylene flanges. The middle flange was especially designed to hold a flat sheet membrane. The two liquid phases were stirred independently in counter current rotation (stirring speed range 100 - 1500 rpm). In order to ensure that the interface was at the membrane surface, the solvent phase was located in the lower half of the cell, which was filled first. With a hydrophobic membrane and solvent in the bottom, the interface is located at the upper side of the membrane as any solvent leaking through the membrane pores floats to the top. With a hydrophilic membrane the interface is in located at the bottom of the membrane as any aqueous fluid leaking through the membrane pores sinks to the bottom. An external reservoir was found to be essential for expelling all the air from under the membrane surface. The membrane was mounted on a holder made of PVC, screwed into place with polypropylene screws and sandwiched between viton gaskets. Problems encountered were small leaks of solvent to the aqueous phase solvent through the sampling line and through the screws. These problems were overcome by proper placement of viton gaskets. 19 Chapter 2 Experimental techniques to water bath stirrer t organic sampling line aqueous digital pump membrane t {----11-- organic reservoir -+--+----11-----11----- organic t UV spectrophotometer to drain .-===11=:::!J stirrer t from water bath Figure 2.7 Schematic diagram of batch Lewis cell type membrane permeation cell 20 Chapter 2 Experimental techniques The experimental protocol used was as follows: Hydrophobic membrane 1. Mount the permeation cell with hydrophobic membrane 2. Fill the organic sample line and reservoir with known amount of solvent. (total volume in sample lines plus reservoir calibrated previously). 3. Pump solvent to the lower compartment until the solvent reaches the membrane. Ensure that volume of solvent pumped into the permeation cell is known. (Usually it was found that air was trapped between solvent and membrane). 4. Circulate solvent between permeation cell and uv-vis cell through solvent reservoir. 5. By gently tilting the permeation cell ensure that the air is transferred from the permeation cell to the reservoir. Note, the solvent reservoir was designed in such a way as to trap air bubbles. 6. Fill the upper compartment with know amount of aqueous and start the experiment. Hydrophilic membrane 1. Mount the permeation cell with a hydrophilic membrane 2. Pump just enough aqueous into the upper compartment so as to wet the membrane and leave a visible aqueous film on top of the membrane. 3. Follow steps 2 to 6 above. 2. 6.3 Permeation experiments Mass transfer experiments between water and heptane were performed with three diffusing substances; toluene, iodine and acetone. The mass transfer of toluene (0.003 M) from water to heptane was measured without a membrane. Both the aqueous and solvent concentrations of toluene were sampled continuously and monitored every two minutes by uv-vis spectrophotometry. Mass transfer measurements with toluene through a hydrophobic membrane (Millipore GVHP) were also attempted. It was found that toluene attacked a component in the sample tubing, which affected the 21 Chapter 2 Experimental techniques toluene solvent absorption readings and the mass balance could not be reconciled. The mass transfer study of this system was therefore abandoned. Mass transfer of iodine (0.001 M) from water to heptane was measured with two types of membranes (Millipore GVHP and WLP). The tests were run for 60-100 minutes and only the solvent concentration was monitored. Mass transfer of acetone (0.007 M) from heptane to water was measured for the two types of membranes by monitoring the aqueous phase acetone concentration. Procedure for cerium(IV) permeation tests Mass transfer of cerium was measured using the membrane permeation cell described in Section 2.6.2. Two types of mass transfer tests were performed, namely extraction and stripping. In the extraction tests the cerium was transported from aqueous to solvent, and in the stripping tests the transport was from solvent to aqueous. In the first instance, the concentration of cerium in the solvent phase was continuously monitored with the uv-vis spectrophotometer, and in the stripping experiments the aqueous concentration was monitored. In selected experiments final samples of solvent were taken and analysed by ICP. The cerium values determined by the two techniques were within 5 percent. For every test, a fresh membrane was used (GVHP for extraction and WLP for stripping). The instrument was calibrated every time the tungsten lamp was switched off. Calibration standards for the aqueous solution were made up from a standardised solution of cerium(IV) at an acid content matching that used for the tests. Calibration of the instrument for the solvent involved using an organic solution of the same composition as that of the test, and contacting with known amounts of cerium sulfate solution. An aliquot of each calibrating standard was stripped and analysed by ICP. Care was taken to use the same flowthrough cell for calibration and experimental tests. Examples of the two types of calibration curve are shown in Figure 2.1. Most permeation tests were run for 2-6 hours. The aqueous phases consisted of cerium (30 - 2900 ppm) in sulfuric acid (0.1 - 0.9 M). The solvent consisted of DEHPA (0.006 - 0.3 M), MEHPA (6 x 10-5 - 0.018 M) in heptane. For stripping tests the solvent was loaded with cerium (280 - 3600 ppm) and the aqueous phase was made up of sulfuric acid only. 22 Chapter 2 Experimental techniques 2.7 Experimental procedures for Chapter 6 2. 7. 1 Hollow fibre contactor experimental apparatus The experimental apparatus consisted of two hollow fibre modules, one for extraction and one for back-extraction or stripping. The tube-in-shell type modules were manufactured at ANSTO, with fibres purchased from Akzo Nobel Fraser AG, Germany. The fibres were first cut into appropriate lengths and threaded through a clear PVC pipe casing. The fibres were potted with epoxy resin and held to the casing by purpose built end pieces, machined from PVC. The physical details of the modules are given in Table 2.2. The rationale for choice of membrane is given in Chapter 6. TABLE 2.2 Hollow module characteristics Module Characteristics Extraction Stripping Module Module Housing material clear PVC clear PVC Potting mix epoxy resin epoxy resin Module internal diameter (mm) 15 35 Module length (m) 0.26 0.33 Packing fraction (%) 62 61 Akzo fibre type Q3/2 1.5/2 Hydrophibicity hydrophobic hydrophilic Fibre material polypropylene polyethersulfone Number of fibres 130 169 Internal diameter (mm) 0.6 1.5 Wall thickness (µm) 200 300 Pore size (µm) 0.2 0.2 The modules and accessory equipment were set up in two frames. A picture of the apparatus is shown in Figure 2.8. Digital pressure gauges (sensitivity 0.1 kPa) were placed at the inlet and outlet of the modules in both the tube and the shell sides. The wetted parts of the pressure gauges in contact with feed and solvent solution were stainless steel. In the aqueous strip circuit special hastelloy pressure gauges were used due to the corrosive nature of the liquid. The flowrate was monitored by an in-line flowmeter connected to a rate/frequency process monitor. The flowmeters were calibrated with water to read in ml min·1. Gear pumps (100-2000 ml min· 1) were used for the feed and solvent solutions, which were circulated through reservoirs. For the aqueous 23 Chapter 2 Experimental techniques flowmeter gear pump controls feed stripping solvent receiver uv-vis solution module reservoir solution spectrophotometer reservoir reservoir Figure 2.8 Photograph of the hollow fibre contactor apparatus 24 Chapter 2 Experimental techniques strip, a masterflex pump was used. All the equipment was connected by colour coded Teflon piping. The cerium feed concentration was continuously monitored by a uv-vis spectrophotometer. The strip and solvent concentrations were determined by periodic sampling and analysis of the liquors by ICP-OES. Batch experiments were run at 22°C for 200-300 minutes. A prolonged experiment was run continuously for 65 hours 2. 7.2 Measurement of entrained organic in the raffinate The volume of entrained organic in the feed was very small and the amount was determined by the following procedure. To each spent feed batch (raffinate), a known volume of diluent was added, generally in the ratio of 20 ml of diluent per litre of raffinate. The two phases were left to equilibrate for two weeks to allow for entrained organic to diffuse to the diluent. A sample of the diluent was then digested in the microwave with concentrated nitric acid and hydrogen peroxide, and the resultant aqueous solution analysed for phosphorous. 2.8 References 1. Jeffery, G. H., Bassett J., Mendham J. and Denny R.C. 1989. Vogel's Textbook of quantitative chemical analysis (Fifth Edition). Great Britain: Longman Scientific & Technical. 2. Partridge J. A. and Jensen R.C. 1969. Purification of di-(2- ethylhexyl)phosphoric acid by precipitation of copper(I I) di-(2- ethylhexyl)phosphate. J. lnorg. Nucl. Chem. 31: 2587-89. 3. Rao Y. R. and Achatya S. 1993. A rapid titrimetric determination of D2EHPA and M2EHPA. Hydrometallurgy- Technical Note 32: 129-35. 25 Chapter 3 Solvent selection and chemistry Summary This chapter discusses the reasons for the selection of a solvent system for the liquid - liquid extraction of cerium(IV) from acidic sulfate solutions, bearing in mind potential commercial usage in a solvent extraction or membrane assisted extraction process. It is shown that DEHPA is the most appropriate reagent for extraction of cerium, taking into consideration factors such as extraction ability, resistance to degradation, solvent selectivity and potential for sulfate transfer into a chloride strip solution. A new detailed study of the chemistry of extraction of cerium(JV) with DEHPA reveals that there are two regions of extraction, which are explained in terms of cation exchange and solvating mechanisms. The data are interpreted by both graphical and numerical analysis. Chapter 3 Solvent selection and chemistry 3.1 Introduction The selection of an extractant in a solvent extraction process, or carrier in a liquid membrane process is of paramount importance. For example, one of the properties that is attributable to an extractant is the degree of selectivity that can be achieved. The solvent extraction literature tends to deal with studies carried out under conditions which are as ideal as possible and differ from the operating conditions in many commercial processes. Choosing a solvent system that may potentially be used commercially is therefore not a straightforward procedure, as both practical and theoretical factors have to be taken into account (Ritcey 1979). This chapter discusses the reasons for the selection of a solvent system for the liquid - liquid extraction of cerium(IV) from acidic sulfate solutions, bearing in mind potential commercial usage in a solvent extraction or membrane assisted solvent extraction process. This work was presented at an international conference on solvent extraction. The detailed publication is shown in Appendix 3. This chapter also presents a new, detailed study of the chemistry of extraction of cerium(IV) with di-2-ethylhexyl phosphoric acid. 3.2 Background Solvent extraction is used to transfer a metal ion from an aqueous phase into a solvent phase. During this process, a metal ion must be rendered lipophilic and hydrophobic. This can be done by means of reactions with an organic extractant. The extraction reactions can be divided into three groups: acidic, basic and solvating. Each of these types of reactions will be discussed in the context of the literature available for extraction of cerium(IV). 3.2.1 Acidic extractants Acidic extractants contain ionisable hydrogen ions which can undergo cation exchange reactions with aqueous metal cations to form a neutral species in the organic phase. An example of such a reaction is given in Equation 3.1. 3.1 where HX is the organic acid. 26 Chapter3 Solvent selection and chemistry Di-2-ethylhexyl phosphoric acid (DEHPA) is an organic acid and is one of the most studied solvent extraction reagents for metal extraction. The literature available for the extraction of trivalent lanthanides with DEHPA is very extensive (Sato 1989), (Mori 1988), (Kopunec 1989), (AI-Janabi 1990), (El-Kot 1991 ), (Shakir 1991 ). Cerium(IV) extraction, however, has been less researched. The extraction of cerium(IV) from high nitrate solutions by DEHPA was first reported four decades ago by (Peppard 1957). This solvent system was utilized to purify 144Ce from other lanthanide fission products. Later studies showed that cerium(III) complexed with DEHPA and oxidized to cerium(IV) when in contact with high concentrations of nitric acid (Bray 1968), (Peppard 1959). Both proposed the formation of a polymerized mixed nitrate complex in the organic phase of the form Ce005N03R2 • The formation of complexes Ce(N03 )(HR2 ) 3 at nitric acid concentrations< 1 M, and Ce(N03 ) 4 (HR) 2 at nitric acid concentrations> 1 M, have been proposed by (EI-Yamani 1993) in studies undertaken on a macro scale. The system cerium(IV)-nitrate-2-ethylhexyl hydrogen 2-ethylhexylphosphonate (HEHEHP) was recently also studied, and Ce(N03 ) 3 HX2 was reported to be formed (Das 1995). Extraction of cerium(IV) with DEHPA from sulfate solution has only been studied by (Tedesco 1967), but this early work was limited to low acidities (pH= 0.5-2). These workers reported some dependence of the distribution coefficient on the equilibrium sulfate concentration, but found no evidence of sulfur in the organic phase, suggesting the extracted species were of the form Ce(HR2 ) 4 and CeO(HR2 ) 2 • More recently (Li 1984) undertook more detailed studies on a similar system of cerium(IV)-sulfate-mono (2-ethylhexyl)2- ethylhexyl phosphonate (HEHEHP). They proposed the formation of CeX2 (HX2 ) 2 at acid concentrations< 5 M H2S04 , and H 2Ce(S04 ) 3 .2HX at acid concentrations > 5 M H2S04• At organic loadings close to saturation (equilibrium 0.5 M H2S04), the extraction of sulfur was reported with the stoichiometry of the extracted complex averaging Ce(S04 ) 05 X 3 .xH20. Cerium(IV) can also be extracted by weakly acidic extractants such as carboxylic acids, but the extraction occurs at much higher pH values, where hydrolysis can become a problem at macro cerium concentrations. These 27 Chapter3 Solvent selection and chemistry reagents have been studied in the context of chromatographic analytical separation of cerium(IV) from other elements (Ray 1982) , (Mitra 1992), (Das 1995) 3.2.2 Solvating extractants This class of reagents extracts neutral inorganic aqueous complexes by virtue of the electron donating capacity of the solvents. The solvation of these neutral complexes increases their solubility in the organic phase. In the search for methods of isolation of 144Ce from other fission products of 235U, many such solvents have been identified for cerium(IV) extraction. Thus cerium(IV) in sulfuric acid medium can be extracted by solvents with high dielectric constants such as o-nitrotoluene and nitrobenzene (EI-Dessouk 1977) or solvents containing carbon - oxygen bonds such as methyl isobutyl ketone (Stamm 1964). Another group of solvating reagents are the esters of organophosphorous acids which contain oxygen bonded to phosphorous. The best known reagent of this type is tributyl phosphate (TBP). The extraction of cerium(IV) from nitrate solutions with TBP was first reported by (Korpusov 1962), and is the basis of most solvent extraction processes for cerium(IV). Two formulations have been suggested for the extracted complex; H 2Ce(N03 ) 6 (TBP) 2 (Korpusov 1962) and Ce(N03 ) 4 (TBP) 2 (Healy 1956). The separation factors for cerium(IV) over many other metal ions impurities is very high, and there has been little incentive to find alternative processes. The TBP/nitrate solvent system is still being proposed in some current investigations for the separation of cerium from Ce Am mixtures (Das 1995) and in the development of new analytical procedures for measuring the cerium content in glasses used in the nuclear industry (Raychaudhuri 1992). Recently the extraction of cerium(IV) from sulfuric acid solutions with Cyanex 923 was also studied and the formation of organic complex Ce(S04 )(HS04 ) 2 .2L proposed. Cyanex 923 is a phosphine oxide type reagent consisting mainly of a mixture of n-octyl and n-hexyl phosphine oxides (Lu 1998). 28 Chapter3 Solvent selection and chemistry 3.2.3 Basic extractants These reagents rely on the ability of the metal ion to form anionic complexes in the aqueous phase. Extraction occurs via an anion exchange reaction as exemplified by Equations 3.2 and 3.3. With the exception of quaternary ammonium extractants, all amine reagents need to be converted to an amine salt prior to extraction. 3.2 where HA is a mineral acid, and R3N represents a tertiary amine. 3.3 A quaternary ammonium reagent (Aliquat 336) was found to be effective for the separation of trace amounts of cerium(IV) from berkelium(IV) at 2 M nitric acid concentration, due to the differing abilities of the metal ions to form aqueous anionic nitrate complexes. The formation of organic complex (R4N)2 Ce(N03 ) 6 was reported (Moore 1969). Extraction of cerium(IV) from sulfuric acid solutions ranging from 0.1 to 2 M with a primary amine (Primene JMT) has also been reported. The proposed formulation of the extracted complexes was (H2 RH) 2 Ce(OH)2 (S04 ) 2 for sulfuric acid concentrations less than 0.1 M and (H2 RH)4 Ce(S04 ) 4 for sulfuric acid concentrations greater t~an 0.1 M (Nekovar 1997). 3.2.4 Solvent extraction processes for cerium(IV) Most of the literature on solvent extraction (SX) processes for cerium deal with extraction of cerium(IV) from nitrate media. Reviews by (Liddell 1984) and (Ritcey 1985) indicate that the process has changed little since the early reports (Korpusov 1962) on the extraction of cerium (IV) with tributyl phosphate. Back extraction is by reductive stripping with hydrogen peroxide, since in acidic solution, cerium(IV) is a stronger oxidizing agent than hydrogen peroxide. TBP (40%) in heptane was used by (Korpak 1971) to extract cerium(IV) (105 g L-1) from 4.5 M HN03• Hydrogen peroxide was used as a reducing agent. A strip liquor of purity 99.95% Ce02 was reported. (Hafner 1977) described the recovery of cerium (IV) from nitrate liquor (7.4 M HN03) obtained from the leaching of bastnasite. H3803 was used to complex fluoride and the strip 29 Chapter3 Solvent selection and chemistry solution was a mixture of 6 M HCI/H202'H3803• (Saleh 1966) recovered spectrographically pure Ce02 using a similar TBP process, with NaN02 as a reducing agent in the strip circuit. In this last process the starting material was monazite, but additional processing steps were included to remove Th, Fe, Si and phosphate. A similar TBP/nitrate process has also been used by (Preston 1996) to recover Ce02 (99.98% purity) from a phosphoric acid by-product. A major variation to the TBP solvent extraction process is the use of electrolytic oxidation of cerium(III) and extraction of cerium(IV) with the solvent. This technique was investigated by (Zhang 1982) to effect the separation of Ce from Pm in nitrate media using a mixed TBP/DEHPA solvent. A similar system for Ce separation from Am was reported by (Kedari 1997). Electro-oxidative extraction and electro-reductive stripping in the cerium-TBP-HN03 system has also been reported by (Ying-Chu 1987) and (Ying-Chu 1988). Limited information is available for the extraction of cerium(IV) from sulfate solutions. Work at the U.S Bureau of Mines (Douglass 1959) showed that sulfate leach liquors derived from bastnasite required the addition of 6-8 M HN03 in order to extract cerium (IV) with TBP. Sulfuric acid was used for stripping and the cerium was recovered by oxalate precipitation yielding a 99.6- 99.8% Ce02 product. More recently, Rhone-Poulenc filed a patent (Horbez 1989) for a process where cerous sulfate was electrolytically oxidised and simultaneously extracted into an organophosphorus acid (chemical name not disclosed). The solvent was stripped with 2.5 - 5 M sulfuric acid. This technique has also generated interest in the field of electrochemistry, where the use of solvent extraction eliminates the need to divide the reactors using a diaphram or membrane. In this context, oxidation of cerium(III) in sulfate media with subsequent extraction into DEHPA was reported by (Horbez 1991 ), (Horbez 1991) and (lbhadon 1992). 3.2.5 Summary of background literature The solvent extraction of the lanthanides has been extensively studied due to the importance of its application in the following areas: • Recovery from primary sources such as monazite and bastnasite and the purification of the 15 lanthanides from each other; and 30 Chapter3 Solvent selection and chemistry • Extraction of long lived actinides and trivalent fission products such as 144Ce. In terms of recovery of lanthanides from primary sources, the two most important reagents are clearly TBP and DEHPA. Extraction of cerium(IV) from nitrate media with these two reagents is well documented both in terms of laboratory studies and processing applications. The use of sulfate media for rare earth processing has attracted limited attention (Alstad 1977), (Zhang 1988). Only a few researchers, notably Deqian Li and coworkers have studied cerium(IV) in this context (Li 1984), (Lu 1998). The application of cerium(IV) to the indirect anodic oxidation of organic molecules, has sparked interest in the electrolytic generation of cerium(IV). In order to improve the efficiency of such a process, solvent extraction has been used to remove cerium(IV) as it is produced in the electrolytic cell in a sulfate medium (Horbez 1991 ). The extractant used for these studies was DEHPA. Details on the extraction chemistry of the cerium(IV)/DEHPA/sulfate system however are scarce, with only a single study at trace levels available in the literature. The work presented in this chapter fills the gap in the current knowledge, providing an assessment of the commercially available reagents for the extraction of cerium(IV) from sulfate media, with a particular emphasis on DEHPA and with due consideration given to important process parameters. 3.3 Results and discussion 3.3. 1 Solvent selection When considering the selection of a solvent system for either a solvent extraction process or a carrier for a liquid membrane process, there are some general requirements that must be satisfied (Thornton 1992), (Tavlarides 1987): • High distribution coefficients of the metal ion of interest must occur in a practical acidity range; • Loaded solvent must be easily stripped; • Good solvent selectivity; 31 Chapter 3 Solvent selection and chemistry • The solvent must be stable and have low solubility in both the feed and strip or receiver solution. • High solvent loadings are required for SX or membrane assisted SX processes, but this is not a pre-requisite for liquid membrane processes; • Acceptable rates of extraction and stripping; • Low cost and availability as a commercial product of consistent quality; • The solvent must be non-volatile and have acceptable flash point and toxicity This work focuses on commercially available reagents as these are known to have limited solubility in aqueous solutions, are of low cost and have, in most cases, been tested in other hydrometallurgical processes. The literature search indicated that acidic and solvating type organophophorous reagents, as well as amine type reagents, have the potential to extract cerium(IV). These solvent systems were therefore studied in detail. 3.3.1.1 Effect of aqueous media and acidity The extraction of cerium with selected neutral and acidic organophosphorous reagents was first studied as a function of aqueous acidity. Extraction curves for three organophosphorus acids, namely; DEHPA (di-2-ethylhexyl phosphoric acid), lonquest 801 (2-ethylhexyl 2-ethylhexylphosphonic acid) and Cyanex 272 (bis-2,4,4-trimethylpentyl phosphinic acid) are shown in Figure 3.1. Details of experimental procedures are given in Chapter 2, Section 2.4.1. With this class of reagent the extraction decreases with increasing acidity, as has been well documented for cerium(III) extraction (Sato 1989) . Cerium(IV) is readily hydrolyzed (Cotton 1980), (Mirza 1980), and therefore a relatively high concentration of acid is required in order to prevent precipitation in the aqueous phase. The order of extraction follows the pKa order of the reagents as shown in Table 3.1 and Figure 3.1. The lower the pKa, the higher the acid concentration where effective extraction can occur. Cerium(IV) extraction with DEHPA > lonquest 801 > Cyanex 272 32 Chapter3 Solvent selection and chemistry ------~-~-~----~--- )!(-· Ce(IIl)/chloride/DEHPA ~ Ce(IIl)/nitrate/DEHP A -o-- Ce(IIl)/sulfate/DEHPA a Ce(IV)/sulfate/lonquest 801 --* -.Ce(IV)/sulfate/Cyanex 272 ---e-- Ce(IV)/sulfate/DEHPA 100 . --··· -- -~lK-o Q lK \ = 80 ...0 ~ -~ i. 60 Q ~ \ i;.:i- ~~ u~ 40 0 ~ *\ = 20 . 0 \\\ 0~ 0 I I I 0.001 0.010 0.100 I.OOO 10.000 [H2S04] / M Figure 3.1 Extraction of cerium from sulfate solution with organophosphorus acids Ce(lll)/chloride/DEHPA: Organic - 0.1 M DEHPA in n-heptane, Aqueous - 0.003 M Ce in HCI Ce(lll)/nitrate/DEHPA: Organic - 0.1 M DEHPA in n-heptane, Aqueous - 0.003 M Ce in HN03 Ce(lll)/sulfate/DEHPA: Organic - 0.1 M DEHPA in n-heptane, Aqueous - 0.003 M Ce in H2S04 Ce(IV)/sulfate/DEHPA: Organic - 0.1 M DEHPA inn-heptane, Aqueous - 0.0024 M Ce Ce(IV)/sulfate/ lonquest 801: Organic - 0.27 M lonquest 801 inn-heptane, Aqueous - 0.014 M Ce Ce(IV)/sulfate/ Cyanex 272: Organic - 0.27 M Cyanex 272 inn-heptane, Aqueous - 0.014 M Ce 33 Chapter3 Solvent selection and chemistry The aqueous media also has an effect, and the extraction of cerium(III) from chloride, nitrate and sulfate media is also shown in Figure 3.1. The extraction curves for chloride and nitrate are superimposed as both form weak complexes with cerium(III). Formation of stronger cerium(III) sulfate complexes causes a shift in the extraction curve to lower acidities. At the high acid concentration required for cerium(IV) extraction, uptake of sulfuric acid itself by the solvent also has to be taken into account. Extraction of acid is undesirable from an economic point of view and because of the potential contamination of strip solution with sulfate ions. Extraction of sulfuric acid from 11 M H2S04 with the three acidic extractants was measured by determination of the sulfur content of the organic phase. The results are presented in Table 3.1. TABLE 3.1 Extraction of sulfuric acid at A:0=1 and [H 2SOJ0=10.8 M Reagent pKa Reference [Reagent] [S]o,9 water for pKa value M M Literature values This work DEHPA 1.9 (Miyake 1990) 0.36 0.056 1.51 (Komasawa 1981) lonquest 801 2.1 (Miyake 1990) 0.29 0.12 3.42 (Binghua 1996) Cyanex 272 6.37 (Sole 1992) 0.22 0.33 The data in Table 3.1 indicate that the reverse sequence to cerium(IV) applies to sulfuric acid extraction, with Cyanex 272 extracting the most acid. This is explained by the different mechanism of extraction of acid (solvating reaction) and metal ions (cation exchange reaction). Cyanex 272 extracts more acid because it is a stronger base by virtue of the electron donating effects of alkyl groups, when compared to the electron withdrawing properties of the ester groups on DEHPA. The generic structure of the reagents is presented in Figure 3.2 34 Chapter 3 Solvent selection and chemistry Acidic Type Reagents organo- organo- organo- phosphoric phosphonic phosphinic acid acid acid 0 0 0 II II II p p / / p OH OH OH ' R// ' R t ' RO ' RO RO R Solvating Type Reagents tri-alkyl di-alkylalkyl tri-alkyl phosphate phosphonate phosphine oxide 0 0 0 II II II p p p / ~ ~ OR R// OR R// ' R RO ' OR OR R Figure 3.2 Generic structure of organophosphorous reagents The arrows give an indication of the nature of the functional groups: Arrow pointing away from P=O for electron withdrawing and towards the P=O for electron donating. 35 Chapter3 Solvent selection and chemistry Of the organophosphorus acids studied, DEHPA is the most suitable for extraction of cerium(IV). High extractions occur at less acidic regions with lonquest 801 and Cyanex 272, where hydrolysis can become a problem. All three reagents require greater than 2 molar acid for stripping, or alternatively reduction of loaded cerium(IV) to cerium(III) which is not extracted at acidities greater than 0.1 M. A number of solvating type reagents were tested. These included: tributyl phosphate (TBP), dibutylbutyl phosphonate (DBBP), bis(2-ethylhexyl 2-ethylhexyl phosphonate (BEHEHP), trioctyl phosphine oxide (TOPO or Cyanex 921), Cyanex 923 (a mixture of phosphine oxides with n-octyl and n hexyl groups) and Cyanex 925 (a mixture of phosphine oxides with n-octyl and 2,4,4-trimethylpentyl groups). The generic structures are shown in Figure 3.2 and extraction of cerium(IV) as a function of acidity is shown in Figure 3.3. The extraction of cerium (Ill & IV) with TBP has previously been reported to only take place at concentrations > 5 M H2S04 (Douglass 1959). This work confirms that there is no significant extraction of cerium(IV) with either TBP, DBBP or BEHEHP in the 0.25-3.5 M sulfuric acid range. Cerium(IV) is well extracted with TOPO. On contact with Cyanex 923 & 925, cerium(IV) was reduced to cerium(III), which was not extracted. Pre-equilibration of the reagents with sodium carbonate eliminated this problem in the case of Cyanex 923 but not in the case of Cyanex 925. Extraction of cerium(IV) with the phosphine oxide type reagents is independent of acidity. Figure 3.4 shows that of the three phosphine oxide reagents, Cyanex 925 extracts the least amount of acid, given similar molar reagent concentrations. However, because of the problem of reduction of cerium(IV), no further tests were undertaken with this reagent. These preliminary tests have indicated that of the solvating extractants, TOPO and Cyanex 923 are likely to be suitable. The extraction of cerium(IV) with a basic reagent, Alamine 336, was also tested (Figure 3.3). It was found that high distribution coefficients could be obtained at sulfuric acid concentrations of less than 1 M. Conversely, stripping or back extraction with greater than 2 M acid is possible. 36 Chapter3 Solvent selection and chemistry -A-- TOPO (0.18 M) o Cyanex 923 (0.47 M) :l!E-- D8BP(1.14M) 100 + • --- Alamine 336 (0.20 M) = .s.... 80 :------t-- $ ---A------&-----1:J. u . ------._~ OS •- - ...... 60 .. ~ • ~ u~ 40 • 0~ 20 0 • 0.1 ______[H 2SO1_.0 4 ) ______/ M ___j 10.0 I Figure 3.3 Extraction of cerium(IV) with solvating extractants TOPO: Organic-0.18 M TOPO inn-heptane, Aqueous-0.005 M cerium(IV) Cyanex 923: Organic - 0.47 M Cyanex 923 in n-heptane, Aqueous - 0.005 M cerium(IV) DBBP: Organic - 1.14 M DBBP in n-heptane, Aqueous - 0.005 M cerium(IV) Alamine 336: Organic- 0.2 M Alamine 336 inn-heptane, Aqueous - 0.016 M cerium(IV) ------~ 0.20 -,------, _ • Cyanex 923 (0.47 M) _ • _ TOPO (0.45M) ~ , . .I 0.15 ---1r------Cyanex 925 (0.42 M) : I ~ DBBP (l.lM) i ----•---- TOPO (0.18 M) - I "i 0.10 Ill /~ ..t' I I ~ • • I/~ 0.05 • A -• -.- c=t.:'. -.- X 0.00 *-----=~;;;.:;;;i~~~=-~------l 0.1 1.0 10.0 ("2S04)/M Figure 3.4 Extraction of sulfuric acid with solvating extractants TOPO (0.18 M) : Organic - 0.18 M TOPO in n-heptane TOPO (0.45 M) : Organic - 0.45 M TOPO in Shellsol 2037 at 40°C ( higher temperature used to increase solubility of TOPO) Cyanex 923: Organic- 0.47 M Cyanex 923 inn-heptane DBBP: Organic - 1.14 M DBBP in n-heptane Cyanex 925: Organic - 0.42 M Cyanex 925 in n-heptane 37 Chapter 3 Solvent selection and chemistry 3. 3. 1. 2 Solvent selectivity In the purification of cerium, it is important to take into account the selectivity of the reagent for cerium(IV) with respect to other impurities which are likely to be present in solution. Primarily the solvent must be able to reject the trivalent lanthanides. Selectivity with respect to Th is also important as this element is present at very high levels in minerals containing rare earths such as monazite (Hart 1988). Selectivity is usually measured in terms of separation factors(~). defined as the ratio of the distribution coefficients of the two metal ions (Rice 1981 ). Table 3.2 shows the separation factors obtained for cerium(IV) over Th and representatives of the light, middle and heavy trivalent rare earths. Separation factors with respect to trivalent lanthanides are very high for Alamine 336, Cyanex 923 and TOPO, and compare well with data obtained by (Preston) for separation of cerium(IV) by TBP from nitrate solution. With DEHPA, selectivity of cerium(IV) over the trivalent rare earths decreases along the series light to heavy. In addition, DEHPA extracts Th preferentially to cerium(IV). However, some opportunity exists for the selective stripping of loaded cerium by reducing it to the trivalent state, and therefore the problem of Th extraction with DEHPA and its separation from cerium(IV) is not insurmountable. TABLE 3.2 Separation factors with selected reagents ( [H 2SOJ0 = 0.5 M, [Ce]0 = 0.007 M ) DEHPA Alamine 336 Cyanex 923 TOPO TBP* 0.15 M 0.062 M 0.24 M 0.18 M 15 vol.% 3 3 3 >103 /3Ce(JV) _ /3Ce(JV) >10 >10 >10 La(//l) - 4600 Ce(Jll) 3 >103 >103 /3Ce(JV) _ /3Ce(/V) 275 >10 Gd(Jll) - 1900 Sm(!//) 3 /3Ce(/V) 2.3 103 >103 >10 /3Ce(/V) = 5900 Y(/11) Yb(l/l) /3Ce(/V) 0.05 >102 8.9 14 Th(/V) TBP* Data reproduced from (Preston), Aqueous: 2 M NH4N03/ 1 M HN03 38 Chapter3 Solvent selection and chemistry 3.3. 1.3 Solvent stability Solvent stability is an important issue in any solvent extraction or liquid membrane process. Cerium(IV) is a very strong oxidising agent with a reduction potential of 1.46 volts in 0.5 M H2S04 (Electrochemical Series 1980). It therefore has the potential of oxidising the extractant and diluent with prolonged contact. TOPO, DEHPA and Alamine 336 were tested for degradation using the procedure outlined in Chapter 2, Section 2.4.2. Heptane was used as a diluent as aromatic compounds have been shown to be easily oxidised (Grover 1969), (Baroncelli 1965). The reduction of cerium(IV) from a nitrate/TBP system has since been measured by (Preston 1996). They confirmed that non-aromatic diluents, such as iso-octane and n-heptane, were the most resistant to oxidation. In this work, the degradation of the solvent was not measured directly, but the Cerium(III & IV) organic and aqueous concentrations were monitored as a function of time. The decrease in cerium(IV) organic concentration was accompanied by a corresponding increase in aqueous cerium(III) concentration, so that the overall cerium concentration in the system was constant. The rate of this decrease was assumed to be an indication of the rate of degradation of the solvent. Plots of the decrease in cerium organic loading (ie [Ce]/[Ce],0 ,a,) as a function of time are shown in Figure 3.5. Data obtained by (Warf 1947) have also been included to provide a basis for comparison with TBP degradation from a nitrate system. Under the experimental conditions shown, the half-life of the solvents (50% decrease in cerium organic concentration) was 12 hours, 25 days, 32 days and 57 days for Alamine 336, DEHPA, TBP and TOPO, respectively. Alamine 336 was also tested after equilibration with Na2C03 to remove possible reducing impurities, but 50% cerium reduction still occurred within 24 hours. Thus on the basis of degradation alone, Alamine 336 is not suitable for extraction of cerium(IV). The degradation rates of the other three reagents are of the same order of magnitude, with TOPO being the most resistant to oxidation. Considering that TBP has been used commercially to extract cerium(IV), its is assumed that the cost of reagent replacement due to degradation of DEHPA and TOPO from a sulfate system will not be prohibitive in a potential process application. 39 Chapter3 Solvent selection and chemistry 3.3.2 Chemistry of extraction of cerium with TOPO and Cyanex 923 The screening tests presented in Section 3.3.1 of this chapter indicated that the solvating reagents TOPO and Cyanex 923 were appropriate reagents for cerium(IV) extraction and therefore further work was undertaken to elucidate the chemistry of the reaction. With both reagents, extraction of cerium(IV) was found to be independent of acidity in the 0.25 - 4 M sulfuric acid range (Figure 3.3), but extraction of sulfuric acid itself was quite significant (Figure 3.4). Extraction of sulfuric acid with TOPO has been measured by (Shuyun 1998), who proposed the formation of the [(TOP0) 3 H 2S04 ] complex in the organic phase. In order to separate the effect of sulfuric acid extraction from sulfate extraction associated with a cerium complex, the results were plotted as the molar ratio of S:Ce measured in the organic (Figure 3.6). The minimum molar ratio of S:Ce in the organic phase occurs at low acidity, where sulfuric acid extraction is not significant, and it approaches a value of two, which is presumed to be an indication of extraction of neutral species Ce(S04) 2. From the above discussion, a general equation for the extraction of cerium(IV) with these solvating reagents is proposed: 3.4 [Ce(SO) L] where L = TOPO or Cyanex 923 and Kex, = 4 2 n [Ce(S04 ) 2 ][Lr D = [Ce] 3.5 [Ce] From Equations 3.4-3.5, it can readily be shown that: logD = logKex, + nlog[L] 3.6 Logarithm plots of the distribution coefficient versus the molar extractant concentration are therefore expected to be straight lines, with slopes equal ton. These plots are shown in Figure 3.7, and are indeed straight lines with slopes close to 2, pointing to the formation of Ce(S04 ) 2 L2 in the organic phase. 40 Chapter3 Solvent selection and chemistry 100 -'.f. 80 I,,. I,,. ~. - 0 I,,. 0 I,,. TOPO 0 DEHPA :r TBP* -a-- Alamine 336 i 0 5 10 15 20 25 30 35 40 Time (days) L ___ ·----~------~ Figure 3.5 Decrease in loaded cerium(IV) as a function of time TOPO: 0.014 M cerium(IV) loaded in 0.1 M TOPO inn-heptane in contact with 0.75 M H2S04 DEHPA: 0.058 M cerium(IV) loaded in 0.15 M DEHPA inn-heptane in contact with 0.75 M H2S04 Alamine 336: 0.02 M cerium(IV) loaded in 0.06 M Alamine 336 inn-heptane in contact with 0.75 M H2S04 TBP*: Data reproduced from Warf - 100% TBP (vacuum distilled) in contact with 1 N HN03 and 0.5 N (NH4)2Ce(N03)s 48 -1r-TOPO(O.l8M) 44 0 l .9 40 ...... Cyanex 923 (0.42 M) I -i. 36 f = ~ u~ 32 I ~ 28 I i. 24 = 0 0 20 c I,,. CJ 16 ·= 12 ~ 0 i.= 8 ~• I,,. 0 4 f···_····_·····_····~~~-·-.·-~~~~~--~·~=~·''·'·~·······_·····~·····_····_·····~····.·_·····~·····_·····~·····~······~.., 0 0.1 1.0 10.0 [H2S04] /M Figure 3.6 Molar ratio of S:cerium(IV) in the organic phase TOPO: Organic - 0.18 M TOPO in n-heptane, Aqueous - 0.005 M cerium(IV) Cyanex 923: Organic- 0.47 Cyanex 923 inn-heptane, Aqueous - 0.005 M cerium(IV) 41 Chapter3 Solvent selection and chemistry Recent studies undertaken by (Lu 1998) suggest that the cerium(IV)-Cyanex 923 complex formed by extraction from 1 - 4 M sulfuric acid is of the form Ce(S04 )(HS04 ) 2 .2L, with a Ce:S ratio of 1:3. This ratio was not measured directly. The structure was proposed due to the presence of characteristic bisulfate absorption frequencies in the IR spectra of the solvent phase. The author does not seem to have considered that the presence of sulfuric acid in the solvent may have given rise to the characteristic bisulfate absorption peaks. In either case, the nature of the extraction reaction of cerium(IV) with Cyanex 923 means that the transfer of sulfate ions from feed to strip solution cannot be prevented. The extraction reaction also indicates that solutions containing low sulfate concentrations would be suitable for stripping. It was found that both TOPO and Cyanex 923 could be effectively stripped with 0.1 M HCI. The solubility of TOPO in aliphatic diluents places a restriction on the maximum solvent loading that is achievable. This is an important constraint in a solvent extraction process, but not so in a liquid membrane process. In solvent extraction processes, the higher the solvent capacity, the cheaper the process becomes. In liquid membrane processes, because extraction and stripping occur simultaneously, the solvent is never fully loaded. At 20°C, the maximum solubility of TOPO in aliphatic diluents is 100 g L-1 TOPO, corresponding to a maximum loading capacity in the solvent of 17 g L-1 Ce. Cyanex 923, being a liquid at room temperature, can be prepared at much higher molar concentrations and will achieve higher solvent loading. Therefore, whilst TOPO maybe quite appropriate for use as a carrier in a liquid membrane process for cerium, Cyanex 923 maybe the best solvent for a corresponding solvent extraction process. 3.3.3 Chemistry of extraction of cerium with DEHPA Of the reagents chosen as potential solvents for the extraction of cerium(IV), DEHPA shows some promise in that it extracts cerium(IV) in the required acidity region, without significant sulfuric acid extraction, and is fairly resistant to oxidation. Since cerium(III) is only weakly extracted in dilute acid, reduction of loaded cerium(IV) provides an option for stripping. The main disadvantage is its lower selectivity with respect to other trivalent rare earths and Th. Further work was therefore undertaken to determine the extraction chemistry, since no extensive studies were found in the literature on this particular system. 42 Chapter3 Solvent selection and chemistry 3.3.3.1 Acidity range of extraction The extraction of cerium(IV) was measured as a function of sulfuric acid concentration. Results are presented in terms of the distribution coefficient and are shown in Figure 3.8. Two extraction regions are apparent; in aqueous solutions ranging from 0.5 to 5 M sulfuric acid, extraction decreases with increasing acid concentration and at greater than 0.5 M acidity, extraction increases with increasing sulfuric acid concentration. In a separate set of experiments, the extraction of sulfuric acid with no metal ion present in the aqueous phase was tested. It was found that significant sulfuric acid extraction occurred only at aqueous sulfuric acid concentrations above 5.5 M H2S04 , coinciding with the second region of cerium extraction. 3.3.3.2 Effect of DEHPA concentration The effect of DEHPA extraction was studied in a series of tests where the sulfuric acid concentration was kept constant and the concentration of DEHPA in the solvent varied. Results for acidity ranges of 0.5 - 5.5 M H2S04 and 5 - 10 M H2S04 are shown in Figures 3.9 and 3.10, respectively. The logarithm plots of the distribution coefficient versus the dimeric DEHPA concentration are straight lines with slopes close to two for the entire acid range studied. 3.3.3.3 Effect of solvent loading All the data discussed previously were obtained under low loading solvent conditions, where the extractant is in excess at a ratio of DEHPA:cerium(IV) > 50. Under these conditions, very low concentrations of sulfur were detected in the organic phase. To test the effect of solvent loading on the type of species formed in the organic phase, 0.15 M DEHPA was progressively loaded with cerium(IV). The equilibrium acidity was kept at 0.75 M sulfuric acid. It was found that the Ce:S molar ratio in the organic phase was greater than 10 at low loading, decreasing to 1.5 at high loading. At maximum loading the DEHPA:Ce molar ratio approaches 3:1, which is in contrast to the DEHPA:Ce ratio dependence found at low loading. The concentrations of cerium and sulfur at various degrees of solvent loading are shown in Figure 3.11. The experimental procedure is described in Chapter 2, Section 2.4.1. 43 Chapter3 Solvent selection and chemistry 1.01 I 0.5 ------Cyanex 923 i '; 0.0 L -+-TOPO 1.£ -0.51 ~ I -1.0 -1.5 ~~~~ -~--~--~~. -2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 log (L] (M) Figure 3.7 Distribution coefficient of cerium(IV) versus extractant (L) concentration TOPO: Aqueous - 0.005 M cerium(IV), 0.2 M H2S04 Cyanex 923: Aqueous - 0.005 M cerium(IV), 0.2 M H2S04 JOO.OOO JO.OOO + + + 0 + b I.OOO I u" D * Cl 0.100 0 *+ f D I 0 * / .\ J_.·· -+ -0.3M 0.010 D 0.\ ;t: o\+o O .. 0.15M D . o 0.08 M 0.001 0.1 1.0 10.0 100.0 [H2S04] / (M) I ~~~~~--J Figure 3.8 Distribution coefficient of cerium(IV) as a function of [H2S04) Solvent: 0.08 - 0.3 M DEHPA in n-hepatne: Aqueous - 0.016 M cerium(IV) 44 Chapter3 Solvent selection and chemistry . ------, +S.25 1.50 J M H2S04 I 1.00 .4.1 M H2S04 ! 0.50 3M H2S04 # x2 M H2S04 0.00 :t;I M H2S04 • l -0.50 e0.5 M H2S04 • )K X Cl -1.00 X Oil • 0 )K ...l -1.50 X • -2.00 • :t: X • '' )I( • -2.50 X • -3.00 • -3.50 -4.00 ' -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 Log [(HR)i] Figure 3.9 Dependence of the distribution coefficient of cerium(IV) on the dimeric concentration of DEHPA: 0.5 - 5.5 M [H2S04] range [H2S04] = 0.5 M, Slope of curve = 2.0, r2 = 0.990, [H2S04) = 1.0 M, Slope of curve = 2.0, r2 = 0.985 [H2S04] =2.0 M, Slope of curve= 2.0, r2 = 0.960, [H2S04) = 3.0 M, Slope of curve= 1.8, r2 = 0.960 [H2S04] = 4.1 M, Slope of curve = 1.9, r2 = 0.999, [H2S04) = 5.25 M, Slope of curve = 2.0, r2 = 0.987 1.00 0.50 • OOO <>5 25 M H2S04 • -0.50 +S.55 M H2S04 • el0.3 M H2S04 -1.00 • ~ Cl oo -1.50 • • 0 ...l .~ -2.00 • -2.50 • -3.00 <> -3 .50 ~ ~ + slope 2 -4.00 +J---.,------,------.-----.- -2.50 -2.00 -1.50 -1.00 -0.50 0.00 Log [(HR)z] Figure 3.1 o Dependence of the distribution coefficient of cerium(IV) on the dimeric concentration of DEHPA: greater than 5 M [H2S04] range [H2S04] = 5.25 M, Slope of curve= 2.0, r2 = 0.987, [H2S04) = 5.55 M, Slope of curve= 2.2, r2 = 0.998 [H2S04] = 10.3 M, Slope of curve= 1.9, r2 = 0.998 45 Chapter3 Solvent selection and chemistry 3.3.3.4 Extraction reaction in the acidity range 0.5 to 5 M H2S04 The reaction of cerium(IV) with DEHPA at low solvent loading can be written in general terms as a cation exchange reaction 3. 7 by analogy with other tetravalent cations (Juang 1993). The dimerisation constant for DEHPA has been measured in n-heptane (Ajawin 1983) and other aliphatic diluents (Huang 1986). All workers have reported that DEHPA exists in these diluents mainly as a dimer. Therefore: where p and q are integers~ 1 and O respectively, giving, K = _[(_Ce_R_4_(H_'R_)-'--q)-'-P_][_H_+_]4_P 3.8 ext [Ce4+Y[(HR)2Y((4+q)/2) The degree of polymerization of cerium species in the organic phase is given by p and can be determined by varying the distribution of cerium at constant acidity and extractant concentration. With low solvent loading the extractant is present in vast excess and the concentration can essentially be considered constant. In this case Equation 3.8 can be expressed as: log[( CeR4( HR) q) P] = p log[Ce 4+] + constant 3.9 A plot of log[Ce] versus log[Ce] is shown in Figure 3.12, and the slope is close to one, indicating that only monomeric species are formed in the organic phase under these conditions of low solvent loading. In sulfate media, cerium(IV) forms strong sulfate complexes. The available literature (Sillen 1964), (Sillen 1971 ), (Hogfeldt 1982) on stability constants of reactions of the type given in Equation 3.10 is summarised in Table 3.3. 46 Chapter3 Solvent selection and chemistry TABLE 3.3 Stepwise formation constants (K) for cerium complexes in sulfate solution Complex log K Media Temp. Reference 4 2 Ce ++ iSOt ~ Ce(S04 )t ; 2 Ce(S04) + 3.3 2 M HCI04 25°c (Sillen 1964) 4.78 5.9 M HCI04 25°c (Sillen 1971) 2.6 variable 19°c (Sillen 1971) Ce(S04h 3.56 5.9 M HCI04 25°c (Sillen 1971) -0.40 variable 19°c (Sillen 1971) Ce(S04h2- 1.88 5.9 M HCI04 25°c (Sillen 1971) 1.17 0.1 M H2S04 35°c (Sillen 1971) 4 Ce ++iHSO; ~ Ce(S04 )i-i; +iH+ Ce(S04)2+ 3.54 2 (W, NaC104) 25°c (Hardwick 1951) Ce(S04h 2.30 2 (W, NaC104) 25°c (Hardwick 1951) Ce(S04h2- 1.30 2 (W, NaC104) 25°c (Hardwick 1951) Ce(S04 ) 2 +iHSO; ~ H;Ce(S04 );~; HCe(S04h- -0.2 variable (H 2S04) 20°c (Sillen 1971) 0.53 5 M LiC104 25°c (Hogfeldt 1982) H2Ce(S04)l· 0.3 variable (H 2S04} 20°c (Sillen 1971} Three complexes of the form Ce(S04 ):-i; (i = 1 to 3) are the most common, with the formation of HCe(S04 ); and H 2Ce(S04 )i- also reported. In this work, only complexes of the type Ce(S04 ):-i; (i = 1 to 3) have been taken into consideration in the treatment of the data. Assuming only a single cerium species is present in the organic phase, the distribution ratio can be written as: 3.11 incorporating Equations 3.10 and 3.8 into 3.11 and rearranging, 47 Chapter3 Solvent selection and chemistry 4 logD + log A= logKext + ( ; q) log[(HR) 2 ]-4log[H+] 3.12 3.13 In this work constants reported by (Hardwick 1951) at ionic strength of 2 have 3 5 7 been used. The values are p1 = 3.4 x 10 , p2 = 6.9 x 10 and p3 = 1.3 x 10 . Equilibrium concentrations of [H+] and [HSO;] were calculated taking into account the second sulfuric acid dissociation constant as outlined in Appendix 1, although the contribution of this reaction to the overall concentration of [HSO;] was not significant in the acidity ranges used for this work. According to Equation 3.12, at constant acidity, a plot of log D versus log[(HR) 2 ] should be a straight line with slope equal to (4+q)/2. Such plots are shown in Figure 3.10, and the slope is two (ie q = 0) for the acidity range studied (0.5 - 5.5 M H 2S04). This points to formation of complex CeR4 in the organic phase according to Equation 3.14. 3.14 The loading curve presented in Figure 3.11, shows that at very high solvent loadings, the cerium concentration in the solvent phase exceeds that expected from a maximum molar ration of HR:Ce = 4: 1, as proposed in Equation 3.14. Clearly, the speciation at low and high solvent loadings is not the same. In order to determine the conditions under which Equation 3.14 is valid, the apparent stoichiometric constant, at various solvent loadings, was calculated using Equation 3.12 and data from Figure 3.11. The degree of solvent loading (a) is defined as per Equation 3.15. The data are presented in Figure 3.13, and it is shown that the apparent Kext remains constant for as 0.1. . [Ce] solventloadmg(a) =--- 3.15 [(HR)2lo 48 Chapter3 Solvent selection and chemistry 7.0 HR:Ce = 3 ::: ---:Ce=4 ::.,__ ~ ::: . : : ::::::r :: _ i 4.0 .--• 0.8 -t:::! s ,. ·-· -~~:::"' -:!9 o 3.o I • o.6 J' S:!. / ,t· ~ 2.0 , 11 0.4 1n ;..• Q2 Ill 0.0 f-~t--L_____j~--+~---l------'-----i ~--+-~·----+----"-----+ 0.0 0 5 10 15 20 25 30 35 40 [Ce].q (g/L) l----~ Figure 3.11 Loading curve for cerium(IV) with 0.15 M DEHPA at 0.75 M H2S04 -2.0 --,------, -2.5 I~ -3.o 0) 0 ....I -3.5 -4.0 -+---,------,----,----,------1 -3.0 -2.8 -2.6 -2.4 -2.2 -2.0 Log [Ce] Figure 3.12 Effect of initial aqueous cerium concentration Solvent: 0.15 M DEHPA in n-heptane: Aqueous - 0. 78 M [H2S04J Slope of curve= 0.97, r2 = 0.996 49 Chapter3 Solvent selection and chemistry The data obtained from the loading curve also indicated that significant amounts of sulfur were transferred into the organic phase with cerium extraction. The amount was dependent on the degree of solvent loading: Ce:S ratios greater than 10 were measured for low solvent loading, but this ratio approached a value of 1.5 at very high loading. Since the presence of sulfur cannot be explained by sulfuric acid extraction, it is reasonable to assume that organic cerium sulfate complexes of the type shown in Equation 3.16 are being formed in the organic phase. 3.16 3.17 In order to determine the exact speciation of the organic solvent, a numerical analysis was carried out on data points obtained at a. ~ 0.1 and ionic strength close to 2. An ionic strength of 2 corresponds to an equilibrium acidity equivalent to 0.66 M H 2S04• Equations 3.8 and 3.17 were incorporated into the distribution coefficient expression (Equation 3.11) as follows: ~q=2 K [(HR) ]'4+q> 12 ~ 1=2 K [(HR) ]< 4+'> 12 [HSO-] D = L.q=O (l,q) 2 + L...i=O (s,I) 2 4 3.18 A[H+]4 A[H+]3 The K values were solved for by minimizing the sum of squares error function as shown in Equation 3.19, and the standard deviation (cr) calculated according to Equation 3.20. The values of q and t were changed from O to 2, assuming either a single or two species presence in the organic phase, and ssresid and cr calculated for each model. SS resid = L (log Dexperimental - log Dcalcu/ated) 3.19 u = ~ssresid/N , N= number of data points 3.20 The minimum standard deviation was obtained for a model with two complexes CeSO R (HR) formed in the organic phase, with an average of CeR 4 and 4 2 2 Ce:{HR)2 ratio of 1:2. This is in keeping with the slope dependence of 2 shown 50 Chapter3 Solvent selection and chemistry in Figure 3.9. The stoichiometric constants for reactions 3.14 and 3.21 are given in Table 3.4. These have been calculated taking into account bisulfate complexation of cerium(IV) in the aqueous phase (Hardwick 1951) and are valid for ionic strength equal to 2 and solvent loading values of a ~ 0.1. The complex CeR 4 is the dominant species. TABLE 3.4 Stoichiometric constants for reaction 3.14 and 3.21 at ionic strength of 2 logKo,o) 8.7 max 8.8 log K maximum calculated as log (K + 3crK) Finally, the dependence of the distribution coefficient over the whole acidity range of 0.5 to 5 M H 2S04 , is presented in Figure 3.14. The data have been plotted in terms of the calculated hydrogen ion activity. This has been done because the whole data set was not obtained at constant ionic strength, and considering the high acid concentrations used, hydrogen ion activities are more meaningful than molar concentrations (Sella 1987). Details of hydrogen ion activity calculations are shown in Appendix 2. The whole data set obtained at low loading (68 points) lies in a single curve, with the slopes at the extremities being close to 4 and 2 at low and high acid concentration ranges, respectively. This confirms the domination of the species CeR4 at low acidity and indicates that the formation of CeS04 R2 (HR) 2 becomes important as the acidity increases. 3.3.3.5 Extraction reaction in the acidity range greater than 5 M H2S04 Figure 3.8 shows that when the aqueous sulfuric acid concentration exceeds 5 M, the extraction of cerium(IV) increases with increasing acidity. In this region, extraction of acid by DEHPA becomes significant. In order to determine the type of complex formed in the organic phase, the distribution of sulfur was measured, with and without the presence of cerium. Results of these 51 Chapter3 Solvent selection and chemistry l.E+13 --r------ 1.E+12 - ~ .J l.E+l 1 .,. - e ...., ~ 8:: l.E+lO _ ...... ! ...... < l.E+09 ~ 1.E +08 +-_ _.__...._ ...... y. _ ___,,____.__...... _.~...J 0.01 0.10 1.00 Solvent loading Figure 3.13 Effect of solvent loading (a) on the apparent extraction constant 10000.0 ·l...... slope4 IOOO.O ·.. \ N -N IOO.O .,. ~ =...... IO.O -Q \tf* 1.0 ·-...... \\· .....f...... + slope 2 ...... , 0.1 ····•······· 0.010 O. IOO I.OOO JO.OOO I {H} Figure 3.14 Effect of hydrogen ion activity on the distribution coefficient of cerium (IV) at acidity 0.5 - 5 M [H2S04] [DEHPA] = 0.02 - 0.3 M, [Ce]o = 0.002 - 0.016 M 52 Chapter3 Solvent selection and chemistry experiments are shown in Table 3.5. Taking into account the amount of sulfur present in the organic phase due to the distribution of sulfuric acid, an approximate S:Ce molar ratio of 3 was obtained for the extracted complex and therefore the reaction proposed is: 3.22 where Kex, [(H2Ce(S04)3)(H~[H+] = 3.23 4 3 [Ce +][HS0;] [(HR) 2 ]2 and D = [(H2Ce(S04 ) 3)(HR)4 ] 3.24 4 [Ce +] + [CeSOt] + [Ce(S04 ) 2 ] + [Ce(S04 );-] In this region of high sulfuric acid concentration and considering the high values for the stability constants of cerium sulfate complexes, it is reasonable to 4 assume that [Ce(S04 );-] >> [Ce +]+[Ceso;+]+[Ce(S04 ) 2 ]. Therefore, incorporating Equation 3.23 into 3.24 and rearranging: D Kex,[(HR) ]2[HS0;]3[Ce4+] = 2 3.25 (fi3[HS0;]3[Ce4+])l[H+]2 where p is the overall stability constant defined by equation 3.26 [Ce(S0 )2-][H+]3 4 3.26 p3 = [Ce 4+][HS0;]3 log D = log Kex, - log p3 + 2 log[(HR) 2 ] + 2 log[ H+] 3.27 According to Equation 3.27, the plot of log D versus log[(HR)2 ] at constant acidity is expected to yield a slope of two. This has already been demonstrated to be the case in Figure 3.10. A similar plot of acid dependency, at constant extractant concentration, should also have a slope of two. Such a plot is shown in Figure 3.15, and indeed, when hydrogen ion activity is used rather than hydrogen ion molar concentrations, a slope of two is obtained. Hydrogen ion activities were calculated from the sulfuric acid molar concentrations as detailed in Appendix 2. 53 Chapter 3 Solvent selection and chemistry --·1-.D 0.12 M DEi-iPA using {H} • 0.12 M DEHPA using [H] t:i. 0.3 M DEHPA using {H} 10.000 .------I.OOO • • Q 0.100 0.010 10 100 1000 [HJ or {H} /M Figure 3.15 Effect of hydrogen ion activity {H} and hydrogen ion molar concentration [H] on the distribution coefficient of cerium(IV) at greater than 5 M H2S04 Solvent: 0.12 M DEHPA, acidity as {H+}, Slope of curve= 2.48, r2 = 0.995 Solvent: 0.12 M DEHPA, acidity as [H+], Slope of curve= 11, r2 = 0.996 Solvent: 0.3 M DEHPA, acidity as {H+}, Slope of curve= 2.05, r2 = 0.995 54 Chapter3 Solvent selection and chemistry TABLE 3.5 Molar sulfur to cerium ratios in the organic [HR] [H2S04]aq [Ce lo,9anic total [S]o,ganic [S]due to acid S:Ce N M g L-1 g L-1 g L-1 Molar 15.0 0.3 0.025 0.036 0.020 2.8 20.0 0.3 0.46 0.67 0.37 2.9 20.8 0.12 0.074 0.20 0.16 2.1 22.7 0.12 0.34 0.58 0.37 2.7 25.1 0.12 0.68 1.27 0.97 1.9 25.2 0.3 1.8 4.2 2.77 3.4 25.2 0.15 1.4 2.0 1.23 2.4 3.4 General discussion In the context of cerium(IV) recovery from sulfate solution, one of the primary requirements for a reagent is that it extract at sulfuric acid concentrations of around 0.5 M and be easily stripped. A number of reagents fulfil this requirement and they include the organophosphorous type acids (DEHPA, lonquest 801 and Cyanex 272), phosphine oxides TOPO and Cyanex 923 and the amine type reagent Alamine 336. The organic acids and the amine reagent can be stripped by contact with higher (> 4 M) acid concentrations. For the two phosphine oxides contact with a solution of low sulfate concentration is sufficient. In all cases, reductive stripping is effective. Acid stripping does not necessarily need to be carried out with sulfuric acid. In fact, stripping with another mineral acid such as hydrochloric acid provides an opportunity to convert the cerium into the chloride form, which is more convenient for subsequent product recovery steps_ Subsequent precipitation products obtained from sulfate media tend to carry considerable amounts of sulfate. With this in mind it is desirable to minimise the transfer of sulfate from feed to strip. Ignoring entrainment, carry over can occur in one of two ways. The first is by the complexation reaction involving cerium sulfate complexes. The second is by extraction of acid by the reagent. Acid extraction does not occur with organophosphorous acids to a significant extent at sulfuric acid concentrations < 5 M, and the lower the pKa of the reagent the less acid it extracts. Acid extraction with solvating reagents TOPO 55 Chapter3 Solvent selection and chemistry and Cyanex 923 is significant even at low (0.2 M) sulfuric acid concentrations. Furthermore, sulfate is involved in the cerium complex in the organic. From these reasons DEHPA has been chosen as the reagent with the greatest potential. Detailed studies with DEHPA revealed that although the most prevalent complex in the organic phase at an equilibrium acidity of 0.5 M H2S04 is CeR4 , involving no sulfate, this is only true when an excess of extractant is available. At high solvent loading, another complex involving sulfate is also present. The use of DEHPA in a solvent extraction process, where high solvent loadings are a prerequisite, will therefore not avoid the transfer of some sulfate from feed to strip solution. An important difference in the application of the solvent DEHPA to a liquid membrane process, lies in the fact that due to the simultaneous loading and stripping occurring, the solvent does not need to be fully loaded and therefore it should be possible to avoid or minimise sulfate transfer. This issue is investigated in Chapter 6, where the DEHPA solvent is applied to a bulk liquid membrane technique using hollow fibres. 3.5 Conclusions It has been shown that DEHPA is the most appropriate reagent for extraction of cerium from a sulfate solution for both a solvent extraction and a liquid membrane process. This conclusion has been reached from consideration of factors such as extraction ability, resistance to degradation, solvent selectivity and potential for sulfate transfer into a chloride strip solution. Commercial reagents of various types were considered. These include organophosphorous acids, phosphine oxide solvating extractants and an amine type reagent. The detailed study of the extraction of cerium(IV) with DEHPA defined the extraction reaction chemistry of the solvent system to be used in the following chapters. It was revealed that there are two distinct regions of cerium extraction. At acidities less than 5 M H 2S04, extraction occurs via a cation exchange mechanism. The formation of two complexes CeR4 and Ce(S04)R2(HR)2 are proposed, with the former being the dominant species at low acidities where the distribution coefficient is high. The presence of sulfur in the organic phase was confirmed experimentally, and it was found that the Ce:S ratio changed with the level of solvent loading. At acidities greater than 56 Chapter 3 Solvent selection and chemistry 5 M H2S04, extraction of cerium(IV) was found to increase with increasing acidity, which was explained in terms of a solvating mechanism. 3.6 Nomenclature Symbol D distribution coefficient {H} Hydrogen ion activity HA general term for mineral acid HX general term for organic acid HR, (HR)2 terms for monomeric and dimeric forms of DEHPA Kext stoichiometric constant K(p,q) stoichiometric constant defined by Equation 3. 7 in the text K(s,t) stoichiometric constant defined by Equation 3.16 in the text L general term for solvating reagent M'" metal cation p degree of polymerization of cerium(IV)-DEHPA complex as defined by Equation 3.7 in the text q number of undissociated DEHPA molecules in cerium(IV)-DEHPA complex as defined by Equation 3. 7 in the text R term for dissociated form of DEHPA ssres sum of squares function defined in Equation 3.19 in the text Greek letters a solvent loading as defined in Equation 3.15 in the text ~ overall stability constant Pt separation factor defined as D A/ D8Ionic strength cr standard deviation Subscripts 0 initial 57 Chapter3 Solvent selection and chemistry Superscripts over bar denotes species in the organic phase Acronyms and abbreviations Alamine 336 BEHEHP bis 2-ethylhexyl 2-ethylhexyl phosphonate Cyanex 272 bis-2,4,4-trimethylpentyl phosphinic acid Cyanex 923 a mixture of phosphine oxides with n-octyl and n-hexyl groups Cyanex 925 a mixture of phosphine oxides with n-octyl and 2,4,4- trimethylpentyl groups DBBP dibutylbutyl phosphonate DEHPA di-2-ethylhexyl phosphoric acid HEHEHP mono (2-ethylhexyl)2-ethylhexyl phosphonate, the main component of lonquest 801 sx solvent extraction TBP tributyl phosphate TOPO trioctyl phosphine oxide 3. 7 References 1. Ajawin I. A., Perez de Ortiz E.S. and Sawistowski H. 1983. Extraction of zinc by di(2-ethylhexyl) phosphoric acid. Chem. Eng. Res. Des 61: 62-66. 2. AI-Janabi M. A. A., Siyamanto K., Khachadoorian K., Abbas E.H. and Kadem A.H.M. 1990. Radiochemical separation of cerium from natural uranium irradiated with thermal neutrons. Journal of Radioanalytical and Nuclear Chemistry, Articles 141, no. 1: 61-67. 3. Alstad J. and Farbu L., inventors. 1977. "Process for separation of the lanthanides." Forskningsgruppe for Sjeldne Jordarter, assignee. United States Patent 4,041,125. 58 Chapter 3 Solvent selection and chemistry 4. Baroncelli F. and Grossi G. 1965. Chemical degradation of aromatic diluents exposed to nitric acid attack. International Conference Solvent Extraction Chemistry of Metals Ed. H. A. C. McKay, T. V. Healyy, I. L. Jenkins and A. Naylor. 5. Binghua Y., Nagaosa Y., Satake M., Nomura A. and Horita K. 1996. Solvent extraction of metal ions and separation of nickel(II) from other metal ions by organophosphorus acids. Solvent Extraction and Ion Exchange 14, no. 5: 849-70. 6. Bray L. A. and Partridge J.A. 1968. Extraction studies of Ce(III) into di(2- ethylhexyl) phosphoric acid and the observation of its oxidation in the presence of concentrated nitrate solutions to form a Ce(IV)-nitrate-DEHP complex. Fifth International Conference on Solvent Extraction ChemistryA. S. Kertes and Y. Marcus, 37-48. 7. Cotton F. A. and Wilkinson G. 1980 Fourth ed. New York: John Wiley & Sons. 8. Das 0. and Roy U. S. 1995. Extraction chromatographic studies of Ce(IV) with carboxylic acid (Versatic-10) and its separation from rare earth elements. Chem. Research Environ. 4, no. 3 & 4: 255-59. 9. Douglass D. A. and Bauer D. J. 1959. Liquid-liquid extraction of cerium, 5513. US Bureau of Mines. 10. EI-Dessouk M. M. and Aly H. F. 1977. Solvent extraction isolation of cerium from some radionuclides. 29, no. 3: 145-9. 11. El-Kot A. M. 1991. Separation of Metal Pairs using Bis(2- ethylhexyl)phosphoric acid (HDEDP); separation of cerium from uranium and from thorium. lsopenpraxis 27: 280-284. 12. EI-Yamani I. S. and EI-Messieh E. N. 1993. Refining of cerium from monazite by solvent extraction. Proceedings to the International Solvent Extraction Conference (ISEC'93). 59 Chapter3 Solvent selection and chemistry 13. Grover V. K. and Gupta Y. K. 1969. Kinetics and mechanism of the oxidation of benzilic acid by Ce(IV) in sulphuric acid solution. J. lnorg. Nucl. Chem. 31: 1403-16. 14. Hafner L., inventor. 1977. "Extraction of rare earths."Germany 2,633,115. (1986). 15. Hardwick T. J. and Robertson E. 1951. Canad. J. Chem. 29: 828. 16. Hart K. P. and Levins D. M. 1988. Management of wastes from the processing of rare earth minerals. Proceedings to Chemeca'BB. 17. Healy T. V. and McKay H. A. C. 1956. Complexes between tributyl phosphate and inorganic nitrates. Ree. Trav.Chim. 75: 730-736. 18. Hogfeldt E. 1982. Stability constants of metal ion complexes - Part A: Inorganic Ligands. IUPAC Chemical Data Series No 21 Pergamon Press, p169. 19. Horbez D. and Storck A. 1991. Coupling between electrolysis and liquid liquid extraction in an undivided electrochemical reactor: applied to the oxidation of cerium(3+) to cerium(4+) in an emulsion. Part I. Experimental. J. Appl. Electrochem. 21, no. 10: 915-21. 20. Horbez D. and Storck A. 1991. Coupling between electrolysis and liquid liquid extraction in an undivided electrochemical reactor applied to the oxidation of cerium(3+) to cerium(4+) in an emulsion. Part II. Cell modeling. J. Appl. Electrochem. 21, no. 10: 922-8. 21. Horbez D., Stork A. and Grosbois J., inventors. 1989. "Procedure of electrochemical oxidation."Australia 31139/89. 22. Huang T. C. and Juang R. S. 1986. Extraction equilibrium of zinc from sulphate media with bis(2-ethylhexyl)phosphoric acid. Ind. Eng. Chem. Fundam. 25: 752-57. 23. Electrochemical Series. 1980. CRC Handbook of Chemistry and Physiscs. 61st ed., J. F. Hunsberg, 0155-0160. 60 Chapter3 Solvent selection and chemistry 24. lbhadon A. 0. and Scott K. 1992. An application of supported liquid membranes in electrochemical processes. 127, no. Electrochem. Eng. Environ. 92: 163-75. 25. Juang R. S. and Lo R. H. 1993. Equilibrium studies of the extraction of zirconium(IV) from sulfuric acid solutions with di(2-ethylhexyl)phosphoric acid. Journal of Chemical Technology and Biotechnology 58, no. 3: 261- 69. 26. Kedari C. S., Pandit S. S. and Ramanujam A. 1997. In situ electro oxidation and liquid-liquid extraction of cerium(IV) from nitric acid medium using tributyl phosphate and 2-ethylhexyl hydrogen 2-ethylhexyl phosphonate. J. Radioanal. Nucl. Chem. 222, no. 1-2: 141-47. 27. Komasawa I., Otake T. and Higaki . 1981. Equilibrium studies of the extraction of divalent metals from nitrate media with di-(2ethylhexyl) phosphoric acid. J. lnorg. Nucl. Chem. 43, no. 12: 3351-56. 28. Kopunec R. and Kovalancik J. 1989. Separation of cerium and europium by extraction with tributyl phosphate and di(2-ethylhexyl)phosphoric acid in n-alkane from nitrate solutions. Journal of Radioanalytical and Nuclear Chemistry, Articles 129, no. 2: 295-303. 29. Korpak W., inventor. 1971. Poland 2,633,115. Chem. Abstracts, 1971, vol. 76, 26860z . 30. Korpusov G. V., Levin V.I., Brezhneva N.E., Prokhorova N.P., Eskevich I.V. and Seredenko P.M. 1962. The separation of cerium by extraction. Russian J. lnorg. Chem. 7, no. 9: 1167-71. 31. Li D., Wan Z., Zeng G. and Xue Z. 1984. Extraction mechanism of cerium(IV) from sulphuric acid solution by mono(2-ethylhexyl) ethylhexyl phosphonate. Journal of the Chinese Rare Earth Society 2, no. 2: 9-18. 32. Liddell K. C. and Bautista R.G. 1984. Hydrometallurgical Process Fundamentals: Nato Conf. Ser. 6Bautista R., 429. 61 Chapter3 Solvent selection and chemistry 33. Lu J., Wei Z. G., Li D. Q., Ma G. X. and Jiang Z. C. 1998. Recovery of Ce(IV) and Th(IV) from rare earths(III) with Cyanex 923. Hydrometallurgy 50, no. 1: 77-87. 34. Mirza M. Y. and Nwabue F. I. 1980. Studies on the extraction of titanium(IV), cerium(IV), thorium(IV) and uranium(VI) with 1-phenyl-3- methyl-4-benzoylpyrazole-5-one from different mineral acids. Separation of thorium from titanium, uranium and rare earths. 27, no. 1: 4 7-50. 35. Mitra K. and Ray U. S. 1992. Extraction chromatographic studies of cerium(IV) with carboxylic acid (n-capric acid) and its separation from rare earth elements. J. Indian Chem. Soc. 69, no. 9: 563-5. 36. Miyake Y., Matsuyama H., Nishida M., Nakai M., Nagase N. and Teramoto M. 1990. Kinetics and mechanism of metal extraction with acidic organophosphorus extractants (I): Extraction rate limited by diffusion process. Hydrometallurgy 23: 19-35. 37. Moore F. L. 1969. New liquid-liquid extraction method for the separation of cerium(IV) from berkelium(IV) and other elements. Analytical Chemistry 41, no. 12: 1658-61. 38. Mori Y., Ohya H., Ono H. and Eguchi W. 1988. Extraction equilibrium of Ce(III), Pr(III) and Nd(III) with acidic organophosphorus extractants. Journal of Chemical Engineering of Japan 21, no. 1: 86-91. 39. Nekovar P., Schroetterova D. and Mrnka M. 1997. Extraction of metal ions with a primary amine. J. Radioanal. Nucl. Chem. 223, no. 1-2: 17-22. 40. Peppard D. F. and Ferraro J.R. 1959. The preparation and infra-red absorption spectra of several complexes of bis-(2-ethylhexyl)-phosphoric acid. J. lnorg.Nucl.Chem. 10: 275. 41. Peppard D. F., Mason G.W. and Moline S.W. 1957. The use of dioctyl phosphoric acid extraction in the isolation of carrier-free 90Y, 140La, 144Ce, 143Pr and 144Pr. J. lnorg. Nucl. Chem. 5: 141-46. 62 Chapter3 Solvent selection and chemistry 42. Preston J. S., Cole P.M., du Preez A.C., Fox M.H. and Fleming A.M. 1996. The recovery of rare earth oxides from a phosphoric acid by-product. Part 2: The preparation of high-purity cerium dioxide and recovery of a heavy rare earth oxide concentrate. Hydrometallurgy 41: 21-44. 43. Preston J. S. and du Preez A.C. Solvent extraction processes for the separation of the rare earth metals. Solvent Extraction 1990 - ISEC'90T. Sekine, 883-94Elsevier. 44. Ray U. S. and Medak S.C. 1982. Extraction of cerium(IV) by liquid cation exchanger (Versatic 10) and its separation from lanthanide elements. J. Indian Chem. Soc. LIX: 392-94. 45. Raychaudhuri A. and Somlai J. 1992. Selective extraction of cerium(IV) on tri-n-butyl phosphate and phenoyltrifluoroacetone loaded poly-urethane foam. J. Radioanal. Nucl. Chem. Letters 166, no. 2: 153-66. 46. Rice N. M. 1981. Recommended nomenclature for solvent extraction (liquid-liquid distribution). Hydrometallurgy 7: 177-99. 47. Ritcey, G. M. and Ashbrook A.W. 1979. Solvent Extraction - Principles and applications to process metallurgy Part II. Elsevier. 48. Ritcey G. M. and Pouskouleli G. 1985. Nonnuclear hydrometallurgical applications. Science & Technology of Tributyl Phosphate-Volume II Selected technical and industrial uses., 71CRC Press, Florida. 49. Saleh F. A. 1966. Separation and purification of cerium from Egyptian monazite sands. Z. Anorg. Alig. Chem. 343: 205. 50. Sato T. 1989. Liquid-liquid extraction of rare earth elements from aqueous acid solutions by acid organophosphorus compounds. Hydrometallurgy 22: 121-40. 51. Sella C. and Bauer D. 1987. Solvent extraction of zirconium(IV) by organophosphorus compounds. Separation Processes in Hydrometallurgy., 207-14ed. G. A. Davies, Ellis Horwood . 63 Chapter3 Solvent selection and chemistry 52. Shakir K., Aziz M. and Beheir G. 1991. Extraction of certain actinides and lanthanide elements from different acidic media by polyurethane foams loaded with di-(2-ethylhexyl)phosphoric acid (HDEHP). Journal of Radioanalytical and Nuclear Chemistry, Articles 147, no. 2: 309-19. 53. Shuyun X., Yonghui Y., Yanzhao Y., Sixiu S. and Borong B. 1998. Extraction of sulfuric acid with TOPO. Journal of Radioanalytical and Nuclear Chemistry 229, no. 1-2: 161-63. 54. Sillen L. G. and Martell A.E. 1964. Stability constants of metal ion complexes - Special Publication No.17.The Chemical Society, London, p 237. 55. Sillen L. G. and Martell A.E. 1971. Stability constants of metal ion complexes - Supplement No.1, Special Publication No.25.The Chemical Society, London, p137. 56. Sole K. C. and Hiskey J. B. 1992. Solvent extraction characteristics of thiosubstituted organophosphinic acid extractants. Hydrometallurgy 30: 345-65. 57. Stamm H. H. 1964. Z. Anal. Chem. 200: 257. 58. Tavlarides L. L., Bae J.H. and Lee C.K. 1987. Solvent extraction, membranes and ion exchange in hydrometallurgical dilute metals separation. Sep. Science & Technol. 22: 581. 59. Tedesco P. H., Rumi V. B. and Gonzalez Quintana J. A. 1967. Extraction of tetravalent metals with di-(2-ethylhexyl) phosphoric acid-Part Ill-Cerium. J. lnorg. Nucl. Chem. 29: 2813-17. 60. Thornton, J. D. 1992. Science and Practice of Liquid-Liquid Extraction. Oxford: Clarendon Press. 61. Warf J. C. 1947. AECD-2524. Tech. lnf. Br. Oak Ridge. 64 Chapter3 Solvent selection and chemistry 62. Ying-Chu H., Tsang-Yang W., Yuh-Yuan W. and Tai-Ming C. 1987. Electro-reductive stripping of cerium in the TBP-HN03 two phase system. Hydrometallurgy 19: 209-25. 63. Ying-Chu H., Yuh-Yuan W. and Tai-Ming C. 1988. Electro-oxidation extraction of cerium in the TBP-HN03 two-phase system. Rare Earths, Extraction, Preparation and ApplicationsBautista R. G. and Wong M. M.The Minerals, Metals & Materials Society. 64. Zhang G., inventor. 1988. "Extracting rare earths elements from sulfuric acid."Chinese patent 86105043. 65. Zhang S. and Deng D. 1982. Electrolytic oxidation of Ce(III) and separation of Ce from Pm by extraction with 0.3 M HDEHP - 0.2 M TBP / kerosene-240. He Huaxue Yu Fangshe Huaxue (Journal of Nuclear and Radiochemistry) 4, no. 4: 243. 65 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane Summary The use of the flat sheet bulk liquid membrane technique for extracting cerium from a sulfate solution is explored in this chapter. Di-2-ethylhexyl phosphoric acid in n heptane is the solvent system investigated here. The effects of membrane type and solvent, feed and receiver solution composition are investigated. lnterfacial tension measurements are discussed and related to the permeation experiments. The cerium permeation experiments demonstrated that the greatest flux was obtained with a hydrophobic membrane at the feed-solvent interface and a hydrophilic membrane at the so/vent-receiver interface and that the overall mass transfer of cerium is controlled by the stripping process. The presence of the impurity MEHPA in the DEHPA solvent was found to have an accelerating effect on the permeation flux. This was attributed to its higher interfacial activity, and the fact that MEHPA is also an active carrier for cerium(IV). Both the interfacial activity and extraction capability of MEHPA were experimentally verified. Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.1 Introduction Liquid membranes have inherent advantages over solvent extraction processes such as lower solvent inventories and the possibility of achieving lower raffinates. There are various types of liquid membranes and their differences are discussed in the background section of this chapter. Out of all the liquid membrane techniques available, the bulk type liquid membrane with microporous membranes as phase separators, is thought to have the greatest potential for rapid development to an applied technology. There are two main reason for this. Firstly, the contacting equipment is simple and is already commercially available. Secondly, from a practical point of view, it seems to provide versatility and long term stability. This liquid membrane technique, in a flat sheet configuration was chosen for the detailed study of cerium transport with DEHPA as a carrier. In this chapter, basic variables in this process are investigated, such as the type of membrane at the interface and the effect of feed, solvent and receiver solution composition. The relationship between interfacial tension and the rate of transport is also explored. The terminology recommended by IUPAC for liquid membrane systems (Koros 1996), differs slightly from the solvent extraction terminology used in Chapter 3. For example, the extractant is termed the carrier, and stripping reactions are referred to as carrier complexation and decomplexation, respectively. In this work, for the sake of consistency, some terminology from solvent extraction has been retained. 4.2 Background This section reviews the different types of liquid membrane processes that are available, highlighting the advantages of each one. The role of interfacial tension in solvent extraction systems and the treatment of interfacial data are briefly discussed. 4.2. 1 Transport mechanisms in liquid membranes A liquid membrane is defined as a liquid film (usually a solvent), separating two miscible phases (aqueous phases in this discussion), through which ions or 66 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane molecules can pass (Boyadzhiev 1990). The driving force for the transport is the difference in chemical potential of the solute across the liquid membrane. The transport of a solute from feed to strip can be passive or facilitated. With passive transport, mass transfer occurs because of solubility differences. This type of transport cannot be used to concentrate solutions. Facilitated transport enables concentration of a solute from feed to strip and is more selective than passive transport. A carrier may be present in the solvent phase, which complexes with a solute at the feed side and decomplexes at the strip side. Facilitated transport is divided into four categories: i) Simple facilitated transport: No carrier is present. The transport of a solute is facilitated by a reaction in the strip phase, which forms a product that is not capable of diffusing back through the membrane into the feed phase. ii) Simple carrier mediated transport: As for simple facilitated transport, with the addition of a carrier that enhances the selectivity of the process. iii) Co-current coupled transport (depicted in Figure 4.1A): The carrier reacts with two species in the feed phase and carries both of them across to the strip phase. The coupling allows one of the species to be transported against its concentration gradient, provided the concentration gradient of the other species is sufficiently large. iv) Counter-current coupled transport (depicted in Figure 4.1 B): The carrier reacts with one species in the feed phase and with another in the strip phase, so that the coupled species are transported in opposite directions. In this thesis only facilitated counter-current coupled transport is studied. 4.2.2 Types of liquid membrane systems The most commonly studied liquid membrane systems are the emulsion and supported liquid membranes, which are briefly discussed below. The third category, broadly termed as bulk liquid membranes differs from the other two in that the membrane is not a thin film but consists of a bulk phase. 67 Chapter4 Transport of cerium with a flat sheet bulk liquid membranes A) Feed Solvent Receiver Figure 4.1A Schematic example of co-current coupled transport A"(metal anion), W(hydrogen ion), R3N (alkyl amine as carrier) B) Feed Solvent Receiver MX ~ "---.-/ HX H+-4--.L------.L-- Figure 4.1 B Schematic example of counter-current coupled transport M"(metal cation), ft(hydrogen ion), HX (organic acid as carrier) 68 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.2.2.1 Emulsion liquid membranes Emulsion liquid membrane technology (ELM), also known as the liquid surfactant membrane, is the most mature of the liquid membrane techniques. The process consists of three main steps: emulsification of the internal or strip phase into drops of solvent; contacting aqueous feed with emulsified strip phase (water in oil in water emulsion) and finally breaking the emulsion generally by electrostatic coalescence. A schematic diagram of the process is shown in Figure 4.2 (Butcher 1994). The recovery and concentration of copper (Li 1983), and uranium (Hartmann 1990) from dilute leach liquors has been widely studied, but the most promising application has been in the area of waste water treatment. Pilot plant studies for the separation of Zn in the viscose and rayon industry, heavy metals from residues of incineration plants and Ni and Cr from electroplating solutions have proved the viability of this technique (Marr 1989). The main advantage of any liquid membrane over conventional solvent extraction, is that it combines the extraction and stripping operations into a single step, thus removing the limitations posed by equilibrium. In emulsion liquid membranes, this allows for very low concentrations in raffinate solutions, even down to the parts per billion range. Other advantages include high selectivity, lower solvent inventory, low solvent losses and high interfacial area. Loss of membrane phase due to solubility in the feed has been reported to be of the order of 1-5 ppm, and interfacial area to strip phase ratio as high as 106 m2/m 3 (Draxler 1990). The two main disadvantages of this technique are emulsion swelling with prolonged contact and membrane leakage to feed due to membrane rupture (Raghuraman 1993). The swelling is attributed to osmotic differences between the feed and strip phases and to entrainment of feed to strip. These effects result in reduced stripping efficiency and decreased selectivity. The leakage of membrane fluid can be minimised by increasing the stability of the emulsion, but this in turn makes the demulsification step more difficult. Finally, design of a full scale plant from pilot plant data is not straight forward when very low effluent concentrations are required. Mass transfer is sensitive to many parameters such as hold-up and interfacial area, which cannot always be very accurately predicted. 69 Chapter4 Transport of cerium with a nat sheet bulk liquid membranes Aqueous Feed Emulsion .,,.,, ~ Receiver phase ~ Membrane phase Contactor -1 mm (10 ~lm thick) Emulsification Solvent Concentrated aqueous product Purified aqueous phase (raffinate) Figure 4.2 Emulsion liquid membrane process (schematic diagram reproduced from Butcher 1994) Feed Receiver phase phase Impregnated membrane Membrane support Figure 4.3 Supported liquid membrane technique 70 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.2.2.2 Supported liquid membranes In the supported liquid membrane process (SLM), the solvent phase is impregnated in a solid support such· as a microporous membrane. The feed and strip solutions are on either side of the liquid membrane. Such an arrangement can be studied in a flat sheet configuration as shown in Figure 4.3. It can also be adapted to continuous operation by the use of spiral wound (Teramoto 1987) or hollow fibre modules (Valenzuela 1999) This technique has been very widely investigated and laboratory bench studies have been undertaken for base metals Cu, Co, Ni and Zn, the platinum group and precious metals (Bromberg 1992), (Hidalgo 1991 ), (Ashrafizadeh 1996), uranium (Babcock 1980) and the rare earth metals (Danesi 1986), (Kondo 1995). A single study on the transport of cerium(IV) has also been published (Chaudry 1996). The carrier used was tri-n-octylamine in an aromatic diluent 1• The inherent advantage of the supported liquid membrane technique, over both emulsion liquid membranes and conventional solvent extraction is its simplicity of operation involving no stirring or phase separation equipment. The major drawback is the lack of long-term stability of the liquid membrane, due to loss of membrane fluid to the surrounding aqueous phases. The causes for this loss are complicated (Wijers 1996), (Zha 1993). Pressure differences over the membrane, progressive wetting by the aqueous phases of the pores in the microporous support, mutual solubility of the aqueous and membrane liquids and emulsion formation of the liquid membrane in the aqueous phases have been proposed as the major mechanisms for membrane loss. Solutions put forward to address this problem include the addition of membrane liquid to the strip phase, continuous re-impregnation of the membrane support and the use of gelled liquid membranes, all of which suffer from practical drawbacks and or decreased permeability. Recently the stability of a SLM used for copper 1 The authors observed that reduction of cerium(IV) in the sulfuric acid feed occurred to such an extent that no cerium(IV) transport was observed. They attributed this to the transport of ascorbic and mandelic acid from the stripping phase to the feed, and the subsequent reduction of cerium(IV) to cerium(III) which is not extracted. Work presented in this thesis shows that cerium(IV) is reduced in the presence of an amine type carrier in an aliphatic diluent. The presence of an aromatic diluent would exacerbate the reduction of cerium(IV). 71 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane transport has been remarkably increased by the application of a hydrophilic polymeric layer to the SLM. No detrimental effect on the permeability and selectivity was found. This technical breakthrough maybe what is required to finally turn the SLM technology into a commercial reality (Wijers 1996). 4.2.2.3 Bulk liquid membranes Bulk liquid membranes consisting of three bulk phases, such as shown in Figure 4.4A, are most often used for laboratory studies to determine the selective properties of a carrier (Zhouri 1995), (El Bachiri 1996). Such a technique has also been used to study the transport kinetics of yttrium across a liquid membrane (Chitra 1997). Rotating film pertraction (Boyadzhiev 1994) is an example of continuous operation of a bulk liquid membrane. The interfacial area to volume ratio in these types of systems is very low, resulting in low fluxes. This configuration of bulk liquid membrane therefore is not likely to compete favourably with other separation processes. The use of membranes as phase separators between the bulk solvent and the two aqueous phases is attracting increasing attention. There seems to be no consensus on the terminology used for this technique, perhaps because there are many variations. The common factor is that the feed, solvent and receiver solutions are arranged in an alternating fashion. The membranes can be flat sheets, spiral wound or hollow fibres. Hollow fibres are favoured for their very high surface area to module volume ratios. In a flat sheet configuration, (Boyadzhiev 1988) have used 4 mm thick foamed viscose as phase separators and called the technique liquid film pertraction (Figure 4.48). (Teramoto 1989) used a similar arrangement with microporous polypropylene and teflon membranes described as flowing liquid membranes. They also applied this technique to spiral wound modules (Teramoto 1989). Sirkar and co-workers (Sengupta 1988), (Majumdar 1992) have developed a technique called hollow fibre contained liquid membrane (HFCLM), in which the bulk solvent phase is stationary on the shell side of a hollow fibre module (Figure 4.4C). The feed and receiver solutions are fed through the tube side of two separate sets of fibres. The microporous membranes can be hydrophobic or hydrophilic. An important aspect of this technology is to balance the pressures in such a way as to immobilise the interfaces at the membrane surfaces and prevent leakages of one phase to another. 72 Chapter4 Transport of cerium with a flat sheet bulk liquid membranes Membrane phase / Feed phase Receiver phase A) Bulk liquid membrane Feed Phase ~ ~ = Membrane Support Membrane Phase Receiver Phase B) Liquid film pertraction Cross section of hollow fibre module Solvent phase Membrane support (hollow fibre) Receiver~....\,-~ phase Feed phase C) Hollow fibre contained liquid membranes Figure 4.4 Types of bulk liquid membrane techniques. 73 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane An alternative to the HFCLM configuration is to have the extraction and stripping operations in separate hollow fibre modules, with the solvent circulating in between. When these contactors are operated in such a way that equilibrium is reached between aqueous and solvent solutions, the process ceases to be a liquid membrane process and essentially becomes a solvent extraction process. Two module extraction and stripping has been referred to as membrane based solvent extraction (Kim 1984), membrane assisted solvent extraction (Hutter 1994), non-dispersive solvent extraction (Ortiz 1996) or simply membrane extraction. The advantage of using two modules instead of one is that the flexibility of the system is much improved and pressure control is easier to maintain. The hollow fibre modules are more readily available in comparison to modules used for HFCLM which are complicated to manufacture. An interesting alternative to using microporous membranes as phase separation has been suggested by (Kedem 1993). They have used ion exchange membranes, in a flat sheet bulk liquid membrane configuration to separate the feed and strip phases. These membranes prevent the loss of the bulk solvent phase to the surrounding aqueous solutions, and do not seem to have a detrimental effect on the overall flux. The greatest advantage of the bulk liquid membrane technique is that in principle it solves the problem of solvent loss experienced by SLM. The solvent inventory is expected to be higher than SLM but much lower than required for solvent extraction due to the absence of settler areas. In a two hollow fibre module configuration, the system is very versatile in that it can be tailored to different rates of transport often found between feed to solvent and solvent to receiver solution. Both membrane type and surface area can be changed to suit. The disadvantage is that the overall resistance is greater than that of an ELM or a SLM system due to the presence of the membrane phase separators and the organic bulk phase. In HFCLM there is also the problem caused by the difficulty in mixing the two sets of fibres to achieve a low contained liquid membrane thickness. 74 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.2.3 The role of interfacial tension In the study of liquid membrane processes, the bulk properties of the liquids are usually considered, because information on interfacial properties and concentrations is very difficult to obtain in practice. An exception to this is the determination of interfacial tension which can be carried out with a number of standard techniques including the Du Nouy ring, the Wilhelmy plate, the Paddy rod and the drop volume method (Weissberg 1949). A large number of extractants used for the solvent extraction and liquid membrane extraction of metal ions are interfacially active. These include hydroxyoxime type reagents used for copper extraction (Szymanowski 1990), organophosphorous (Qiang 1989) and amine reagents (Lee 1995). The interfacial activity of the reagents can sometimes be used to explain equilibria and kinetic data. lnterfacial tension measurements of values of r as a function of extractant concentration can be described by the Gibbs adsorption Equation 4.1 (Vandegrift 1980): r=-1- ar 4.1 RT a(Ina) where r is the surface excess of extractant at the interface or the interfacial extractant population, r is the interfacial tension and a is the activity of extractant in the bulk phase, R is the gas constant and T the absolute temperature. In dilute solutions the activity can be substituted by the molar concentration. From plots of interfacial tension versus the natural logarithm of the bulk concentration two parameters Cmin and rmax can be calculated. Cmin is the minimum bulk extractant concentration necessary to saturate the interface and r max is the maximum extractant interfacial concentration that can be obtained in the system. Due to high differentiation errors, only the surface excess at saturation can be calculated using the Gibb adsorption equation. The interfacial data can also be matched to various adsorption equations which are then introduced into the Gibbs equation and solved analytically. The Szyszkowski isotherm, shown in Equation 4.2, is one type of mathematical expression which has been used to describe interfacial data (Szymanowski 1994). 75 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane C y = y0 [I-Bln(-+ 1)] 4.2 A where B and A are constants, Yo is the interfacial tension of the two phases in the absence of extractant, and C is the bulk concentration of extractant. The constants A and B have a physicochemical meaning and can be used to calculate the surface excess at the saturated interface and the free energy of adsorption as shown in Equations 4.3 and 4.4. r = Bro 4.3 max RT !1G 0 d = RTlnA 4.4 Incorporating the Szyszkowski equation into the Gibbs equation and differentiating, the surface excess can be calculated as a function of extractant concentration as shown in Equation 4.5 (Abrantes L.M. 1996). r-Bro(-5._) 4.5 RT C+A 4.3 Results and discussion The use of the flat sheet bulk liquid membrane technique for extracting cerium from a sulfate solution is explored in this chapter. A batch permeation cell was used for this purpose because of the ease with which membranes can be substituted at either the feed/solvent or solvent/receiver interface. The effects of membrane type and solvent, feed and receiver solution composition are investigated. lnterfacial tension measurements are discussed and related to the permeation experiments. The solvent system investigated here is the di-2- ethylhexyl phosphoric acid - heptane system identified in Chapter 3 as being the most appropriate for cerium recovery. Details of the experimental procedures are given in Chapter 2, Section 2.5. 76 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.3.1 Membrane characterisation and selection Flat sheet microporous membranes were used for this work. The manufacturer's specifications are shown in Table 4.1. The choice of membranes was limited because of the chemical compatibility requirement for the membranes with respect to acidic solutions (5.5 M H2S04 and 2 M HCI), strong oxidizing conditions due to the presence of cerium(IV) and to the organic solvent n-heptane. Under these conditions common materials such as cellulose acetate, nylon and polycarbonate are not suitable. TABLE4.1 Characteristics of flat-sheet membranes Pore Porosity Thickness Name Supplier Material size % µm µm HYDROPHOBIC Durapore GVHP Millipore 1PVDF 0.22 75 125 Gelman FP-200TM Gelman 1PVDF 0.20 45-50 HYDROPHILIC Durapore GVWP Millipore 1PVDF 0.22 70 125 Durapore WLP Millipore 1PVDF 0.10 70 125 GHPolypro Gelman 2pp 0.20 1 PVDF = Polyvinylidene fluoride 2pp = Polypropylene Membrane thickness The thickness of the flat sheet membranes were measured with a micrometer, by stacking up two and three filters and averaging the readings. Measurements before and after 24 hour contact with heptane, in the case of hydrophobic membranes, and 5 M sulfuric acid or 2 M hydrochloric acid in the case of hydrophilic membranes were taken. The results are presented in Table 4.2. The results show that the hydrophobic membranes are not significantly affected by contact with heptane. Some degree of swelling was measured (< 2%) for the two Millipore membranes but this was within the error associated with the readings. The hydrophilic membranes shrunk when in contact with 5 molar 77 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane sulfuric acid. This effect was more noticeable with the GHPolypro membrane (15% reduction) made up of polypropylene than on the polyvinylidene fluoride membranes (4-11 % reduction). Contact with hydrochloric acid also caused a reduction in the measure thickness of the hydrophilic membranes, but to a lesser degree. The effect of sulfuric acid on the three hydrophilic membranes is illustrated by the scanning electron microscopy pictures shown in Figure 4.5. It appears that the surface of the GVWP and WLP membranes has been eaten away, while the acid has caused a closing-up of pores in the GHPolypro membrane. TABLE 4.2 Measured thickness of membranes (1,1m) Name Specifications in contact with as received n-heptane 5 M H2S04 2M HCI HYDROPHOBIC Durapore GVHP 125 113 114 Fluoropore FGLP 175 149 151 Gelman FP-200TM 154 151 HYDROPHILIC Durapore GVWP 125 111 106 109 Durapore WLP 125 104 92 96 GHPolypro 99 84 91 Pore size distribution The pore size distributions of selected membranes were measured with a Coulter Parameter. Results obtained are shown in Figure 4.6 and Table 4.3. The mean pore size measured by this method was in agreement with the manufacturer's specification in the case of the GHPolypro membrane. However, with the other three membranes, the mean pore size measured was much higher than specified. The narrowest pore size distribution was obtained for the Gelman GHPolypro membrane. 78 Chapter4 Transport of cerium with a flat sheet bulk liquid membr, Durapore GVWP Durapore VVLP Figure 4.5 SEM pictures of flat sheet hydrophilic membranes. A) Prior to contact with solution B) After 24 hours of contact with 5 M sulfuric acid Chapter4 Transport of cerium with flat sheet bulk liquid membranes -A--- Gelman GH-Polypro -e-- Durapore WLP -o-- Gelman FP-200TM ---+- Durapore GWvP n 5.00 T I 4.50 I 4.00 .-~ ii 3.50 ~ C ! 3.00 ! 2.50 i5... 2.00 I!? u::: 1.50 1.00 1\ b \ ' 0.50 I ------ Figure 4.6 Pore size distribution of various membranes ---..------l.... Diffusion boundary layers Receiver Solvent Feed Phase uI I Phase uI I Phase I I (HR)2 Ce4+ !Qi: _[J: I I I I H+ I I I Ce4+ Ce4+,..::n: : ~LI~ ~ Microporous membrane as phase separators Figure 4.7 Counter-current facilitated transport of cerium(IV) (HR) 2 (dimeric di-2-ethylhexyl phosphoric acid as carrier) 80 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane TABLE4.3 Measured pore size (1,,1m) Name Specifications Measured mean minimum maximum Durapore GVWP 0.22 0.40 0.32 0.63 Durapore WLP 0.1 0.30 0.24 0.46 GHPolypro 0.2 0.19 0.16 0.27 FP-200TM 0.2 0.35 0.26 0.60 4.3.2 Permeation experiments The mass transport of cerium was measured in a three compartment permeation cell described in Chapter 2, Section 2.5.2. The three phases, aqueous feed, solvent and aqueous receiver solutions were physically separated by two flat sheet microporous membranes. The solvent used was a solution of di-2-ethylhexyl phosphoric acid (DEHPA) in n-heptane. Both hydrophobic and hydrophilic membranes were tested on the feed/solvent and solvent/receiver interfaces. Unless stated otherwise the feed was a solution of cerium(IV) sulfate in 0.55 M H2S04 and the receiver solution was 5.5 M H2S04• These conditions were chosen to maximize extraction of cerium in the feed and stripping to the receiver solution, according to the distribution coefficients as presented in Chapter 3. The concentrations of cerium in the feed and receiver solutions were monitored continuously with a UV-VIS spectrophotometer. During the permeation experiment the cerium is transported from the feed to the receiver solution (Figure 4.7). There is also a transport of hydrogen ions in the reverse direction. Implicit in the extraction and stripping reactions (see Chapter 3, Equation 3.7), is that for every mole of cerium transported from feed to receiver solution, four moles of hydrogen ions are transported from the receiver to the feed. These changes in hydrogen ion concentrations were not measured because they are very difficult to detect, due to the high acid concentrations in both feed and receiver solutions relative to the amount of hydrogen ions being transported. An example of the cerium concentration profiles obtained is shown in Figure 4.8. The concentration profiles appear to be continuous lines because a reading was taken every two minutes and many data points were collected. In the initial stages of the experiment, the decrease 81 Chapter4 Transport of cerium with a flat sheet bulk liquid membranes ,-- 200 Feed 150 Receiving E 0. 0. -';" 0.... 100 0 ~ ~ ~ ~ 1~1~1~1~ Minutes Figure 4.8 Cerium(IV) concentration profiles in permeation experiment [DEHPA]=0.2 M, Feed: Durapore GVHP, [Ce]o=1.4 mM, 0.55 M H2S04, Receiver: Durapore GVWP, 5.5 M H2S04 82 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane in cerium concentration in the feed solution is accompanied by a corresponding increase in the solvent concentration. There is a lag time before the cerium is transferred from the solvent to the receiver solution due to membrane diffusion at the solvent/receiver interface and the uptake of cerium by the solvent. The receiver solution concentration then increases with time. In this particular experiment, after approximately 10 hours 50% of the cerium has been transported from the feed to the receiver solution. After this point the cerium is being transported "uphill" against its concentration gradient. Through the first part of the cerium feed curve, a straight line can be drawn. Similarly for the cerium receiver curve, ignoring the initial lag time. The overall flux JR and the flux of cerium transfer from feed to solvent JF was determined from the slope of the straight lines according to Equations 4.6 and 4.7. J _ VR a[Ce]R 4.6 Receiver - A at R J - VF a[Ce]F 4.7 Feed - A at F The fluxes JR and JF are not the same due to the large volume of solvent, which has the capacity to store the total amount of cerium in the system. The counter-current transport of cerium from feed to receiver solution in the flat sheet bulk liquid membrane permeation cell consists of eight steps: i) Diffusion of cerium(IV) from bulk to the feed/solvent interface through aqueous boundary layer; ii) Complexation of cerium(IV) with carrier at the feed/solvent interface; iii) In the case of hydrophobic membrane in the feed side, diffusion of extracted complex through membrane pore; iv) Diffusion of extracted complex through solvent boundary layer at the feed side; v) Diffusion of extracted complex through solvent boundary layer at the receiver side; vi) Decomplexation of extracted complex at the solvent/receiver interface; 83 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane vii) In the case of hydrophilic membrane in the receiver side, diffusion of cerium(IV) through membrane pore; viii) Diffusion of striped cerium(IV) through the receiver aqueous boundary layer The permeation experiments were run under various conditions to determine the relative importance of the various steps. 4.3.2. 1 Effect of stirrer speed The degree of stirring of the three bulk phases has an impact on the thickness of the boundary layers at the two interfaces. As mentioned above, four of the eight resistances in a flat sheet bulk liquid membrane are due to boundary layer diffusion, and therefore the bulk stirring speed will affect the overall cerium transport. The effect of feed stirrer speed on the cerium flux from feed to solvent (JF) is shown in Figure 4.9. For these experiments the third compartment in the permeation cell was blocked, and only one membrane was used. The cerium flux increases with increasing stirrer speed reaching a plateau at approximately 600 rpm. This means that at low stirrer speed, the resistance due to diffusion through the aqueous feed boundary layer is significant and affects the overall transport from feed to solvent. As the stirrer speed increases, the feed boundary layer thickness 88 decreases to a minimum value, and other resistances become more important. The stirrer speed of the solvent phase did not have a marked effect on the cerium flux at rpm values between 385-600 rpm indicating that diffusion through the boundary layer, 80 , at least for the extraction reaction is not significant. A detailed investigation was not done on the effects of stirrer speed of the receiver phase. The influence of receiver phase boundary layer thickness is generally not as pronounced as that of the feed, because the concentration of hydrogen ions (for counter ion transport) in the receiver phase, is much higher than that of cerium ions in the feed (Babcock 1980). It was therefore assumed that 600 rpm would provide sufficient agitation to minimise the receiver phase boundary layer. In order to minimise all boundary layer thickness, all further experiments were undertaken at a stirring speed of 600 rpm for all three phases. 84 Chapter4 Transport of cerium with flat sheet bulk liquid membranes 9.E-06 - 8.E-06 ...---:.- • ';" 7.E-06 U) ~ E 6.E-06 0 E ->< 5.E-06 :::s u. 4.E-06 3.E-06 0 200 400 600 800 1000 Aqueous rpm Figure 4.9 Effect of aqueous stirrer speed on the cerium(IV) flux from feed to solvent. Solvent stirrer speed = 385 rpm. [DEHPA] =0.2 M, Feed: 0.55 M H2S04, [Ce]o = 1.4 mM + GH-Polypro O VVLP tt. GVWP .GVHP I 50 40 E ~ 30 cDD -a) cDD cDD ~ 20 cDD cDD ccD I cDD 10 cc00 i I 0 0 20 40 60 80 100 120 140 160 180 200 2=-2240 Minutes '------ Figure 4.1 O Effect of type of membrane at the solvent/receiver interface on cerium concentration in the receiver solution Membrane at the feed/solvent interface: hydrophobic GVHP 85 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.3.2.2 Effect of membrane type The hydrophibicity of the membranes placed at the interfaces of a three phase system is expected to have a significant effect on the overall flux. Four configurations of either hydrophobic or hydrophilic membrane at the feed/solvent and solvent/receiver interface are possible. 4.3.2.2.1 Feed-solvent interface Two membranes were tested at the feed/solvent interface. Results are presented in Table 4.4. The cerium flux JF measured from feed to solvent was 3.7 times faster for the hydrophobic GVHP membrane, compared to the hydrophilic WLP membrane. TABLE4.4 The effect of membrane type on JFeed Membrane at feed/solvent interface JFeed(mol m-2s- 1) Hydrophobic - GVHP 7.38 X 10-6 Hydrophilic - WLP 1.97 X 10-6 Conditions: [DEHPA] = 0.2 M, [MEHPA] = 0.01 M, [Ce]0 = 0.014 M, 5.5 M H2S04 in receiver These results can be explained by analogy to a membrane based solvent extraction system. Although these permeation experiments were undertaken with simultaneous stripping, the solvent volume is sufficiently large so that the initial stages of the permeation experiment, (where the measurement for JF is taken), is essentially independent of the stripping reaction. Overall mass transfer coefficients for flat sheet membrane based extraction without chemical reaction have been described by (Prasad 1992) as•: • For a more detailed discussion on membrane based extraction see Chapter 5, Sections 5.3.2 and 5.3.5 86 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane extraction with hydrophobic membrane 4.8 where kd is the distribution coefficient, k8 , k0 and km are the mass transfer coefficients for the aqueous and organic diffusion layers and membrane respectively. Ka is the overall mass transfer coefficient based on the aqueous phase. extraction with hydrophilic membrane 4.9 Under extraction conditions, where kd >> 1, the terms containing kd in the denominator will be relatively small. If these terms are ignored it becomes apparent that the overall resistance to mass transfer is greater with hydrophilic membranes than with hydrophobic. Since the resistance due to chemical reaction at the feed/solvent interface is independent of membrane type, it follows that similar conclusions can be drawn for mass transfer with chemical reaction, where the kd >> 1. In the case of cerium(IV) transfer from 0.55M H2S04 to a solvent phase containing DEHPA, the distribution coefficient is much greater than 1 and therefore the flux measured with the hydrophobic membrane is greater than that measured with a hydrophilic membrane. 4. 3. 2. 2. 2 Solvent - receiver interface To study the effect of membrane type on the overall mass transfer JR, a hydrophobic GVHP membrane was used at the feed/solvent interface. Permeation experiments were then undertaken with four different membranes at the solvent/receiver interface. Results are presented in Table 4.5 and Figure 4.10. The results show that: • The overall flux is approximately three times slower with the hydrophobic membrane compared with the three hydrophilic membranes tested. Again by analogy to membrane based stripping, overall mass transfer with no chemical reaction can be described by Equations 4.10 and 4.11 (Prasad 1992). The terms containing kd in the nominator are small because kd <<1 for effective stripping and therefore the overall mass transfer resistance will be higher for hydrophobic membrane due to an extra term. 87 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane stripping with hydrophobic membrane 4.10 stripping with hydrophilic membrane 4.11 TABLE4.5 The effect of membrane type on JR Membrane at solvenUreceiver interface JReceiver(mo/ m-2s-1) Hydrophobic - GVHP 8.36 X 10-7 Hydrophilic - GVWP 2.56 X 10-6 Hydrophilic - WLP 2.13 X 10-6 Hydrophilic - GH-Polypro 2.57 X 10-6 Conditions: [DEHPA] = 0.19 M, [MEHPA] = 0.01 M, [Ce]0 = 0.014 M, 5.5 M H 2S04 in receiver, Hydrophobic - GVHP at the feed/solvent interface • The three hydrophilic membranes tested differ in mean pore size and pore size distribution (Table 4.1 ). Another major difference, as shown by the SEM picture in Figure 4.5, is that the GHPolypro membrane has a lower porosity and well defined pores, leading to lower tortuosity. Minor differences in membrane thickness between the three membranes were also measured (Table 4.2). However, no major differences were found between the fluxes of these three membranes, under similar permeation conditions. Differences in membrane porosity, thickness and turtosity effect the value of km. Since, the term containing km in Equation 4.11, is not the most significant one, it follows that the nature of the hydrophilic membrane will not effect the overall flux as was observed. However, the same properties in a hydrophilic membrane, if used at the feed/solvent interface, will have a major effect since in that case the term km is significant. Conversely, the nature of the hydrophobic membrane at the solvenUreceiver interface will also have a major effect on the overall flux. • It was observed that the lag time was much longer with the hydrophobic membrane (200 minutes) and the same for all three hydrophilic membranes (40 minutes). The lag time is defined here as the time between the start of the permeation experiment and the start of the stripping process. The longer lag 88 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane time for the hydrophobic membrane is due to the need for the cerium-DEHPA complex to diffuse through the membrane pores before it is available for stripping. 4.3.2.2.3 Leakage of one phase into another In a two phase system containing a microporous membrane at the interface, the pores of the membrane are impregnated with aqueous solution if the membrane is hydrophilic and solvent solution if the membrane is hydrophobic (Figure 4.11 ). Therefore, a hydrophobic membrane will tend to leak solvent whilst a hydrophilic membrane will tend to leak aqueous solution. This effect was observed by (Kim 1984), who used two cellulose acetate hollow fibre modules to concentrate a copper sulfate solution. He found that there was an accumulation of aqueous droplets in the solvent recirculating line. To prevent the leakages it is necessary to keep the pressure of the non-impregnating phase higher than the pressure of the impregnating phase, but below a certain critical pressure. Above the critical pressure the non-impregnating phase will force its way through the pores of the membrane. The types of leaks observed during the transport of cerium(IV) in the three compartment cell operated under different cell configurations are detailed in Table 4.6. The source of the aqueous leak was identified by measuring the acidity of the leak. TABLE4.6 Expected direction of leakage in flat sheet bulk liquid membrane permeation cell Membrane Type at the interface Expected direction of leakage Observed leakage Feed/Solvent Solvent/Receiver Hydrophobic Hydrophobic Solvent to feed and receiver No leaks Hydrophobic Hydrophilic Solvent to feed and receiver to Receiver to solvent solvent Hydrophilic Hydrophilic Feed and receiver to solvent Receiver to solvent 89 Chapter4 Transport of cerium with flat sheet bulk liquid membranes Hydrophobic Hydrophilic Membrane Membrane - ',,. I I ' . ,,,.,--' r \,...'-'-- - r \,_I' 1/ _: // r - I \ ', /,./, . //'// , . ": I , r \ Aqueous' .- // Solvent,., I . • .• non-impregnating ' 1mpregnat1ng phases -phase-,,.·, , - I \ / / / ,,,.,--\.,...''' r -- r ,_ r \ I' ,-, :: , - I \ l' ~ I ., / - ~ \ ,, / Figure 4.11 Wetting properties of hydrophobic and hydrophilic microporous membranes r e Purified DEHPA J I 1,::·04 §~~~~~~~~~~~~ i ~U) ------~------ - 1 · E-o:0"""00-0--0-.0..,.o-so--o.-,01-o-o-[-M-~-;P,..1:-~-,M-0-.0..,.2-00 ___ -. -~o.~02-s~o-_-_ -~""""!.0300 ! [__ ------~ Figure 4.12 Effect of MEHPA concentration on the flux of cerium(IV) from feed to solvent. [DEHPA] =0.2 M, Feed: 0.55 M H2S04, [Ce]o = 1.4 mM, Membrane GVHP 90 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane No leaks of solvent to feed or receiver were observed with the hydrophobic Durapore GVHP membrane. This is attributed to the slightly lower pressure in the solvent compartment compared to the feed and receiver compartments, due to the lower density of the solvent. The leak from receiver to solvent was observed for the three different hydrophilic membranes. In each case aqueous solution was accumulated in the solvent phase. The leak did not occur immediately, but coincided with the start of the back extraction process. Over a twenty hour period, the size of the leak was of the order of 2.6% of receiver solution volume lost to solvent, which is quite significant. The effect of the leakage on the measured overall permeation flux is expected to be less than 1%, since generally only the readings taken over the first sixty minutes of stripping were used for flux determination. These observation highlight the need for pressure control, which was lacking in the batch permeation cell. This issue is addressed in the application of the bulk liquid membrane technology to hollow fibre modules, presented in Chapter 6 of this thesis. 4.3.2.3 Effect of mono-2ethyl hexyl phosphoric acid Mono 2-ethylhexyl phosphoric acid (MEHPA) is a by-product of the manufacturing process of the extractant DEHPA. The AR grade reagent supplied by Merck used for this work contained approximately 1% of the mono ester, although one batch was higher at 4%. Higher fluxes were measured for the batch containing the greater concentration of MEHPA. Considering that MEHPA is an acidic extractant itself, it was decided to investigate the effect of MEHPA on the permeation of cerium. A range of permeation experiments were undertaken at constant feed and receiver solution composition and DEHPA solvent concentration. The concentration of MEHPA was changed from 0.002 to 0.027 M whilst DEHPA concentration was kept at 0.2 M. This represents a change in molar ratio of MEHPA to total MEHPA + DEHPA of 1 to 13.3 %. As a baseline point for DEHPA with no MEHPA impurity, AR grade DEHPA reagent was purified by precipitation of the copper-DEHPA complex as detailed in Chapter 2, Section 2.3.5. The MEHPA concentration in the purified sample was determined by potentiometric titration as 0.9%. Results for fluxes measured from feed to solvent are presented in Figure 4.12. It was found that the 91 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane concentration of MEHPA had a significant effect, increasing the flux of cerium from feed to solvent. The rate of extraction for the purified DEHPA reagent was found to be approximately the same as that measured for the 1% MEHPA AR grade Merck reagent, indicating that the purification procedure is unnecessary for this grade of reagent. The increased rate of extraction was also reflected in the overall flux measured as shown in Figure 4.13. Increased MEHPA concentration had a positive effect on the overall flux up to a concentration of 0.01 M (5% MEHPA). A further increase in MEHPA was found to decrease the overall flux. MEHPA itself is capable of extracting cerium(IV) and is in effect a stronger extractant than DEHPA, as is exemplified in Figure 4.14, where the extraction of cerium(IV) at 0.55M H2S04 with a mixture of MEHPA (0.11 M) /DEHPA (0.08 M) is higher than that of DEHPA (0.08 M) alone. Furthermore, whilst extraction of cerium(IV) decreases with increased acidity in the range (0.5 - 3 M H2S04), extraction of cerium(IV) remains high throughout this acid range when a significant amount of MEHPA is present. This means that higher concentrations of acid are required to strip cerium from MEHPA/DEHPA mixtures, with relatively high MEHPA concentrations. When MEHPA concentrations are low with respect to DEHPA, up to 5% molar ratio, the extraction of cerium(IV) by MEHPA does not significantly affect the distribution coefficient kd. However, as this ratio increases, back extraction becomes less favourable at the fixed concentration of 5.5 M sulfuric, and this is the reason for the drop in JR shown in Figure 4.13. The increased rates of extraction translating into higher measured fluxes JF, however, cannot be explained by a significant change in the distribution coefficient, but are thought to be related to the interfacial activity of MEHPA. This issue is explored further in Section 4.4.3. Figure 4.13 also illustrates the influence of the type of membrane used on the solvent/receiver interface. It is evident that regardless of the accelerating effect of MEHPA, hydrophobic membranes used at the back extraction interface reduce the overall flux, confirming results presented earlier in Section 4.4.2.2. It is also evident from these results that the mass transfer from solvent to receiver solution is slower than from feed to solvent. 92 Chapter4 Transport of cerium with flat sheet bulk liquid membranes o JF 1:,,. JR (Hydrophilic) + JR (Hydrophobic) • HCI as stripping agent 1.E-04 -"'• "e 1 ..,o: 1.E-05 all -,... - H- - -~------~------1• 1.E-06 - a--· - \1, 1.E-07 1------...... ------~ 0.001 0.01 0.1 [MEHPA] /M Figure 4.13 Effect of MEHPA concentration on the flux of cerium(IV) from feed to solvent (JF) and from feed to receiver (JR) [DEHPA] =0.2 M, Feed: 0.55 M H2504, [Ce]0 =1.4 mM, Receiver: 5.5 M H2S04 Legend: JF: hydrophobic GVHP at feed/solvent interface JR (hydrophilic): hydrophobic GVHP at feed/solvent interface, hydrophilic at solvent/receiver interface JR (hydrophobic): hydrophobic at feed/solvent and solvent/receiver interface HCL as stripping agent: same conditions as JR (hydrophilic), but 2 M HCI/ 0.5M H20 2 mixture used as receiver solution I• 0.11 M MEHPA + 0.08 M DEHPA a 0.08 M DEHPA \ 10.000 -,,------, • • •• I.OOO • • 0.100 • ~ • • 0.010 • 0.001 0.10 1.00 10.00 I (H2S04] /M L Figure 4.14 Distribution coefficient for cerium(IV) 93 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane Finally, the addition of another additive commonly used in solvent extraction systems as a modifier to improve phase separation was tested. The additive was tributyl phosphate (TBP). It was found that the flux from feed to solvent through the same hydrophobic membrane decreased from 3.84 x 1o-s to 1.02 x 10-s with addition of 4 vol.% TBP to a 0.2 M DEHPA solution. This decrease is discussed later in this Chapter. 4.3.2.4 Effect of receiver solution composition Cerium(IV) is extracted from sulfuric acid solutions according to Equation 4.12. It follows that low acid concentrations are required for extraction and high acid concentrations for stripping. The concentration of sulfuric in the receiver solution was fixed at 5.5 M, since it is at this value that the distribution coefficient is at its lowest (see Chapter 3, Figure 3.8). At higher concentrations, cerium(IV) extraction increases with increasing acidity due to the formation of a different cerium complex. 4.12 Cerium stripping is also possible via a reductive stripping reaction shown in Equation 4.13. 4.13 This reaction reduces the cerium(IV) to cerium(III) which is not extractable and therefore there is no mechanism for cerium to diffuse back to the feed solution. This effectively reduces the cerium(IV) concentration in the receiver solution to zero, providing an opportunity to further push the equilibrium of Equation 4.12 to the right. There is also the advantage that cerium can be concentrated from a sulfate medium to a chloride medium, which tends to be more favourable for further cerium processing. With these processing advantages in mind, a couple of tests were carried out with a receiver solution of 1 M HCI and 0.5 M H20 2• A membrane combination of hydrophobic in the feed and hydrophilic in the strip side was chosen. The results are presented in Figure 4.13 and show that the same curve was obtained as with sulfuric acid. Thus a change in stripping reaction has had no effect on the overall cerium permeation. 94 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.3.2.5 Effect of cerium concentration in the feed Figure 4.15 shows the effect of cerium concentration in the feed on the flux between feed and solvent, and on the overall flux from feed to strip. At low cerium concentrations (below 5 mM cerium), the flux increases linearly with cerium concentration. A plateau is reached at approximately 1O mM cerium, where both fluxes are independent of cerium concentration in the feed. In the linear part of the curve, the diffusion of cerium through the aqueous boundary layer is likely to play a significant role in the overall transport. As the curve reaches a plateau the resistance due to chemical reaction becomes more important when compared to aqueous diffusion. However, what is not clear from such a graph is the role played by membrane diffusion relative to chemical reaction and aqueous diffusion. In order to quantitatively determine the contributions of all these factors to the overall resistance, the relevant mass transfer coefficients, as well as the chemical reaction kinetics must be determined. This requires better defined hydrodynamics than is available on the permeation cell used for the current tests. Such a study has been undertaken with a modified Lewis cell and is the subject of Chapter 5 in this thesis. 4.3.3 lnterfacial measurements The interfacial properties of the solvent system in contact with the aqueous phase is very important since it affects interfacial concentrations of carrier at the interface. This concentration in turn has a significant effect on the rate of interfacial chemical reactions. lnterfacial tensions measurements of DEHPA and MEHPA under various conditions have therefore been studied, in order to determine whether there are significant differences that can explain the accelerating effect of MEHPA on the extraction of cerium(IV) with DEHPA. lnterfacial activity of DEHPA The Du Nuoy ring method was used for interfacial tension measurements. Details are presented in Chapter 2, Section 2.5.4. Results for DEHPA in contact with 0.55 Mand 5.5 M H2S04 solutions are shown in Figure 4.16. The interfacial tension decreases with increasing concentration of DEHPA in the bulk phase. This indicates that DEHPA is interfacially active and the decrease of interfacial tension is a measure of the population density of DEHPA molecules at the interface. The molecules are attracted to the interface 95 Chapter4 Transport of cerium with flat sheet bulk liquid membranes ~eed ,ic Receiver ] ------4.0 .,, -0 .<>····· ...... ~ ...... 3.0 ,.,· ,.···· >< ... ··' ':" Ill "! 2.0 E 0 g 1.0. >< ::I u::: 0.0 0 5 10 15 20 25 1--~- [Ce] (mmol dm·3 ) Figure 4.15 Effect of cerium(IV) concentration in the feed on flux from feed to solvent (JF) and from feed to receiver (JR) [DEHPA] = 0.2 M, Feed: 0.55 M H2S04, Receiver: 2 M HCI / 0.5 M H20 2 ··---~-~------[o 5.5 M H2S04 o 0.5 M H2S04 I 50 E 45 t ~ o o 8"· i;·~'"··1!L. z4o.j 19 g .§.~ ·~o 8w ~ -~ 25 J=lo0 o GI (Ill> 0 .0.. I- 20 90 "ii ·;:; 15 ra i 10 I ~~-E--0-9_1_E--08_1_E--0-7 1E-06 1E-05 1E-04 1E-03 1E-02 1E-~OO [DEHPA] M Figure 4.16 lnterfacial tension measurements of the DEHPA/heptane system The dotted curves have been fitted using the Szyskowski equation 96 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane because the polar part of the molecule is hydrophilic. The concentration of carrier molecules at the interface, known as the surface excess is theoretically given by the Gibbs equation (shown in Equation 4.1 ), as discussed in Section 4.2.3 of this chapter. According to this treatment, the maximum extractant interfacial concentration that can be obtained in the system r max (mol m-2), can be estimated from the slope of the linear part of (r O - r) vs ln C plot as shown in Figure 4.17. Cmin• which is the minimum bulk extractant concentration necessary to saturate the interface is estimated to be at the point where the plot deviates from linearity. The estimation of Cmin by this method is subject to high errors. Results for the estimation of Cmin• r max and the saturated interfacial area (A2 molecule-1), for the DEHPA data are given in Table 4.7. The interfacial activity of DEHPA has been studied by many researchers and details of DEHPA systems in aliphatic diluents are given in Table 4.7. The data presented in this work are in reasonable agreement with that obtained by other workers. It is apparent that the maximum surface excess is not significantly affected by the nature of the aqueous media with respect to both the ionic strength and the type of aqueous media, namely, chloride, nitrate or sulfate. The value for Cmin has not been presented for the 5.5 M sulfuric acid data because of the large errors associated with its determination. TABLE 4.7 lnterfacial parameters estimated from the Gibbs equation rmax saturated Gibbs Media/Solvent cmin interfacial area Reference M x106 mol m-2 A2 molecu1e-1 5 0.55 M H2S04' heptane 5.5 X 10- 0.89 186 This work 5.5 M H2S04' heptane 0.79 209 This work 0.001 M HN03' 8 X 10-3 0.92 181 (Vandegrift 1980) dodecane 0.1 M HN03' dodecane 2 X 10-3 0.90 184 (Vandegrift 1980) 1 M HN03' dodecane 2.5 X 10-4 0.91 182 (Vandegrift 1980) 0.01 M HN03' 6 X 10-3 1.2 135 (Cox 1983) dodecane 0.01 M HCI / heptane 2.5 X 10-4 0.84 198 (Lim 1996) 97 Chapter4 Transport of cerium with flat sheet bulk liquid membranes ----·-----·------~-·------ 30 25 y= 2.2003x+ 26.119 ...... 20 R2 = 0.9869 .... ~ .. ·e 15 z ••• .§.10 , . Cl i- •• .... 5 0 -5 f--"-~+-"-~--+--"~'---+--"~-'--l-~-'--4-'-'~I-L-~+-'-l____j__+-·_L__j______!---'-"~~ -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 Ln [DEHPA] {M) -----·------' Figure 4.17 Gibbs treatment of interfacial tension data of the DEHPA/heptane/ 0.55 M H2S04 system [Li. 5.5 M H2S04 + 0.55 M H2S04 I .. 2.0 ....0 1.8 >< if' 1.6 E 1.4 0 1.2 .§. 1.0 "'CII "'CJ 0.8 >< w 0.6 CII CJ 0.4 ~ ::I 0.2 U) 0.0 ...... •• 1E-08 1E-07 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+OO [DEtFA] M Figure 4.18 Surface excess for the DEHPA/heptane system calculated from Equation 4.5 98 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane A better way to determine Cmin is by calculating the surface excess over the entire concentration range. This involves fitting the data to a mathematical expression as explained in Section 4.2.3. The Szyszkowski equation (Equation 4.2) was used for this purpose and the fitted curves are shown in Figure 4.16. In order to precisely calculate the surface excess, additional information on the system being studied is required. For instance, an acidic extractant like DEHPA can undergo various types of equilibria: i) partitioning between the aqueous and organic phase where Kp is the apparent partitioning constant. The overbar indicates the presence in the organic phase. K - HR< e >HR 4.14 ii) acid dissociation where Ka is the apparent acid dissociation constant 4.15 iii) dimerization in the organic phase 4.16 Cote and Szymannowski (Cote 1992) presented the case for consideration of these reactions in the treatment of interfacial data and incorporation into the Gibbs equation as follows: TABLE4.8 Equilibrium constants for DEHPA in heptane measured at 25°C in contact with 0.1 (Na,H) N03 (Sella 1988) 1.49 3.2 4.5 99 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane The apparent equilibrium constants for DEHPA reactions 4.14 - 4.16, are given in Table 4.8. The numbers indicate that DEHPA has a low solubility in the aqueous phase and in acidic conditions it is not dissociated. In this case only the dimerisation reaction need be considered and the interfacial tension can be plotted as a function of monomer HR, dimer (HR) 2 or total organic DEHPA concentration, as recommended by Cote & Szymannowski (Cote 1992), and shown in Equations 4.18-4.20. 4.18 4.19 4.20 where m =.!.+---1-- HR 2 2(1+8KdC)l/2 The value of m changes from 1 to 0.5, reflecting the relative amounts of monomer and dimer present at a given total analytical DEHPA concentration. At very low DEHPA concentration, the monomer species is predominant and m is 1, while at high DEHPA concentrations, the dimer is the main species and m is 0.5. It is evident from Equation 4.20, that when the data are plotted against its analytical concentration, as is commonly done, without consideration of the dimerisation constant, it can give rise to discrepancies in the calculation of surface excess. Bearing in mind the above discussion, the surface excess was calculated as a function of DEHPA concentration (shown in Figure 4.18). The maximum surface excess and the free energy of adsorption have also been computed and are shown in Table 4.9. The results are in very good agreement with DEHPA interfacial data treated in this manner, but are about double the values obtained from the Gibbs equation, not taking into account the dimerisation of DEHPA. They confirm that the surface excess is not sensitive to aqueous media as discussed previously. Furthermore the nature of the aliphatic media does not seem to have a major influence. Results obtained for this work where heptane was used as the diluent were similar to those published for dodecane and hexane. 100 Chapter4 Transport of cerium with a nat sheet bulk liquid membrane TABLE4.9 lnterfacial parameters for DEHPA calculated using the Szyszkowski equation rmax tiGat1 Aqueous/solvent Sz Reference x106 mol m·2 kJ mo1·1 DEHPA system 0.55 M H2S04' heptane 1.7 -30 This work 5.5 M H2S04' heptane 1.6 -30 This work pH 1 / hexane 1.6 -35.7 (Cote 1992) 1 M HN03 / dodecane 1.7 -33.1 (Cote 1992) An interesting point to come out of this data treatment is the value of Cmin· As discussed previously this is around 5.5 x 10-5 M DEHPA using the Gibbs equation. The value is increased to 3 x 10-4 M DEHPA by reading off the surface excess curve at> 98% saturation, using the Szyszkowski equation to describe the interfacial tension. When a correction is made for the dimerisation of DEHPA, Cmin increases again to 8.6 x 10·3 M DEHPA as shown in Figure 4.18. This illustrates that the value read from the Gibbs plot is not very meaningful. lnterfacial activity of MEHPA The interfacial activity of MEHPA was also investigated. Unfortunately, no AR grade MEHPA reagent could be obtained. The reagent that was used was a mixture of MEHPA and DEHPA. The exact proportions of each component were determined by potentiometric titration. Two experimental approaches were taken. One was to dilute the reagent as received to a range of concentrations, thus keeping the MEHPA/DEHPA ratio constant at 57% MEHPA. The other approach was to prepare a range of solutions containing a constant amount of DEHPA but varying amounts of MEHPA. Experimental data obtained are shown in Figure 4.19. The theoretical treatment of the data is discussed below. The Gibbs adsorption equation for two dilute solutes is given by (Chattoraj 1984): 101 Chapter4 Transport of cerium with flat sheet bulk liquid membranes [ • DEHPA O 57% MEHPA • Fixed 0.01 M DEHPA [ 50 Ie 45 __ z 40 • .§. 35 -'OD•• + +41t + ~ 30 ":?! 25 a ·~ .... {!!. 20 l! 15 ODD ••• u • ~ 10 ° I 1 ~ +--~LU.ilJ-~~--~.u.u+~-'--LI..LUJ---" """I ' """'I ' "~QC~ .. ""I ' ""~'I 1 E-09 1 E-08 1 E-07 1 E-06 1 E-05 1 E-04 1E-03 1 E-02 1 E-01 1 E+OO [DEi-FA] + [MetPA] M Figure 4.19 lnterfacial tension data for DEHPA and mixed DEHPA/MEHPA Legend: DEHPA: DEHPA/heptane/0.55 M H2S04 system 57% MEHPA: DEHPA+MEHPA/heptane/0.55 M H2S04 system Fixed 0.01 M DEHPA: DEHPA (0.01 M) + MEHPA (0.001-0.013 M) /heptane/0.55 M H2S04 system j <> 0.5 M H2S04 • DEHPA loaded with Ce j 50 ~------~ e 45 <> • <> <> e<> <> <> <> <> z 40 • • S. 35 5 30 "! 25 CD 1- 20 't:~ 1015 j I 5 0 '' ""I ''""'I ""'I '"""I "'I ''"""I ''"""I ''"""I 1E-10 1E-09 1E-08 1E-07 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+OO [DEHPA] M Figure 4.20 lnterfacial measurements for DEHPA and DEHPA loaded to 80% of its capacity with cerium 102 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 4.21 At constant DEHPA concentration, the surface excess of MEHPA is given by Equation 4.22. 1 0 4.22 r MEHPA = - - -(---r--J RT olnCMEHPA CDEIIPA In the case where the proportion of the two solutes is kept constant, only the total surface excess can be calculated. 1 or 4.23 rtota1 = -RT 81 C n total where ctotal = C DEHPA + C MEHPA From the data shown in Figure 4.19, it is apparent that MEHPA has a higher interfacial activity than DEHPA. This is in keeping with the chemical structure of the two compounds. MEHPA has an extra hydroxy group compared to DEHPA, which increases its attraction to the aqueous interface. As the MEHPA is adsorbed at the interface, the interfacial tension decreases. The curve for the DEHPA/MEHPA mixture reaches a point, known as the critical micelle concentration (CMC), where the interfacial tension is at its lowest, and does not change with a further increase in extractant concentration. In this case, this occurs at a MEHPA + DEHPA concentration of 0.01 M. Beyond the CMC concentration, the extractant molecules form micelles, with their polar groups directed to the center of the micelles. The micelles do not adsorb at the interface. In the case of DEHPA, the CMC was not reached at a concentration of 1 M and therefore it is inferred that the micelle formation in the mixed curve is due to MEHPA. The calculations for the surface excess are shown in Table 4.10. The maximum surface excess of MEHPA was calculated from the Gibbs Equation 4.22 as 2.5 x 1o-s mol m-2• This is higher than the value obtained for DEHPA. The total surface excess for the MEHPA/DEHPA mixture is similar at 2.7 x 10-s mol m-2, with the value calculated using the Szyszkowski equation, being in good agreement with that obtained from the Gibbs equation. From these results it is not possible to determine the actual proportion of MEHPA and 103 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane DEHPA at the interface for any given mixture. However, since the population density for the MEHPA/DEHPA mixture is approximately 1.7 times higher than that due to DEHPA alone, and MEHPA is more interfacialy active than DEHPA, it is probable that the presence of MEHPA actually decreases the adsorption of DEHPA at the interface. TABLE4.10 lnterfacial parameters for MEHPA 0.55 M H2S04' heptane rmax rmax Experimental approach Sz Gibbs x106 mol m-2 x106 mol m-2 Constant DEHPA at 0.01 M 2.5 Surface excess of MEHPA calculated ------Constant molar ratio 57% MEHPA 2.9 2.7 Total surface excess of MEHPA + DEHPA calculated lnterfacial activity of cerium-DEHPA complex During mass transfer there maybe a change of interfacial tension, depending on whether the products of a reaction are more or less interfacially active than the respective reactants. Any change in interfacial tension causes an interfacial tension gradient. If the interfacial tension increases with mass transfer the system is said to be surface tension positive. Conversely, a system is surface tension negative if the interfacial tension decreases and surface tension neutral if there is no change. lnterfacial tension gradients are important because they affect interfacial stability. It has been shown that large interfacial gradients promote increased mass transfer, but in surface tension negative systems this may lead to increased solvent losses due to emulsion formation in the membrane pores (Zha 1993). The interfacial activity of the cerium-DEHPA complex was investigated in order to determine the type of interfacial tension gradient present at the feed/solvent interface. A 0.3 M DEHPA solution was loaded to about 80% of its capacity corresponding to 11.5 g L-1 cerium in the organic. The solution was then diluted to various concentrations with heptane and the values compared with those of DEHPA alone. Results are shown in Figure 4.20. The curves for DEHPA and 104 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane DEHPA loaded with cerium are very close and differ by only a few percent. However, since there are still free DEHPA molecules in the system (ie. the cerium-DEHPA complex was not isolated and purified), one cannot calculate the surface excess of the cerium-DEHPA complex. It can nevertheless be stated that the system has no significant interfacial tension gradient during extraction. The effect of mass transfer on the interfacial tension gradients for both extraction and stripping was further investigated by determination of the interfacial tension before and after equilibrium. A DEHPA concentration of 0.2 M was chosen, which is high enough to guarantee full saturation of the interface by DEHPA molecules. Results are shown in Table 4.11. For these particular conditions, the surface tension increased slightly (2.5%) with cerium loading confirming that the feed/solvent interface the system is surface tension neutral or slightly positive. At the solvent/receiver interface the system is surface tension negative, with the interfacial tension gradient being more pronounced with reductive hydrochloric acid stripping than with sulfuric acid stripping. TABLE 4.11 lnterfacial parameters for 0.2 M DEHPA in heptane % Aqueous / solvent system 'Y mNm·1 change aqueous: 0.55 M H 2S04, 300 ppm Ce before equilibration 23.5 solvent: no Ce after equilibration 24.1 +2.5 % aqueous: 5.5 M H 2S04, no Ce before equilibration 23.7 solvent: loaded with - 300 ppm Ce after equilibration 22.7 -4.2 % aqueous: 2 M HCI & H 20 2, no Ce before equilibration 22.6 solvent: loaded with - 300 ppm Ce after equilibration 20.7 - 8.4 % lnterfacial activity of DEHPA I TBP mixture Addition of TBP to the solvent phase was found to decrease the cerium flux from feed to solvent, while that of MEHPA had the opposite effect. The effect of addition of TBP on the interfacial tension was investigated, and is shown in Figure 4.21. It apparent that TBP, like MEHPA decreases the interfacial tension of the system, although this effect is not as pronounced as that of 105 Chapter4 Transport of cerium with flat sheet bulk liquid membranes ------:=-======: ----~--- I A TBP - -<> -- MEHPA I 30-~------~-~---- e 25 z .§. 20 C 0 "iii C 15 ~ iii ~ ·u 10 <> i! -~ 5 <> % 0.00 0.05 0.10 [MEHPA] / [TBP] M ____ _J Figure 4.21 Effect of additives MEHPA and DEHPA on the interfacial tension Legend: TBP: addition ofTBP to 0.13 M DEHPA/heptanel0.55 M H2S04 MEHPA: addition of MEHPA to 0.01 M DEHPA/heptanel0.55 M H2S04 106 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane MEHPA. The interfacial tension trend cannot therefore explain the opposite effects on cerium mass transfer. Other factors contribute to this effect and these are discussed below. 4.4 General Discussion In this chapter, the transport of cerium was optimised with respect to a range of different physical and chemical parameters. The flat sheet bulk liquid membrane permeation cell has proved to be very useful for this purpose. The one drawback of the experimental set-up was the lack of external pressure control, with the consequent leaking of one phase into another in prolonged experiments. With hydrophilic membranes at the solvent/receiver interface, serious loss of receiver solution to the membrane phase was observed. In order to prevent such loss, the non-impregnating phase (ie the solvent) has to be kept at a higher pressure than the impregnating phase, and that can only be done by applying some external pressure. Another factor that may have contributed to the leakage is the negative interfacial tension gradient of the system at the solvent/receiver interface. In a study on the instability of SLM (Zha 1993), it has been argued that a cf system causes interfacial turbulence due to Marangoni effects and stimulates instability at the interface. This instability promotes emulsion formation. In this work, the receiver/solvent interface is at the solvent side of the hydrophilic membrane support and once the emulsion is formed it will be incorporated into the bulk solvent phase by sheer forces, and hence the observed accumulation of receiver solution in the bulk organic. The presence of the bulk solvent enabled measurements of both the flux between feed and solvent and the overall flux between feed and receiver solutions to be measured. It was found that the overall mass transfer of cerium is slower than mass transfer from feed to solvent, indicating that the stripping process is rate determining. This information is useful for design of a continuous process. In the design of the hollow fibre, bulk liquid membrane apparatus, discussed in Chapter 6, the lower permeation rate was compensated by providing a greater surface area in the strip module. Two chemically different stripping processes were compared but found to have the same overall permeation rates. The mass transfer of cerium from feed to solvent was dramatically increased by the presence of MEHPA. Detailed 107 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane interfacial tension measurements showed that MEHPA decreases the interfacial tension of the DEHPA/Heptane/0.55 M H2S04 system, and its presence increases the overall surface excess. On the other hand, TBP also reduces the surface tension of the system, but has an opposite effect on the cerium mass transfer. The difference lies in the ability of the additives in forming cerium complexes that are soluble in the solvent phase. It was shown in this work that MEHPA is a stronger complexing agent for cerium than DEHPA. Furthermore, it is known that TBP does not extract cerium(IV) from sulfate solutions (Douglass 1959). So whilst both additives are attracted to the interface, MEHPA increases the carrier concentration at the interface, while TBP decreases it, and therefore slows down the rate of mass transfer. It is probable that MEHPA is acting as a phase transfer agent, complexing cerium at the interface. The Cerium-MEHPA complex diffuses to the bulk phase, where rapid ligand exchange with DEHPA occurs, due to the higher bulk DEHPA concentration. Once the MEHPA molecules are free of cerium they are again attracted to the interface, displacing DEHPA molecules. If the bulk MEHPA concentration increases beyond approximately 0.01 M, it has a negative effect on the overall mass transfer. When the bulk MEHPA concentration is high, the cerium-MEHPA complex remains in the solvent phase and does not strip readily because its equilibrium distribution coefficient is high even at 5.5 M sulfuric acid concentration. 4.5 Conclusions The cerium permeation experiments demonstrated that the greatest flux was obtained with a hydrophobic membrane at the feed-solvent interface and a hydrophilic membrane at the solvent-receiver interface. Three different types of hydrophilic membranes were tested but these had no significant effect on the overall flux. Leakage of receiver solution to the solvent phase was observed due to the lack of control of the pressures applied on the either side of the membrane and the negative interfacial gradient of the system. The overall mass transfer of cerium was found to be slower than that of feed to solvent, indicating that the stripping process is controlling the rate of cerium transport. 108 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane The presence of the impurity MEHPA in the DEHPA solvent was found to have an accelerating effect on the permeation flux. This was attributed to its higher interfacial activity and population, and the fact that MEHPA is also an active carrier for cerium(IV). Both the interfacial activity and extraction capability of MEHPA were experimentally verified. 4.6 Nomenclature Symbol A membrane area (m2} C molar concentration (mol L· 1) flux measured from feed to solvent (mol m·2 s· 1) flux measured from feed to receiver (mol m·2 s·1) overall mass transfer coefficient (m s· 1) acid dissociation constant for DEHPA dimerization constant for DEHPA partition constant for DEHPA mass transfer coefficient (m s·1) distribution coefficient gas constant (8.3143 J mo1·1 K· 1) time (s) absolute temperature (°K) volume (m-3) Greek letters 8 boundary layer or membrane thickness (m) y interfacial tension (mN m·1) r surface excess (mol m·2) Subscripts a aqueous F feed m membrane support 109 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 0 solvent R receiver 0 initial Acronyms and abbreviations CeR4 cerium-DEHPA complex CMC critical micelle concentration DEHPA di-2-ethylhexyl phosphoric acid ELM emulsion liquid membrane HFCLM hollow fibre contained liquid membrane HR &(HR)2 monomeric and dimeric forms of DEHPA MEHPA mono 2-ethylhexyl phosphoric acid TBP tri-butyl phosphate 4. 7 References 1. Abrantes L.M., Paiva A.P. and Cote G. 1996. lnterfacial tension data of organophosphorous ligands in the solvent extraction of silver. ISEC'96- Value Adding Through Solvent Extraction. 2. Ashrafizadeh S. N. and Demopoulos G.P. 1996. Supported liquid membrane vs conventional liquid-liquid extraction of Rh(II I) from chloride media. ISEC'96 - Value Adding through Solvent ExtractionVol 2, 941-46. 3. Babcock W. C., Baker R. W., Kelly D. J. and LaChapelle E. D. 1980. Coupled transport membranes for uranium recovery. International Solvent Extraction Conference (ISEC'BO)Vol 2, 80-90. 4. Babcock W. C., Baker R.W., Lachapelle E.D. and Smith K.L. 1980. Coupled transport membranes Ill: The rate-limiting step in uranium transport with a tertiary amine. Journal of Membrane Science 7: 89-100. 5. Boyadzhiev L. 1990. Liquid pertraction or liquid membranes - State of the art. Separation Science and Technology 25, no. 3: 187-205. 110 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 6. Boyadzhiev L. and Alexandrova S. 1994. Recovery of copper from ammoniacal solutions by rotating film pertraction. Hydrometal/urgy 35: 1059-121. 7. Boyadzhiev L. and Bezenshek E. 1988. Liquid film pertraction. Recovery of iodine from iodine-containing aqueous solutions. Journal of Membrane Science 37: 277-85. 8. Bromberg L., Lewin I. and Warshawsky A. 1992. Membrane extraction of silver by di (2-ethylhexyl)dithiophosphoric acid. Journal of Membrane Science 70: 31-39. 9. Butcher C. 1994. Liquid membranes: The time is right. The Chemical Engineer. 21-22. 10. Chattoraj D. K. and Birdi K.S. 1984. Adsorption and the Gibbs surface excess. 83-89. New York: Plenum Press. 11. Chaudry M.A., Amin S. and Malik M. T. 1996. Tri-n-octylamine-Xylene based supported liquid membranes and transport of Ce(IV) ions. Separation Science and Technology 31, no. 9: 1309-26. 12. Chitra K. R., Gaikwad G.D., Surender G.D. and Damodoran A.O. 1997. Studies on complexation and ion transport mechanism of yttrium in a liquid membrane system. Hydrometallurgy 44: 377-94. 13. Cote G. and Szymanowski J. 1992. Processing of interfacial tension data in solvent extraction studies. lnterfacial properties of various acidic organophosphorus extractants. J. Chem. Tech. Biotechnol. 54: 319-29. 14. Cox M., Elizalde J.C., Castresana J. and Miralles N. 1983. ISEC'83 New developments in liquid-liquid extractors. 15. Danesi P. R. and Reichley-Yinger L. 1986. Origin and significance of the deviations from pseudo first order rate law in the coupled transport of metal species through supported liquid membranes. Journal of Membrane Science 29: 195-206. 111 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 16. Douglass D. A. and Bauer D. J. 1959. Liquid-liquid extraction of cerium, 5513. US Bureau of Mines. 17. Draxler J. and Marr R. 1990. Emulsion liquid membrane for waste water treatment. ISEC'90 - Solvent extraction 1990Part A - 37-48. 18. El Bachiri A., Hagege A. and Burgard M. 1996. Recovery of silver nitrate by transport across a liquid membrane containing dicyclohexano 18 crown 6 as a carrier. Journal of Membrane Science 121, no. 2: 159-68. 19. Hartmann D., Pareau D., Chesne A. and Durand G. 1990. Extraction of uranium from heap leaching solutions by liquid surfactant membranes. ISEC'90- Solvent extraction 1990Part 8, 1567-72. 20. Hidalgo M., Masana A. and Salvado V. 1991. Accelerated mass transfer of palladium(II) through a selective solid-supported liquid membrane containing Cyanex 471. Analytica Chimica Acta 251: 233-39. 21. Hutter J.C., Vandegrift G. F., Nunez L. and Redfield D. H. 1994. Removal of VOCs from Groundwater Using Membrane-Assisted Solvent Extraction. A/CHE Journal 40, no. 1: 166-77. 22. Kedem 0. and Bromberg L. 1993. Ion-Exchange Membranes in Extraction Processes. Journal of Membrane Science 78, no. 3: 255-64. 23. Kim B. M. 1984. Membrane-based solvent extraction for selective removal and recovery of metals. Journal of Membrane Science 21: 5-19. 24. Kondo K., Hashimoto T., Sumi H. and Matsumoto M. 1995. Mechanisms of samarium extraction with diisostearylphosphoric acid and its permeation through supported liquid membrane. Journal of Chemical Engineering of Japan 28, no. 5: 511-16. 25. Koros W. J., Ma Y. H. and Shimidzu T. 1996. Terminology for membranes and membrane science.(Reprinted from Pure and Appl. Chem., vol 68, 1996, 1479-1489). Journal of Membrane Science 120, no. 2: 149-59. 112 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 26. Lee C. J. and Wang S. S. 1995. Penicillin G extraction by Amberlite LA-2 - A study of interfacial reaction kinetics by interfacial tension measurements. Journal of Chemical Technology and Biotechnology 64, no. 3: 239-44. 27. Li N. N., Cahn R.P., Naden D. and Lai R.W.M. 1983. Liquid Membrane Processes for Copper Extraction. Hydrometallurgy 9: 277-305. 28. Lim, T. M. 1996. "Kinetic studies of solvent extraction of rare earths into di- 2-ethyl hexyl phosphoric acid (DEHPA)." PhD Thesis, The University of New South Wales. 29. Majumdar S. and Sirkar K.K. 1992. Hollow-Fiber Contained Liquid Membrane in Membrane Handbook. Eds W. S. W. Ho and K. K. Sirkar, 764-808. New York: Van Nostrand Reinhold. 30. Marr R., Bart H. J. and Draxler J. 1989. Solvent extraction and liquid membrane permeation - a critical comparison. Proc. lnt. Cont. Sep. Sci. Techno/.Baird M. H. I. and Vijayan S. 2, 403-10, Ottawa, Ont.: Can. Soc. Chem. Eng. 31. Ortiz M. I., Galan B., Alonso A. I. and lradien J. A. 1996. Simultaneous Extraction and Back Extraction of Cr(IV) in Hollow Fibre Modules. Value Adding Through Solvent Extraction, Proceedings of /SEC '96, 2, 905-10, Melbourne: The University of Melbourne. 32. Prasad R. and Sirkar K. K. 1992. Membrane Based Solvent Extraction in Membrane Handbook. Eds W. S. W. Ho and K. K. Sirkar, 727-63. New York: Van Nostrand Reinhold. 33. Qiang Y., Jingfen Y., Jiufang L. and Teng T. 1989. lnterfacial characters of organophosporic and phosphonic acid extraction systems. Rare Metals 8, no. 2: 5-10. 34. Raghuraman B. and Wiencek J. 1993. Extraction with Emulsion Liquid Membranes in a Hollow-Fiber Contactor. A/CHE Journal 39, no. 11: 1885- 89. 35. Sella C. and Bauer D. 1988. Solvent Extr. Ion Exch. 6: 819-33. 113 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 36. Sengupta A., Basu R. and Sirkar K. K. 1988. Separation of Solutes from Aqueous Solutions by Contained Liquid Membranes. A/ChE Journal 34, no. 10: 1698-708. 37. Szymanowski J. Prochaska K. Alejski K. 1990. lnterfacial behaviour of LIX 65N and surface kinetics of copper extraction. Hydrometallurgy 25: 329- 48. 38. Szymanowski J., Cote G., Sobczynska A., Firgolski K. and Jakubiak A. 1994. lnterfacial Activity of Decyl Pyridinemonocarboxylates. Solvent Extraction and Ion Exchange 12, no. 1: 69-85. 39. Teramoto M., Matsuyama H., Takaya H. and Asano S. 1987. Development of spiral-type supported liquid membrane module for separation and concentration of metal ions. Separation Science and Technology 22, no. 11: 2175-201. 40. Teramoto M., Matsuyama H., Yamashiro T. and Okamoto S. 1989. Separation of ethylene from ethane by a flowing liquid membrane using silver nitrate as a carrier. Journal of Membrane Science 45: 115-36. 41. Teramoto M., Tohno N., Ohnishi N. and Matsuyama H. 1989. Development of a spiral-type flowing liquid membrane module with high stability and its application to the recovery of chromium and zinc. Sep. Sci. Technol. 24, no. 12: 981-99. 42. Valenzuela F., Basualto C., Tapia C. and Sapag J. 1999. Application of hollow fibre supported liquid membrane technique to the selective recovery of a low content of copper from a Chilean mine water. Journal of Membrane Science 155: 163-68. 43. Vandegrift G. F. and Horwitz E. P. 1980. lnterfacial activity of liquid-liquid extraction reagents - I. J. /nor. Nucl. Chem. 42: 119-25. 44. Techniques of organic chemistry. 1949. Physical methods of organic chemistry Part 1.A. Weissberg, 363. New York: lnterscience. 114 Chapter4 Transport of cerium with a flat sheet bulk liquid membrane 45 Wijers C. 1996. IMSTEC'96 - Proceedings of the international membrane science and technology conference, 1. 46. Wijers, C. 1996. "Supported liquid membranes for the removal of heavy metals: Permeability, selectivity and stability." PhD Thesis, University of Twente. 47. Zha, F. 1993. "Stability and applications of supported liquid membranes." PhD Thesis, The University of New South Wales. 48. Zhouri A., Burgard M. and Lakkis D. 1995. The use of dicycloheaxano 18 crown 6 as an extractant-carrier for the recovery of chromic acid. Hydrometallurgy 38: 299-313. 115 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction Summary The mass transfer rate of membrane assisted extraction of cerium into a solvent phase containing di-2-ethylhexyl phosphoric acid and its back-extraction into an acidic aqueous phase is investigated with a modified Lewis-type membrane permeation cell. This work quantitatively determines the relative contributions of aqueous and organic boundary layer diffusion, diffusion in the membrane pores and the rate of the forward and reverse chemical reactions to the mass transfer of cerium through microporous membranes. It is shown that stripping is slower than extraction. The importance of the distribution coefficients, chemical reaction rate and membrane morphology in the overall process is discussed. Chapters The role of reaction kinetics in membrane assisted solvent extraction 5.1 Introduction In developing membrane extraction processes, it is very important to understand the chemical reactions that occur, both from a thermodynamic and a kinetic point of view. The kinetics and the mechanism of extraction and back extraction reactions are especially important because one of the major differences between conventional solvent extraction and liquid membrane processes is that the latter are governed by kinetics rather than by equilibrium. In membrane assisted solvent extraction, diffusion in the membrane pore is likely to play a significant role in the kinetics of the overall process, and has often been assumed to be the dominant resistance to mass transfer in supported liquid membrane systems. The present work aims to quantify the contribution made by both diffusion and chemical processes to the overall membrane extraction and back extraction of cerium with di-2-ethylphosphoric acid. The author would like to acknowledge the contribution of two undergraduate summer vacation students, Catherine Chan and Kate MacFarlane, who carried out some of the experimental work presented in this chapter. 5.2 Background In metal solvent extraction systems, a chemical reaction accompanies the transfer of a solute from one bulk liquid phase to another. The overall chemical reaction can be broken down into many individual reaction steps, such as extraction aggregation, dissociation and distribution; ligand exchange, metal complexation, etc. These reactions occur either in the bulk phase or in the interfacial zone and can in principle all be slow enough to determine the rate of reaction. The kinetics of solvent extraction, however, are a function of both chemical reaction and diffusion, and both of these processes have to be taken into account when studying the kinetics of metal solvent extraction (Danesi 1984). Experimentally, kinetic studies involve the measurement of changes in metal concentrations as a function of time. The data are then arranged into an empirical rate expression and correlated with the most likely extraction mechanism. Since the rate of mass transfer is a function of both hydrodynamics and the concentrations of the species involved in the chemical reaction, different apparatus and experimental conditions will give rise to 116 Chapters The role of reaction kinetics in membrane assisted solvent extraction different rate expressions. These difficulties are often the source of disagreement in proposed reaction mechanisms. Excellent reviews on the subject have been given by (Coleman 1971) and by (Danesi 1980) and (Danesi 1992). 5.2. 1 lnterfacial Zones In practice, in a two phase system, the bulk phases are sufficiently agitated to be considered homogeneous, but as the interface is approached diffusion of reacting species towards and away from the interface has to be taken into consideration. In their review of the kinetics of metal solvent extraction, Danesi and Chiariza used the two film theory to describe the interfacial zone, consisting of two stagnant diffusion layers adjacent to the interfacial plane (Danesi 1980). The thickness of these two films depends on the hydrodynamics of the experimental set-up as well as the physico-chemical properties of the solutions. Effective film thicknesses are generally in the range of 1 - 100 µm (Danesi 1992) for well stirred systems, but can be as large as 3000 µm (Danesi 1980) for systems with poor hydrodynamics Many extracting reagents, such as organophosphorous acids, exhibit surface active properties and adsorb at the interface (Vandegrift 1980), (Qiang 1989). The adsorbed extractant molecules are thought to align at the interfacial plane, with the hydrophobic part of the molecule remaining in the solvent side of the interface and the hydrophilic part in the aqueous side. In polar extractants such as the organophosphorous acids, the hydrophobic part of the molecule consists of the alkyl chains, while the hydrophilic part consists of the phosphoric acid functional group. Some researchers have suggested that the interface is not a monolayer, but can be considered as an "extended surface" of a few molecular layers (Chen 1994). Qiang et al. have gone further and developed a new model called the two-film adsorption model, where it is suggested that the interfacial zone consists of two adsorption layers either side of the interface, in addition to the diffusional layers (Qiang 1989). This zone is microscopic in character with the adsorption film thickness being of the order 1o-s m. In contrast, the stagnant diffusion films are macroscopic, since the diffusion distances exceed the molecular dimensions by several orders of magnitude (Danesi 1992). A schematic diagram depicting the interface is shown in Figure 5.1. 117 Chapters The role of reaction kinetics in membrane assisted solvent extraction When referring to species at the interface, it is important to differentiate between interfacially adsorbed species and interfacial concentrations. The species adsorbed at the interface are expressed as moles per unit area, and in this case it is assumed that the interface is a surface of zero thickness. lnterfacial concentrations (moles per unit volume), on the other hand, refer to concentrations at the extreme limit of the diffusional layers. bulk organic organic diffusion layer extreme limit of a{ organic diffusion layer 1-100 ~tm ------ adsorbed interface extractant microscopic scale molecules 2-10 nm aqueous ------extreme limit of diffusion layer a.{ aqueous diffusion layer 1-100 µm bulk aqueous Figure 5.1 Schematic representation of interfacial zone When a microporous membrane is introduced between two phases, then the position of the interface depends on the nature of the membrane as shown in Figure 4.11, Chapter 4. Dispersion free solvent extraction, where the two phases contact each other in the pores of a microporous membrane has been extensively studied by Sirkar, Prasad and co-workers (Kiani 1984), (Prasad 1988), (Prasad 1986), (Prasad 1988), (Prasad 1987). They have demonstrated that when the membrane is hydrophobic, and the aqueous phase pressure does not exceed a critical value, the pores are preferentially wetted by the solvent and the interface is on the aqueous side of the membrane. Conversely, a hydrophilic membrane is preferentially wetted by the aqueous phase and therefore the interface is on the solvent side of the membrane. 5. 2. 2 Extraction Regimes Since the kinetics of metal extraction are a function of both mass transfer and chemical reaction, it is important to determine which is the controlling process, 118 Chapters The role of reaction kinetics in membrane assisted solvent extraction under a given set of experimental conditions. On the one extreme, when the chemical reactions are sufficiently fast, the rate of the extraction process is determined by diffusion in the two stagnant diffusion films and the system is then said to be operating in a "diffusion regime". On the other hand, if diffusional processes are fast relative to the rates of chemical reactions, the system is said to be operating under a "chemical regime". When neither process can be ignored with respect to the other, then the system is operating under a "mixed regime", which is complicated to describe mathematically. The aim of many kinetic studies on metal solvent extraction is to elucidate the mechanism of the chemical reaction. The mechanism of the reaction cannot be determined directly, since the concentrations of the reacting species at the interface are not known. A common approach is to obtain experimental data under a kinetic regime and postulate a mechanism which is consistent with the experimentally determined rate equation. Unfortunately, it is often very difficult to unambiguously operate under a chemical regime, and therefore obtain an "intrinsic" rate equation, with no contribution from diffusional processes. Moreover, different mechanistic models will, under certain experimental conditions, lead to similar rate equations (Mendes-Tatsis 1986). The pitfalls and difficulties associated with kinetic measurements have been thoroughly discussed by (Danesi 1980). Guidelines for experimental procedures and methods that help to discriminate between various kinetic models have also been provided by other researchers (Hughes 1984), (Harada 1986) and (lrabien 1990). 5.2.3 Experimental Techniques Several experimental techniques have been used to measure metal extraction kinetics and the most common are discussed individually in the following paragraphs. They differ in the efficiency of stirring of the two phases and in the hydrodynamic conditions close to the interface. 5.2.3.1 Stirred vessels with constant interfacial area In this technique, the aqueous and the solvent phases are agitated independently so as not to perturb the horizontal interfacial area, which remains constant. The interfacial area is determined by the geometry of the apparatus. The original design was used by (Lewis 1954) and this type of cell is often named the Lewis cell. In earlier works, the reacting species concentrations 119 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction were determined from samples periodically removed from the reaction cell, leading to volume changes (Vandegrift 1977). Continuous monitoring of aqueous and organic concentrations is therefore preferred, and can be achieved by using optical (Aparicio 1989), radiometric (Saad 1998) or electrochemical techniques. Many variations have been introduced, such as different stirrer and baffle designs to improve internal hydrodynamics (Chen 1994). Other modifications include the placement of meshes close to the interface to allow operation at relatively high Reynolds numbers without disturbance to the interface (Nitsch 1978). The Lewis type cell offers some advantages including known interfacial area and the ability to vary the stirring speeds of both phases over a wide range and independently of each other. It is therefore possible to obtain information about the type of regime under which the system is being operated. The major drawback is that the diffusion films close to the interface are not well defined and are equipment specific. Some researchers have calibrated their apparatus by studying the rate of mass transfer of an inert substance having a known diffusivity (Tarasov 1976), (Lim 1996). This has enabled them to determine the thickness of the diffusion layers as a function of stirrer speed, and to compare the mass transfer coefficients of the species of interest with that of the calibrating substance. Another restriction of the Lewis cell is that it can only be operated under conditions where the agitation of the bulk phases does not disturb the interface. In the case of very fast chemical reactions, the diffusion layer thickness at the limiting stirrer speeds may not be thin enough to ensure that the system is under chemical control (Hughes 1984). This technique is also not applicable to very slow reactions because of the limited surface area available for extraction. Nevertheless, Lewis cells have been widely used, and provided the limitations are taken into account, are very useful in the study of metal extraction kinetics. Recently, Juang and Lo (Juang 1994) introduced a membrane permeation cell for the study of kinetics of solvent extraction. The main difference between this cell and the Lewis type cell is the presence of a microporous hydrophobic membrane at the interface, which contributes to better defined hydrodynamics. The hydrodynamics near the aqueous organic interface were characterized by the study of diffusion of acetic acid and iodine, enabling the thickness of the diffusion layers to be determined. The authors used the Levich proportionality between mass transfer coefficients, diffusivity and kinematic viscosity 120 Chapters The role of reaction kinetics in membrane assisted solvent extraction (Equation 5.1) (Levich 1962) to estimate the mass transfer coefficients of the reacting species. This enabled them to estimate concentrations close to the interface and obtain an intrinsic rate equation, free from the effects of diffusion. This technique has the advantage of being relatively simple to use. With very fast chemical reactions however, resistance due to diffusion in the membrane pores would be expected to control the overall mass transfer, increasing the errors associated with the estimation of interfacial concentrations. k or k oc D 213 y-116 5.1 a " 5.2.3.2 Highly agitated vessels In this type of reaction vessel, the two phases are intimately mixed, with very small droplets of one phase being dispersed in the other phase. The reaction vessel can be a simple shaking separating funnel. In order for this technique to be effective, however, the reaction must be either slow enough, such as the extraction of Al3+ with DEHPA (Sato 1978), or a complexing agent is added to the aqueous phase to increase the reaction time (Ma 1991). More commonly, complex experimental systems are used such as the AKUFVE apparatus, where a stream of the biphasic mixture is continuously centrifuged and the concentration of the reacting species are also monitored continuously (Rydberg 1973). Phase selective membrane filters have also been used very effectively to continuously sample the biphasic mixture in a very rapidly stirrer tank (Amankwa 1989) (Persaud 1987). This technique, in its most sophisticated form, can measure rates of reactions with half lives as short as ten seconds (Danesi 1980). The main disadvantage is that the hydrodynamic conditions are not well defined and it is difficult to control the interfacial area through which mass transfer takes place. 5.2.3.3 Moving drops This technique measures the mass transfer that takes place when a single drop travels along a tube filled with another phase. When the tube is filled with an organic solvent, the aqueous drop will travel down the tube (falling drop method), and when the tube is filled with an aqueous phase, the solvent drop will rise to the top of the tube (rising drop method). This technique allows for control of drop size and therefore interfacial area and the residence time can also be accurately determined. However, with very fast reactions, much of the 121 Chapters The role of reaction kinetics in membrane assisted solvent extraction extraction can take place during drop formation which maybe difficult to account for in the interpretation of results. Another disadvantage is that with very slow reactions, very long columns are required to provide sufficient residence time. 5.2.3.4 Rotating diffusion cell The rotating diffusion cell was introduced by Albery for the study of kinetics of copper extraction (Albery 1981). In this type of cell, a cylinder filled with the solvent is dipped into an aqueous phase. The bottom of the cylinder contains a hydrophobic membrane. The solvent preferentially wets the pores of the membrane. The interface is therefore at the aqueous side of the membrane surface. The cylinder can be rotated at varying speeds. The assumption with this type of cell is that the hydrodynamics are very well characterized. The rotating cylinder creates a laminar flow field that can be described mathematically and the thickness of the diffusion layer can be calculated using the Levich Equation 5.2. The main disadvantage is that the membrane introduces another resistance, and some of the membranes commonly used are themselves not that well characterised with respect to tortuosity, porosity and membrane thickness (Lazarova 1995). This problem has been addressed by Meng and co-workers, who used a 50 µm thick metallic membrane and calibrated the apparatus by studying the diffusion of potassium chloride through the membrane (Meng 1996). Another point to be born in mind is that, according to the opinion of some researchers, no studies have conclusively shown that it is valid to describe the hydrodynamics inside the cylinder as those of a perfect rotating disk. Further, the assumption that the flows inside and outside the cylinder are the same has also not been proven (Simonin 1991 ), (Danesi 1992). b'=0.643w-112D113y it6 5.2 5. 2. 3. 5 Hollow fiber membrane extractor Yoshizuka et al. introduced the use of a single hollow fiber membrane placed in a glass tube for the kinetic study of metal extraction. The aqueous solution is fed to the lumen of the fiber and the solvent is run co-currently on the shell side. A hydrophobic membrane of 445 µm thickness was used. The team has applied this technique to study of extraction and stripping of copper, zinc and the rare earths holmium and yttrium (Yoshizuka 1986), (Yoshizuka 1986), (Sato 122 Chapters The role of reaction kinetics in membrane assisted solvent extraction 1988), (Sato 1989), (Yoshizuka 1992). The experimental results have been explained with a diffusion model for metal extraction with interfacial reaction. In the case of copper extraction and stripping with 2-ethylhexyl-phosphonic acid mono-2-ethylhexyl ester, good agreement was found between results obtained with a traditional stirred transfer cell and the membrane extractor. This technique is particularly useful in understanding the mechanism governing extraction and back extraction in hollow fibers, with potential application in membrane assisted solvent extraction or supported liquid membrane technology. However, the mathematical treatment of the data is complicated. 5.2.4 Kinetic studies with di-2-ethy/hexyl phosphoric acid The extraction kinetics of metal ions with di-2-ethylhexyl phosphoric acid (DEHPA) has been extensively studied due to the importance of this reagent in hydrometallurgical and analytical applications. A summary of the literature available for the non-ferrous metals is shown in Table 5.1. Coleman and Roddy (Coleman 1971) reviewed works published in this area prior to 1970 and Danesi and Chiaizia reviewed papers up to 1980 (Danesi 1980). No studies on the kinetics of Ce4+ extraction have been found in the literature. The solvent extraction of metal ions, with acidic extractants such as DEHPA, involves some form of ligand substitution reaction. The rate at which these reactions occur should depend on the nature of both the ligand and the metal ion involved. Metal ions or complexes that readily exchange ligands or water molecules are termed labile, while inert complex ions only exchange their ligands very slowly. In homogeneous liquid phase reactions it has been found that the relative rates at which different aqua ions exchange water molecules with solvent water is of primary importance in two significant ways (Cotton 1980): • For a given metal ion the complex formation rates are usually about a factor of 1O lower than those for water exchange. The water exchange rates themselves are dependent on the metal ion and span about 10 orders of magnitude; • For a given metal ion the rates show little dependence on the identity of the ligand. 123 Chapters The role of reaction kinetics in membrane assisted solvent extraction TABLE 5.1 Summary of literature on the kinetics of metal extraction with DEHPA Metal Aqueous Method Suggested Mechanism year Reference cation media 2 Ca + (N03) Lewis Diffusion with chemical 1977 (Vandegrift 1977) ,_, ______reaction: ------Cu2+ acetate Lewis lnterfacial reaction 1983 (Komasawa 1983) Cu2+ 1988 (Ihm 1988) Cu2+ 1990 (Miyake 1990) Cu2+ ,_, ______(S04) 1993 (Juang 1993) ------2 ------· ------Zn + (N03) Lewis lnterfacial reaction 1983 (Cianetti 1983) Zn2+ (S04) Lewis lnterfacial reaction 1983 (Ajawin 1983) Zn2+ (S04) Lewis Film diffusion & interfacial 1985 (Ajawin 1985) reaction Zn2+ 1M Lewis ITSCR model: interfacial two 1989 (Aparicio 1989) (CI04) step consecutive reaction Zn2+ (CI04) RDC Kinetics too fast: limited by 1989 (Dreisinger 1989) mass transfer Zn2+ (S04) Lewis Diffusion of organic complex is 1990 (Svendsen 1990) rate determining ------Eu3+ ~------1 M (Cl) Lewis Both interfacial reaction and 1981 (Danesi 1981) diffusion coupled with fast chemical reaction models fit the data Ce3+, Pr3+, 0.2 M Lewis No mechanism suggested 1986 (Mori 1986) Nd3+ (Cl) y3+ Acetate- Shake lnterfacial reaction 1991 (Ma 1991) EDTA flasks Ho3+, y3+ 0.1 M HF Diffusion & interfacial reaction. 1992 (Yoshizuka 1992) (N03) Eu3+ (P04) Lewis lnterfacial reaction 1995 (Daoud 1995) Ce3+, Eu3+, Cl Lewis lnterfacial reaction 1996 (Lim 1996) Gd3+, Tb+, y3+ ------Co2+ acetate Lewis lnterfacial reaction 1983 (Komasawa 1983) 2 Co + (N03) Lewis lnterfacial reaction 1983 (Cianetti 1983) Co2+ (S04) Rising Mass transfer with chemical 1985 (Teramoto 1989) drop reaction Co2+ (S04) Lewis Diffusion and chemical 1985 (Golding 1985) reaction Co2+ (C104) RDC MTWCR model - mass 1989 (Dreisinger 1989) transfer with chemical reaction in aqueous film layer 124 Chapters The role of reaction kinetics in membrane assisted solvent extraction TABLE 5.1 (CONT.) Summary of literature on the kinetics of metal extraction with DEHPA Metal Aqueous Method Suggested Mechanism year Reference cation media uo/+ (N03) Lewis lnterfacial reaction 1988 (Huang 1988) ------2 ------Ni + (N03) Lewis lnterfacial reaction 1983 (Cianetti 1983) Ni2+ acetate Lewis lnterfacial reaction 1983 (Komasawa 1983) Ni2+ (S04) Lewis Diffusion and chemical 1985 (Golding 1985) reaction ------vo2+ 0.5 M Shake Chemical reaction in aqueous 1980 (Islam 1980) (S04) flasks film layer vo2+ 0.5 M Stirred Chemical reaction in aqueous 1990 (lpinmoroti 1990) (S04) vessel film layer vo2+ 0.5 M RDC MTWCR model - mass 1991 (Hughes 1991) (S04) transfer with chemical reaction in aqueous film layer vo2+ 0.5M MPC* lnterfacial reaction 1994 (Juang 1994) (S04) I'------3 ------Al + H2S04 Shake lnterfacial reaction 1978 (Sato 1978) flasks *MPC - Membrane permeation cell HF - Hollow fiber One would thus expect metal ions with relative low water exchange constants such as vo2+, Fe3+ or Al3+ to have lower extraction rates than Ln 3+ and Zn 2+, which have relatively high water exchange constants. This is indeed the case for extraction with DEHPA. However, such comparisons do not necessarily hold when other slow processes such as adsorption-desorption of chemically reactive species at the interface are the rate determining steps in the solvent extraction process (Danesi 1992). The place where the chemical reaction takes place is therefore very important but there seems to be no general agreement as to where the chemical reaction occurs. A summary of the mechanistic studies undertaken with DEHPA for zn2+, Cu2+, Co2+, Ln 3+ and vo2+ is shown Table 5.1. In most studies interfacial reactions play an important role, but some researchers have explained their results in terms of chemical reactions taking place in the aqueous phase. Although many different techniques have been used for the kinetic studies, there can be good agreement on the form of the empirical rate expression 125 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction obtained. An example is the extraction of vo 2+ with DEHPA which has been studied using separating funnels (Islam 1980), a stirred vessel (lpinmoroti 1990), a rotating diffusion cell (Hughes 1991) and a membrane permeation cell (Juang 1994). For the first three methods, the forward rate expression was of the form shown in Equation 5.3. 5.3 The two studies that presented an expression for the back extraction were in agreement and the back rate expression is shown in Equation 5.4 (Islam 1980), (Juang 1994). [V02+][H+] R =k -===-- 5.4 r r [(HR)z]1'2 With the membrane permeation technique, a similar expression to Equation 5.3 was obtained, with the exception of the hydrogen ion dependence, which was found to be inversely proportional to the half power. According to the authors, the reason for the discrepancy was the particular concentration conditions chosen for the experimental study. Despite the similarity in their rate expressions, widely varying mechanistic assumptions have been used to explain the data. (Juang 1994) postulated that both the interfacial dissociated and undissociated extractant reactions with the aqueous vo2+ metal ion complex are the rate determining steps (Equations 5.5 & 5.6). The other authors postulated that the rate determining step was the reaction between vo2+ and the dissociated extractant in the aqueous phase (Equation 5.7). vo2+ + HR. <=> VOR+ + n+ I I 5.5 vo2+ + R. <=> VOR+ I I 5.6 vo2+ + R <=> VOR+ 5.7 No kinetic studies have been found on the extraction of Ce4+ with DEHPA or other acidic commercial solvent extraction reagents, but there are a few publications dealing with the extraction of trivalent rare earths (Ln 3+). (Danesi 126 Chapters The role of reaction kinetics in membrane assisted solvent extraction 1981) applied the interfacial two-step consecutive reaction (ITSCR) model to explain the kinetic data obtained in a Lewis type cell, for Eu3+ extraction into DEHPA-dodecane from NaCI-HCI media. The model originally developed from data for DEHPA extraction of Ca 2+, essentially consists of the following steps (Vandegrift 1977): a) Formation of an interfacial complex between the interfacially adsorbed molecules of the extractant k~ Eu 3++2(HR); ;,>(EuR;); +2H+ -I 5.8 5.9 - k; ____ (EuR3 ); +3(HR)2 f EuR3 (HR) 3 +3(HR); 5.10 where k: Mass transfer coefficients of the form shown in Equations (5.11-5.12) can be derived by defining the slow steps of the interfacial reactions as per Equations (5.8 and 5.10). 5.11 A[H+]3 K =------5.12 3 3 0 B[(HR)2 ] +[H+] The authors also used another model to analyze the experimental data. This model, developed by (Danesi 1980) assumes that diffusion through the two stagnant layers adjacent to the interface is rate controlling and it is coupled with a fast chemical reaction. They found that the mass transfer coefficients for Eu3+ could be described by Equations 5.13 and 5.14, which have the same functional dependance as Equations 5.11 and 5.12. 127 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction K (ka)[(HR)2J3 = 5.13 a [(HR)2J3 +~[H+]3 Kdko K = (ko)[H+]3 5.14 o KAko/ka)[(HR)2J3 +[H+]3 Data were also obtained by (Lim 1996) with a Lewis cell for the extraction of Ce3+, Eu 3+, Gd 3+, Tb3+ and y 3+ from chloride media. The initial forward rate for all five elements was described by Equation 5.15, which is similar to Equations 5.11 and 5.13, except for the power dependence on [(HR) 2 ]. The ITSCR model was also used to explain the results (Lim 1996). 5.15 (Yoshizuka 1992) used a hollow fiber technique to study the kinetics of extraction of Ho3+ and y 3+ from nitrate solution with DEHPA. They concluded that over a very wide range of concentrations, the kinetics could be explained by a diffusion model accompanied by an interfacial reaction. Their experimental results, in the ranges of low permeability, showed apparent orders for the permeabilities of extraction, of 2, 1 and 2 with respect to pH, [Ln 3+] and [(HR) 2 ], respectively. These are different from those observed by (Danesi 1980) and (Lim 1996). The above discussion highlights the fact that there is no general agreement on the mechanism of extraction for lanthanides with DEHPA. The apparent orders of reaction vary depending on the range of concentrations studied, and even with similar concentration ranges there are some differences in the experimental results reported. Mechanisms of extraction are difficult to prove and often more than one mechanism will adequately explain the experimental data. From the literature available, it is not possible to predict which are the major parameters controlling the mass transfer of cerium(IV) in a membrane assisted solvent extraction process. 128 Chapters The role of reaction kinetics in membrane assisted solvent extraction The work in this chapter aims to study the mass transfer rate of the cerium(IV)/sulfate/DEHPA system through microporous membranes, quantifying both diffusion and chemical rate contributions to mass transfer. 5.3 Results and discussion 5.3. 1 Choice of experimental technique As discussed in Section 5.2.3 of this chapter, three experimental techniques used for the kinetic study of metal solvent extraction incorporate the use of a membrane. The experimental apparatus includes the membrane permeation cell, the rotating diffusion cell and the single microporous hollow fiber membrane extractor. Hollow fibers are the most appropriate extractors for membrane assisted solvent extraction from the processing point of view. However, the single hollow fiber technique was not used in the current study because of the complicated mathematical treatment required to separate the effects due to diffusion and chemical reaction (Yoshizuka 1986), (Yoshizuka 1986), (Yoshizuka 1992). Both the rotating diffusion cell and the membrane permeation cell technique are appropriate for such a study, and the latter was selected on the basis of the simplicity of both the experimental set-up and the mathematical treatment. Permeation data for the cerium system presented in Chapter 4, showed that extraction is faster with a hydrophobic membrane and back extraction is favoured by a hydrophilic membrane. Accordingly, the hydrophobic Millipore GVLP membrane was chosen for extraction and the hydrophilic Millipore WLP for back extraction. Both membranes have been characterized in Chapter 4, Section 4.3.1. Details of the membrane permeation cell apparatus and the experimental procedure used are presented in Chapter 2, Section 2.6. 5.3.2 Hydrodynamic characteristics of membrane permeation cell The hydrodynamic characteristics of the membrane permeation cell were determined by studying the mass transfer rates of iodine from water to n heptane, and acetone from n-heptane to water. Since no chemical reaction is involved in the mass transfer of these substances, and they do not exhibit strong interfacial activity, the process can be described by considering only diffusion processes (Prasad 1992). 129 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction The calibration of the membrane permeation cell has two purposes. The first is to minimise the effect of the diffusional resistance of both the aqueous and organic layers adjacent to the membrane. This is done by optimisation of the stirring speed of the bulk solutions, which affects the thickness of the stagnant films. The second purpose is to quantify the individual aqueous and organic film and membrane mass transfer coefficients, for the calibrating substances, thus providing a means of quantifying those parameters for the system being studied. The concentration profiles for iodine transfer from aqueous to organic through a microporous hydrophobic membrane are represented in Figure 5.2. The first resistance is in the aqueous layer adjacent to the membrane. This aqueous film has a thickness of 8a. With a hydrophobic membrane, the organic solution wets the pores of the membrane and therefore the interface is on the aqueous side of the membrane. Once iodine has been delivered to the interface, it partitions according to its distribution coefficient Kd. This is assumed to occur instantaneously. In the organic phase, iodine diffuses through the membrane and the organic diffusion film layer of thickness 80 • No discontinuity in the organic iodine concentration at the organic side of the membrane has been assumed, since the pore size of the membrane (0.22 µm), compared to the molecular size if iodine (2.7A0 ), is sufficiently large not to cause hindered diffusion. The mass transfer rate or flux is defined as 5.16 5.17 where K = [I] and [J]x is the hypothetical aqueous iodine d UL concentration at equilibrium with the organic iodine concentration [I]. when Kd >> 1, the flux close tot= 0, N 1 =KJl] 0 5.18 130 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction Hydrophobic Aqueous Membrane Organic ~ .. ----~ Wm,ai [I] [IJi a/o interface+ Figure 5.2 Concentration profiles for diffusion of iodine from an aqueous to a solvent phase, through a hydrophobic membrane. • 300 rpm 0200 rpm x 120 rpm +82 rpm 040 rpm I 0.35 ~------~ 0.30 0.25 "a_ 0.20 e - 0.15 B 0.10 30 40 50 60 70 Time / minutes Figure 5.3 Mass transfer of iodine from water to n-heptane. Iodine concentration in n-heptane measured as a function of time and aqueous stirrer speed. [l]o = 0.98 mol m-3, organic stirrer speed = 200 rpm. 131 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction From Figure 5.2 the flux at steady state can also be described as: 5.19 = k,,, ([ /] 111 ,0 ; - [ f],,,, 0 ;) 5.20 = k0 (t f],,,, 0 ; - [ /]) 5.21 From Equations (5.17 and 5.19-5.21), the following relation can be derived [/] + ([/]; -[J]; _ [J]m,oi + [J],,,,o; J_ [/] 1 = __:______Kd Kd_____:_:___ Kd [/]m,ai [/],,,,o; [Iln,,o; [/] ------1 Kd Kd Kd Kd =-+ + ka NI NI 5.22 Thus the overall resistance to mass transfer can be described in terms of the individual resistance due to diffusion through the membrane and aqueous and organic layers adjacent to the interface (Kiani 1984). The mass transfer of iodine from water to heptane was measured at various stirrer speeds in a mass transfer permeation cell as described in Chapter 2, Section 2.6. The concentration of iodine in the organic phase was monitored as a function of time. Linear relationships were found between [/] and t in the first 30 minutes of the reaction. Typical plots are shown in Figure 5.3. The initial rate of extraction was calculated from Equation 5.16. This technique, known as the method of initial slopes, is commonly used in solvent extraction kinetic studies (Chen 1994), (Aparicio 1989) and (Komasawa 1983). The method is valid at the initial stages of the reaction, when the reverse reaction can be ignored. 132 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction The effect of increasing the bulk aqueous stirring speed on the overall aqueous mass transfer is shown in Figure 5.4. The overall aqueous mass transfer coefficient increases with increasing stirrer speed until it reaches a plateau at 150 rpm. This type of curve has been commonly obtained for Lewis type cells and the plateau region ascribed to a decreased effect of aqueous film layer diffusion, relative to chemical reaction (Hughes 1984). In the case of a membrane permeation cell, the plateau is an indication that aqueous diffusion is minimised with respect to membrane diffusion and or chemical reaction. The measured Ka values obtained by (Komasawa 1980), also shown in Figure 5.4, for mass transfer of iodine from water to n-heptane, using a Lewis type cell, follow a similar trend to those obtained in this work. The absolute values, however, are approximately 50% higher, presumably due to different hydrodynamic conditions in the two cells. Additional mass transfer data were obtained for the chemical system toluene/ water/ heptane using the same membrane permeation cell as that used for the iodine experiments. In this instance, however, the membrane was eliminated thus effectively turning the permeation cell into a Lewis type cell. The overall mass transfer coefficient measured as a function of aqueous stirrer speed is shown in Figure 5.5. It was found that at low stirrer speed (50 rpm), the same Ka value was obtained for toluene in the Lewis cell as for iodine in the membrane permeation cell. At the higher stirrer speed, the Ka values for these two systems only differed by less than 20%. The effect of changing the organic stirrer speed was also investigated (Figure 5.5). It can be seen that while the aqueous stirring speed has a marked effect on Ka, it was found that changing the organic stirrer speed (range 10 - 150 rpm) had no effect on Ka. These factors indicate that the major resistance to mass transfer in both membrane permeation with a hydrophobic membrane and the Lewis cell containing no memebrane, is in the diffusion through the aqueous diffusion layer. For conditions where Kd >> 1 and the stirrer speed is low, Equation 5.22 approximates to Equation 5.23. The distribution coefficients for iodine and toluene between water and heptane are 43 (Juang 1994) and 20.4 (measured in this work), respectively. 133 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction o Komasawa et al. (Lewis cell) • Membrane Permeation Cell 1E-04 T - I ~ ~---~+--~~~~~---~--~---+--t~~ ~,E-05 @_-~~-~---~--I+--~-~:-~--+~l=_i+=:i+s_;,~:[-+[lJ.t!t/~t~D~--_§·'-·~~~·t~t~~·_g- -li~tt~...... I 1E-06 .____ .....__....._ ...... _,___._...... _...... _ ____. _ __.__.,___.,___l...... l-...J.-J.....1 I 10 100 1000 aqueous stirring speed L ~~~~~~~______J Figure 5.4 Mass transfer coefficients of iodine from water to n-heptane measured as a function of aqueous stirrer speed. This work: Iodine (1 mol m-3) transfer measured in membrane permeation cell (hydrophobic membrane). Data reproduced from Komasawa for iodine (0.7 mol m-3) transfer measured in Lewis type cell. o Constant organic rpm eConstant aqueous rpm __ ] 1E-04 1---+---+-----l- -+----+--+--+-----+---+----+---+-·+-+--+-+--l '111 ------~·--+----+----+--+--+-+--+--+----+ .§. • 1 E-06 L...-__.....,__.,___ ...... _._ ...... __ ___. _ ___.._.,___..___L.....,I,...... 10 100 1000 aq/org stirring speed Figure 5.5 Mass transfer coefficients of toluene from water to n-heptane in a Lewis type cell. Dependence of Ka on aqueous stirrer speed (at constant 150 rpm organic stirrer speed), and on organic stirrer speed (at constant 150 rpm aqueous stirrer speed). 134 Chapters The role of reaction kinetics in membrane assisted solvent extraction 1 1 5.23 To optimise the organic phase stirring conditions, the mass transfer of acetone from n-heptane to water was also studied. The mass transfer can be represented by the concentration profiles shown in Figure 5.6. The mass transfer rate or flux is defined as 5.24 5.25 where Kd = [AcL and [AcL is the hypothetical organic acetone concentration [Ac] at equilibrium, where the aqueous acetone concentration is [Ac]. when Kd << 1, the flux close to t = O , 5.26 From Figure 5.5, and following the methodology established by (Kiani 1984), the flux at steady state can also be described as: NAc = k)[Ac]; -[Ac]) 5.27 5.28 5.29 From Equations (5.25) and (5.27-29), it can readily be shown that 5.30 when Kd << 1, then, 5.31 135 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction Hydrophobic Aqueous ..Membrane .. Organic I Oa Orn ... Oo ..I I [Ac]i I [Ac] [Ac] [Ac]m,ai a/o interface+ Figure 5.6 Concentration profiles for diffusion of acetone from a solvent to an aqueous phase, through a hydrophobic membrane. oLewis-Cell )KT.M.Lim -'u, .§. V 0 ::K ~1E-05 *I ~~~--§§~~~~~±=!-~_§§-,-----,-, I 1E~1_0______10_0 ______- __ ~ I organic stirring speed ~ Figure 5.7 Mass transfer coefficients of acetone from n-heptane to water, measured as a function of organic stirrer speed. This work: acetone (7 mol m-3) transfer measured in membrane permeation cell (hydrophobic membrane) and Lewis type cell. Data reproduced from Lim for acetone (6 mol m-3) also using a Lewis type cell. 136 Chapters The role of reaction kinetics in membrane assisted solvent extraction The mass transfer of acetone from n-heptane to an aqueous phase was measured in both a Lewis cell and the membrane permeation cell containing a hydrophobic membrane. The concentration of acetone in the aqueous phase was measured as a function of time. Details of the experimental procedure and analytical techniques used are presented in Chapter 2, Section 2.6. The initial flux was calculated according to Equation (5.24) by the method of initial slope, under conditions where the reverse reaction can be ignored. The overall organic mass transfer coefficient was calculated according to Equation (5.26). Results are shown in Figure 5. 7. In the case of the membrane permeation cell, the overall organic mass transfer constant is not very sensitive to the organic phase stirring speed for values ranging from 100 - 400 rpm. The K0 values measured for the membrane permeation cell were lower than those measured for the Lewis cell, as expected from Equation (5.31 ). Lewis cell data obtained by (Lim 1996), for the transfer of acetone from n-heptane to water (Figure 5.6), agrees reasonably well with the data presented in the current work. In conclusion, it has been shown that in order to minimise the effect of film diffusion in both phases, a stirring speed above 150 rpm is adequate for the experimental set-up used in this work. All further experiments were therefore undertaken at a stirring speed of 200 rpm for both the aqueous and organic phases. From the study of mass transfer of iodine and toluene from water to n-heptane with both a Lewis cell and a membrane permeation cell containing a hydrophobic membrane, it has been demonstrated that: a) where no chemical reaction is involved, the main resistance to extraction (ie Kd >>1 ), lies in the diffusion through the aqueous diffusion layer b) the use of a hydrophobic membrane of 125 µm thickness as a phase separator did not affect the overall resistance to mass transfer at low stirrer speed. The study of mass transfer of acetone from n-heptane to water with both a Lewis cell and a membrane permeation cell containing a hydrophobic membrane, demonstrated that: 137 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction c) where no chemical reaction is involved, the overall mass transfer of back-extraction (ie. Kd <<1 ), is governed by both diffusion through the organic diffusion layer adjacent to the interface and the diffusion through the hydrophobic membrane itself. d) the use of a hydrophobic membrane in this case greatly increases the overall resistance to back-extraction. 5.3.3 Calibration of permeation cell with microporous hydrophobic Millipore GVHP membrane The above discussion has made no mention of how different types of membranes, with varying characteristics such as porosity, tortuosity and membrane thickness affect the overall mass transfer. The current section deals with these issues. Diffusional resistance within a flat membrane can be described by Equation 5.32. This relation holds provided there is unhindered diffusion of the solute and the membrane is symmetric and completely wetted by the solvent phase (Prasad 1992). Unhindered diffusion is considered to be taking place if the solute dimensions are at least two orders of magnitude smaller than the pore dimensions (Prasad 1988). k = DAB 6 5.32 Ill s: u 111 r where c and 8m are known membrane parameters of porosity and membrane thickness and r is the unknown tortuosity factor. DAs is the diffusivity of a particular species A in solvent B and various correlations are available to estimate this value. In order to determine the tortuosity parameter for the Millipore GVHP hydrophobic membrane, the flux through the membrane of the cerium-DEHPA complex in heptane was measured. The solvent n-heptane was used for both the feed and the receiver solution and therefore in these experiments there was no interface. The flux can be described by Equation 5.33, taking into account diffusion through the two organic diffusion layers, adjacent to the membrane. 138 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction 5.33 1 2 1 where -----=----+----- Ko(CeR4 -heptane) ko(CeR4 -heptane) km(CeR4 -heptane) The plot of flux as a function of Ce concentration in the feed solution is shown in Figure 5.8. The slope of the curve was measured as 1.44 ± 0.05 E-06 m s· 1• The porosity and membrane thickness for Millipore GVHP are 0.75 and 1.25E- 04 m respectively. The value of kO(CeR,-heptane) however is not known and therefore further information is required to estimate this value in order that rcan be calculated. The mass transfer data for iodine and acetone at 200 rpm presented in the previous section can be used for this purpose as follows. The transfer of iodine from water to n-heptane is given by: 1 1 1 1 -----=---+------+------Ka( I - water I heptane) k a(!- water) K d(!- water I heptane )km(!-hep tan e) K d(!- water I heptane ik o(!-hep tan e) 5.34 and the transfer of acetone from n-heptane to water is given by: ___1___ = 1 + 1 + Kd(Ac-water/heptane) 5.35 Ko(Ac-heptanelwater) ko(Ac-heptane) km(Ac-heptane) ka(Ac-water) The three Equations 5.33 - 5.35 have r in common incorporated in their K K respective km terms. The values of o( CeR 4 -heptane) ' a( I - water I heptane) and K o(Ac-heptane I water) were experimentally determined. All the ka and k0 terms are related through the Levich relationship (Levich 1962) described below (Equation 5.36). The Levich relationship holds provided the hydrodynamic conditions and temperature in the cell remain unchanged. where y is the kinematic viscosity 5.36 Once the kinematic viscosity and diffusivity is calculated for each system, Equations 5.33-5.35 can be solved simultaneously for the three unknowns. 139 Chapters The role of reaction kinetics in membrane assisted solvent extraction 9E-06 --' ~ • ~e 7E-06 -eQ SE-06 -=~ 3E-06 •• -~ IE-06 0 2 3 4 5 6 [Ce] (mol m·3 ) Figure 5.8 Flux of Ce in DEHPA/n-heptane solution through a Millipore GVHP membrane. [DEHPA] = 186 mol m·3 220 organic 200 180 160 140 &, 120 Q. l ~ 100 8, 80 60 40 aqueous 20 I 0 0 50 100 150 200 250 300 350 400 450 500 l Time (minutes) Figure 5.9 Extraction of Ce in membrane permeation cell 3 3 3 193 mol m· DEHPA, 15 mol m· MEHPA, 550 mol m· H2S04 140 Chapters The role of reaction kinetics in membrane assisted solvent extraction This method has been used by (Juang 1994) in the calibration of a membrane permeation cell. Diffusivities for iodine and acetone in n-heptane and water (Table 5.2) were estimated by the Hayduk and Minhas correlation, which is recommended by (Reid 1987) as yielding the lowest errors. Molar volumes at the normal boiling point were calculated by the Tyn & Caius method (Reid 1987). The following values were used for the equilibrium distribution coefficient Kd(t-water/heptaneJ = 43 (Juang 1994) and Kd(Ac-waterlheptane) = 0.1. TABLE 5.2 Diffusivity DAe at 22°c calculated by Hayduk-Minhas Correlation Diffusivity Molar Volume of solute Solvent Solvent m2s-1 cm3 mo1·1 D20 Viscosity cP 4 I-heptane 5.69E-09 56.3 0.6837 0.400 I-water 1.28E-09 56.3 0.9978 0.955 Ac-heptane 4.76E-09 77.0 0.6837 0.400 Ac-water 1.09E-09 77.0 0.9978 0.955 The mass transfer coefficients calculated by this method are shown in Table 5.3. A number of 2.36 was obtained for the tortuosity of the membrane GVHP. This value agrees well with that obtained by (Juang 1994) (r= 2.2) for the same GVHP membrane, and values of r = 2 - 2.8, obtained by other workers for different hydrophobic membranes (lmato 1981 ), (Kiani 1984), (Prasad 1986). The numbers in Table 5.3, when substituted into equation 5.22, also show that for mass transfer of iodine, diffusion in the aqueous boundary layer accounts for 97% of the overall resistance. These results support the arguments presented in Section 5.3.2, which state that the presence of the GVHP membrane has no significant effect on the mass transfer of iodine. For acetone transfer, it was calculated that the membrane organic boundary layer and aqueous boundary layer contribute 79%, 15% and 6%, respectively, to the overall resistance. In this case the presence of the membrane lowers the flux and the membrane characteristics are expected to have a significant effect. 141 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction TABLE5.3 Mass Transfer Coefficients for Iodine and Acetone Measured in Membrane Permeation Cell using GVHP Membrane Mass Transfer Coefficients Experimental Calculated System ms-1 System ms-1 Ka (I-water/heptane) 1.70E-05 ka (I - water) 1.67E-05 km (I - heptane) 1.45E-05 ko (I - heptane) 7.39E-05 Ka (Ac-heptane/water) 8.84E-06 ka (Ac - water) 1.SOE-05 km (Ac-heptane) 1.21E-05 ko (Ac - heptane) 6.56E-05 The thickness of the stagnant films adjacent to the interface can also be calculated by assuming Whitman's two film theory is applicable. When the concentration of the diffusing components is very low compared to that of the solvents, then ka = D/oa and k0 = D/00 • At stirrer speeds of 200 rpm in both phases, the aqueous and organic diffusion layers calculate at 77 and 73 µm respectively, which is within the range of 20 -100 µm measured by other workers using Lewis cells. 5.3.4 Extraction of cerium with a membrane permeation cell The liquid-liquid extraction of cerium from a 0.55 M sulfate aqueous solution with di-2-ethylhexyl phosphoric acid in n-heptane was studied with the membrane permeation cell containing a hydrophobic Millipore GVHP membrane. The aqueous and organic phases were stirred at 200 rpm. Details of the experiments are given in Chapter 2, Section 2.6.3. The initial rate of extraction was measured by monitoring the aqueous cerium concentration as a function of time and the forward extraction rate calculated by Equation 5.37. The concentration in the organic phase was calculated by difference. Since the [CeR4 ] 0 concentration was zero, the back extraction was considered to be negligible in the initial stages of the extraction reaction. In selected experiments, the final aqueous and organic concentrations were 142 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction determined by ICP analysis and the mass balance closed within 5%. Cerium concentration profiles in both the aqueous and organic phase over a 8 hour period are shown in Figure 5.9. Only data in the linear part of the curve (ie the first 30 minutes) were used for the flux calculation. N __ Va d[Ce] 5.37 1 - A dt A schematic representation of the concentration profiles in the aqueous, membrane and solvent for all the reacting species is shown in Figure 5.10. In this system, cerium is extracted via cationic exchange reactions, as discussed in detail in Chapter 3. In this work, it has been demonstrated, that at low solvent saturation levels, (ie. HR:Ce molar ratio >10), formation of [CeR 4 ] and [CeS04 R2 ] occurs, with the former being the most abundant. For simplicity, in this treatment only the formation of [CeR 4 ] (Equation 5.38) is considered. 5.38 The flux across the membrane N 1 equals the mass transfer rate of the forward reaction minus the reverse reaction. Since the initial concentration of cerium in the solvent is zero the reverse reaction can be ignored. Under steady state conditions, the flux can be described in terms of the mass transfer rate as per Equations (5.39-41). 1 5.39 5.40 1 ( ) =-K-2 (HR),-heptane [( HR )2 ]h - [( HR )2 ],,am. 5.41 The subscript i indicates interfacial concentrations (moles per unit volume) close to the interface, that is, at the extreme limit of the diffusion layers. This is not to be confused with the interfacially adsorbed species at the interface ,which are expressed as moles per unit area. 1 The diffusion of DEHPA through the solvent diffusion layer and the membrane is described by equation 5.41. 143 Chapters The role of reaction kinetics in membrane assisted solvent extraction Hydrophobic Aqueous ..Membrane .. Organic ,. Oo .. 1 [Ce [CeRt] i a/o interface Figure 5.10 Concentration profiles for reacting species in the extraction process of Ce4+ from a 550 mol m·3 sulfate aqueous solution to a solvent phase of DEHPA in n-heptane measured in a membrane permeation cell containing a hydrophobic membrane. 144 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction 1/ K(HRiJ-heptane is the mass transfer coefficient across the membrane and organic diffusion layer and is given by: 1 1 1 ----= +----- 5.42 K- (HRJi -heptane ko((HR)i -heptane) km{(HR)i-heptane) Under steady state conditions, the flux across the membrane can also be written in terms of the forward chemical rate, which is a function of the interfacial concentrations of cerium, hydrogen ion and dimeric DEHPA. The interfacial concentrations can be obtained from the measured fluxes using Equations 5.39-5.41. Using this methodology, it is therefore possible to obtain an empirical chemical rate expression from flux measurements which has been isolated from the contributions made by the diffusion processes. 5. 3. 4. 1 Determination of chemical extraction rate equation The chemical extraction rate is dependent on the concentration of the reacting species. DEHPA is known to exhibit significant interfacial activity (Vandegrift 1980), but does not readily partition into the aqueous phase. The distribution coefficient for partition between organic and aqueous phase was determined by (Komasawa 1983) to be 1.6x 10-4, which indicates that the concentration of DEHPA in the aqueous phase is very low. These factors would suggest that metal extraction reactions with DEHPA are likely to take place at the interface or in the layers adjacent to the interface. Kinetic studies undertaken on the extraction of rare earths with DEHPA (referenced in Table 5.1 ), have highlighted the importance of interfacial reactions. However, the concentrations close to the interface are not known, and are dependent on the diffusion of the reacting species away from and toward the interface. As already mentioned in the previous section, in order to eliminate the effect of diffusional resistances on the determination of the extraction rate equation, the concentrations close to the interface have been estimated using Equations (5.39-42). Examples of such calculations are shown in Appendix 4. The values of the mass transfer coefficients used for theses calculations are shown in Table 5.4 and were estimated using Equations 5.32 and 5.36, by reference to the values obtained for iodine and acetone mass transfer. The diffusivity of the dimeric DEHPA molecule in heptane was estimated using the Hayduk-Minhas correlation and the molar volume at the normal boiling point 145 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction calculated as 798.4 cm 3 mo1·1, by the Tyn and Caius method. This value is similar to those published in the literature (9.99 E-10 (Juang 1994), 1.003 E-09 (Dreisinger 1989). The diffusivities of the hydrogen ion and the Ce-DEHPA complex were calculated from the Stokes-Einstein equation2. The molecular radius of Ce-DEHPA was estimated from its molecular volume, assuming the complex has four times the volume of monomeric DEHPA. TABLE 5.4 Calculated Mass Transfer Coefficients for Ce, H+, CeR4 and (HRh System Mass Transfer Diffusivity Ionic radius Viscosity Kinematic Coefficient solute solvent Viscosity ms·1 m2s-1 AD 10-3 N s m·2 m2 s-1 k0 (Ce-0.55 M 1.07E-05 16.70E-10 21.11 1.08E-06 sulfuric) k0 (W-0.55 M sulfuric) 4.67E-05 6.08E-09 W=0.32 k0 ((HRh-heptane) 2.33E-05 1.01E-09 20.40 5.85E-07 km ((HRh-heptane) 2.56E-06 k0 (CeR4-heptane) 1.78E-05 6.70E-10 km (CeR4-heptane) 1.70E-06 Diffusivity of Ce in sulfate estimated to be the same as that of Cu in 0.5 M sulfate (Awakura 1988) 2 (R.C.Weast (Ed.) 1981) The mass transfer rate of cerium was measured under various conditions, using the calibrated membrane permeation cell and GVHP Millipore hydrophobic membranes. The initial aqueous cerium and sulfuric acid concentrations ranged from 1.4 to 13.4 and from 110 to 880 mol m·3, respectively. MEHPA is an impurity present in both commercial and AR grade DEHPA. Repeated crystallization with copper sulfate is required to achieve higher than 99% purity (Partridge 1969). MEHPA also extracts cerium, but due to its much lower concentration when compared to DEHPA, it has a very limited effect on the overall equilibrium (for equilibrium data on cerium extraction with MEHPA see Section 4.4.2.3, Chapter 4). However, as already discussed previously MEHPA has a dramatic effect on extraction rates of cerium(IV) and was 2 The Stokes Einstein equation used to calculate hydrogen ion diffusivity assumes diffusion by Brownian motion with hydrodynamic drag. A hopping mechanism is more appropriate to describe hydrogen ion diffusion. In that case higher hydrogen diffusivity maybe calculated 9 x 10-9 compared with 6 x 10-9 (Table 5.4) assumed for these calculati~ns. Thi~ change has no significant effect with differences between [H]; and the bulk concentrations varying only by 1.1 % in both instances. 146 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction therefore treated as an independent parameter. The MEHPA concentrations ranged from 1.7 to 19 mol m-3• The change in MEHPA concentration was achieved by doping AR grade DEHPA with a mixed DEHPA/MEHPA product. The effect of each of these parameters on the cerium flux is shown in Figures 5.11-5.14. The data are presented as plots of flux versus interfacial concentrations and logarithmic scales are also shown. The interfacial concentrations of [Cet, [H+]h [(HRh]i.om and [(HRh]i,am were calculated as described earlier using the mass transfer coefficients shown in Table 5.3 and Equations 5.39-42 as shown in Appendix 4. Significant differences were found between bulk and interfacial cerium concentrations [Ce]i (average 44%) and DEHPA concentrations [(HR)2]i,am (average 13%), indicating that diffusion of cerium through the aqueous layer and diffusion of [(HR)2] through the membrane play a significant role in the overall mass transfer process. The [H+t and [(HRh]iom were found to differ from their bulk concentrations by a maximum of 3.5 % only, indicating that diffusion of hydrogen ion through the aqueous boundary layer and [(HRh] diffusion through the organic boundary layer are not major contributors to the overall resistance. An empirical chemical rate expression was then derived to describe the experimetal data in terms of interfacial concentrations. The form of the rate expression was obtained from previous studies on rare earth extraction with DEHPA (Danesi 1981). The reaction orders with respect to [Ce], [(HR)2] and [~] were determined as 1, 1 and -1 respectively, from the slopes of the experimental log-log plots. The constants were obtained by minimising the error between calculated and experimental data points. 3 Rf = {1.7 x 10- !Ce];[(HR) 2 ];,0 ,,, }[MEHPA]o.3 5.43 49.8 [H ]; + [(HR)2L, 0111 The average error for the data set spanning three orders of magnitude is 27%. The largest deviations are associated with experimental data points with high fluxes and high concentrations of MEHPA. Possible reasons for these deviations are: i) At high fluxes the it is more likely that the chemical reaction rate is not as important a parameter in the overall mass transfer process and therefore the equation 5.43 is not as relevant and error become more significant; 147 Chapters The role of reaction kinetics in membrane assisted solvent extraction ,- 1------ --'u, ":"U)- c:oE c:oE §. 1.E-05 !..E-05 >< :,>< :, u::: u::: en en .2 .2 0.01 0.10 1.00 10.00 100 1000 10000 log [H]1 (mol m·3) Figure 5.11 Effect of [Ce] Figure 5.12 Effect of [H2S04 open symbols: [DEHPA] = 160 mol m-3, closed symbols: [DEHPA] = 160 mol m-3, [MEHPA] = 18 mol m-3 , [H2S04) = 550 mol m-3 [MEHPA] = 18 mol m-3 , [H2S04) = 550 mol m-3 closed symbols: [DEHPA] = 200 mol m-3, [MEHPA] =2 mol m-3 , [H2S04) = 550 mol m-3 1.E-04 ~------~ 3 -'u, ":"; 1.E-05 - c:oE c:oE §. 1.E-05 .§. >< >< :, :, u::: u::: j 1.E-06 j 10 100 10 100 1000 log [MEHPA] (mol m-3) log [(HR)zl;.am (mol m·; Figure 5.14 Effect of [MEHPA] Figure 5.13 Effect of [DEHPA] closed symbols: [Ce]o = 1.4 mol m-3, open symbols: [Ce]o = 1.4 mol m-3 , [DEHPA] =185 mol m-3 , [H2S04) = 550 mol m-3 [MEHPA] = 10 mol m-3, [H2S04) = 550 mol m-3 closed symbols: [Ce]o = 1.4 mol m-3, [MEHPA] < 1.2 mol m-3, [H2S04) = 550 mol m-3 148 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction ii) The MEHPA concentration has a significant effect on the rate and it is probable that the rate dependency on MEHPA concentration is more complicated than indicated by Equation 5.43, thus giving rise to relatively high errors in predicted rates. The general form of the rate expression, ignoring the MEHPA component, is similar to those obtained by (Danesi 1980), (Danesi 1981) and (Lim 1996), for the extraction of trivalent rare earths with DEHPA. The power dependencies with respect to [[(HR)2 ], and [H+] differ from the above works, but this is not surprising as the interfacial reaction mechanism is not expected to be the same for cerium(IV) as for the trivalent lanthanides. No attempt has been made in this work to determine the actual chemical reaction mechanism for the reasons given below. The introductory sections of this chapter outlined the difficulties in deducing chemical reaction mechanisms from rate equations. In the present study, the situation is further complicated by the presence of MEHPA, which has been shown in Chapter 4 to be surface active and therefore compete with DEHPA for limited interfacial sites. In addition, MEHPA is also an extractant for cerium(IV). Under these circumstances, the current work is not considered to be extensive enough to enable a reaction mechanism to be proposed. Of interest here is to quantify the contributions made by the various diffusion resistances and comparing them to the contribution made by the resistance due to chemical reaction. For this purpose, the empirical rate equation obtained is useful, provided the analysis is limited to the concentration ranges experimentally tested. 5.3.4.2 Transport mechanism of membrane liquid-liquid extraction process The preceding sections have identified four main resistances as contributing to the overall resistance to mass transfer for extraction of cerium with chemical reaction from an aqueous to an organic phase, through a microporous membrane. 1. Resistance due to diffusion through aqueous stationary layer. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: 149 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction Na = ka(0.5Msulfuric)([Ce]h -[Ce];) ~ ka(0.5Msulfuric) [Ce]b 5.44 __ [CeR ]b since [Ce]; (( [Ce]h when Kd ---4 is high [Ce]; 2. Resistance due to chemical reaction. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: 5.45 3. Resistance due to diffusion through the membrane pores. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: - km(CeR, -heptane)Kd [ Ce ]b 5.46 (1 + Kd) [CeR ] where Kd = 4 ,,am [Ce]h -[CeR4Lam 4. Resistance due to diffusion through the stationary organic layer. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: _ ko(CeR4 -heptane)KACe ]h 5.47 (1 + Kd) Following the methodology used by (Juang 1998), (Juang 1997), the fractional resistance due to aqueous layer diffusion (~a), chemical reaction (~c), membrane diffusion (~m) and organic layer diffusion (~0 ), can be quantified by Equations 5.48-5.51: 11 Na ~ =-----__c______5.48 a l/ Na +l/ Ne +l/ Nm +l/ N 0 150 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction ~ =----I_I_N~e___ _ 5.49 e II Na +II Ne +II Nm +II N 0 5.50 ~ = ____I_I_N__.::. 0 ____ () II Na +II Ne +II Nm +II No 5.51 Given the availability of equilibrium data, it is therefore possible to quantitatively determine the contribution of both diffusion and chemical processes to the overall extraction process, under specific experimental conditions. The equilibrium data used for these calculations was obtained from the model for extraction developed in Chapter 3. Results are presented in Table 5.5. The data indicate that, for the experimental conditions studied, the liquid - liquid extraction process for cerium is mainly controlled by membrane diffusion. The rate of chemical reaction can contribute up to 36% to the overall resistance. The accelerating affect of MEHPA on the flux is evident in Figures 5.13-14. This has been explained in terms of MEHPA's increased interfacial activity and, at low concentration, its ability to act as a catalyst. A discussion on this matter is presented in Chapter 4. The results in Table 5.5 also show as the MEHPA concentration is increased, the relative contribution of chemical reaction to the overall mass transfer decreases. The highest distribution coefficient is calculated for the lowest acidity and this data point also has the highest calculated aqueous diffusion contribution. Conversely the lowest distribution coefficient occurs at the lowest DEHPA concentration and it is at this point where the aqueous diffusion resistance is at its lowest. On average, contributions to the overall resistance are distributed as follows: diffusion through the membrane pores contributes 62%, chemical reaction contributes 21 %, diffusion through the aqueous layer contributes 10% and diffusion through the organic layer contributes 7%. 151 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction TABLE 5.5 Fractional Resistances of Aqueous and Organic Diffusion Layers, Diffusion in Membrane Pores and Chemical Reaction for the Extraction of Cerium [DEHPA] [MEHPA] [Ce [HJ mol m-3 mol m-3 mo1m-3 mol m-3 ~a(%) ~c(o/o) ~m(o/o) ~o(o/o) 200 2 0.22 1100 10 30 54 6 200 2 0.35 1100 10 30 54 6 200 2 0.64 1100 10 30 54 6 200 2 1.4 1100 10 30 54 6 179 19 1.3 1100 12 12 69 7 179 19 2.4 1100 11 16 67 7 179 19 4.4 1100 11 20 63 6 193 13 6.5 1100 10 26 58 6 179 19 9.8 1100 10 27 57 6 ------185 15 1.4 220 17 6 70 8 185 15 1.4 440 16 11 66 7 185 15 1.4 660 15 15 64 7 185 15 1.4 1100 12 19 62 7 185 15 1.4 1760 6 17 69 8 ------5.6 2 1.4 1100 0 7 84 9 10 2 1.4 1100 0 11 80 9 20 2 1.4 1100 1 20 72 8 50 2 1.4 1100 3 33 58 6 100 2 1.4 1100 6 36 54 6 100 10 1.4 1100 7 26 60 7 140 10 1.4 1100 10 24 60 7 185 10 1.4 1100 11 21 61 7 240 10 1.4 1100 13 18 62 7 300 10 1.4 1100 14 16 63 7 ------185 1.7 1.4 1100 10 27 57 6 185 5 1.4 1100 11 19 63 6 185 10 1.4 1100 11 15 66 7 185 15 1.4 1100 12 13 68 7 185 19 1.4 1100 12 12 69 7 In liquid membrane extraction processes, both supported and emulsion membrane diffusion has been assumed by many authors to be rate controlling. One would thus expect membrane diffusion also to be rate controlling in any 152 Chapters The role of reaction kinetics in membrane assisted solvent extraction membrane extraction device such as a hollow fiber contactor. In the present work, this has been quantitatively demonstrated to be the case for cerium extraction through a 125 µm thick hydrophobic membrane. It can also be concluded from the above results that cerium extraction is a relatively slow reaction, and therefore the contribution of the chemical reaction to the overall extraction processes cannot be ignored. The importance of the chemical reaction would be accentuated if small concentrations of MEHPA were not present. In conclusion, there are two major contributors to the resistance to extraction of cerium with DEHPA/MEHPA using a well stirred membrane contactor. These are the diffusion of the extracted complex in the membrane pore and the chemical reaction rate. The membrane thickness is expected to be a major parameter determining which of these is dominant. 5.3.5 Calibration of permeation cell with microporous hydrophilic Millipore WLP membrane In this section, the membrane permeation cell was calibrated by measuring the transfer of iodine from water to n-heptane through a hydrophilic WLP Millipore membrane. This transfer can be treated in similar terms to that already described for the case where the microporous membrane is hydrophobic. The only difference is in the position of the interface. The hydrophilic membrane is wetted by the aqueous phase and therefore the interface is on the solvent side of the membrane. The concentration profiles for iodine are shown in Figure 5.15A. The overall mass transfer rate of iodine is defined as (Prasad 1987) 5.52 5.53 5.54 5.55 153 Chapters The role of reaction kinetics in membrane assisted solvent extraction A Hydrophobic B Hydrophobic Aqueous Membrane Organic Aqueous Membrane Organic 111 .. ------.. .. ------6 I ~ I [l [AJ [Ac] [Ilm,oi a/ o interface+ a/o interface+ Figure 5.15A & B Concentration profiles for diffusion of iodine from an aqueous to a solvent phase and acetone from a solvent to an aqueous phase, through a hydrophilic membrane A [aBOO ppm <>400 ppm A320 ppm x200 ppm :iic100 PP±j 10 l 8. a a a a E 6 a a 0. ....0. 4 a a a -Cl) ~ 2 a a a a a a a a i ; ; 0 -2 Time (minutes) 1·e 5..------, co ....Cl 4 • >< 3 ":'en- ~ 2 E 0 .§. >< 0 +----...... ,r------,----,-----,-----,------1 I :::, u::: 0 2 3 4 5 6 I I______cc_e_J_(m_o_l _m_-3_> ______j Figure 5.16 Diffusion of Ce in aqueous 0.55 M H2S04 through a hydrophilic membrane A Cerium concentration in the receiver solution as a function of time and [Ce]o B Effect of initial cerium concentration on flux 154 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction From Equations (5.52-55), the following relation can be derived [ /] + ([/] . - [/] . - [/]; + [I]; ) - [/] m,ai m,a, K K K 1 =---='------____;:d d d __ 5.56 when Kd >> 1, Equation 5.56 approximates to Equation 5.57, indicating that iodine transfer through a hydrophilic membrane will be slower than with an hydrophobic membrane 5.57 Similarly, the mass transfer rate of acetone from a solvent to an aqueous phase is defined as (Figure 5.15 B): NAc = K)[Ac]-KAAc]) 5.58 = k 0 ~Ac]m,ai -[Ac]) 5.59 = km~Ac]m,oi -[Ac]m,a;) 5.60 = k)[Ac]-[Ac];) 5.61 From Equations 5.58-5.61, it can readily be shown that 5.62 The hydrophilic membrane used in this work was the Millipore microporous membrane WLP. The membrane thickness (125 µm) and porosity (70%) are given by the manufacturers, but the tortuosity is unknown. This parameter was 155 Chapters The role of reaction kinetics in membrane assisted solvent extraction determined by a method similar to that used in Section 5.3.3. Three overall mass transfer coefficients were experimentally determined. The first was for the transfer of iodine from water to heptane through the membrane WLP at 200 rpm stirring speed in both phases. A value of 1. 77E- 06 was obtained for K 0 u-watertheptane>, which can be described as: I I I I -----=---+---+----- 5.63 Ka(/-waterlheptane) ka(/-water) km(/-water) Kdko(/-hepatane) The mass transfer of acetone from heptane to water through the same membrane, described by Equation 5.64 was 1.95E-05. I Kd Kd I -----= + +---- 5.64 Ko(Ac-heptane/water) ka(Ac-water) km(Ac-water) ko(Ac-hepatane) Finally the cerium flux from an aqueous cerium sulfate solution to an aqueous receiver solution through the hydrophilic WLP membrane was also measured. The acidity in both sides was 0.55 M H 2S04 . Results are presented in Figure 5.16. It was found that the flux has a linear dependence on the initial cerium concentration, with Ka equal 7.80E-07 m s·1, measured from the slope of the flux versus cerium concentration plot according to Equation 5.65. 5.65 I 2 I where - = +----- 5.66 Ka ka(Ce-0.55Msulfate) km(Ce-0.55Msulfate) In Equations 5.63- 5.66, ka and k0 values where calculated assuming that the hydrodynamic conditions, and therefore the aqueous and organic boundary film thicknesses, were the same for both the GVHP and the WLP membranes. Given these data, the tortuosity parameter was optimised to satisfy all three Equations 5.63-64 and 5.66. A value of t = 2.8 was obtained which is similar to values reported for other hydrophilic membranes, which range between 1.5- 2.7 (Prasad 1987). No tortuosity values were found in the literature for this particular membrane. It is interesting to note that when Kd << 1, the last two terms in Equation 5.62 can safely be ignored, which means that the mass transfer rate is only 156 Chapters The role of reaction kinetics in membrane assisted solvent extraction dependent on the organic boundary layer resistance. This conclusion is not valid for the case of acetone, where Kd = 0.1. In this case the aqueous, organic and membrane diffusions contribute 10, 22 and 68%, respectively, to the overall resistance. 5.3. 6 Stripping of cerium with a membrane permeation cell The back-extraction of cerium with sulfuric acid from a solvent phase containing di-2-ethylhexyl phosphoric acid inn-heptane was measured with the membrane permeation cell. A hydrophilic membrane (Millipore WLP) was used because the work presented in Chapter 4 showed that the back extraction rate was enhanced by the use of a hydrophilic membrane. The back extraction mass transfer rate was measured by contacting the solvent phase, loaded with a know amount of cerium, with a receiver solution of sulfuric acid containing no cerium, in a membrane permeation cell with a WLP membrane. The bulk solutions were stirred at 200 rpm, in keeping with the conditions used to determine the mass transfer coefficients for iodine and acetone. Experimental details are given in Chapter 2. The aqueous cerium concentration was measured as a function of time. Plots of [Ce] versus time were found to be linear close to the start of the reaction, indicating that the contribution of the reverse reaction is not significant in that particular time period. The mass transfer rate was calculated from the initial slope of the [Ce] versus time plot as per Equation 5.59. Typical data points obtained are shown in Figure 5.17. N _ Va d[Ce] 5.67 1 - A dt The concentration profiles for the reacting species in this system are represented in Figure 5.18. Under steady state conditions, the transport flux equations are as follows: 5.68 5.69 5.70 157 Chapters The role of reaction kinetics in membrane assisted solvent extraction where K1H.1 is the mass transfer constant across the membrane and aqueous static layer and is given by: 1 1 1 = +----- 5.71 K[H+] k aUH+ ]/2M .,·ulphuric) k m([H+]/ 2M sulphuric) 5. 3. 6. 1 Determination of chemical reverse rate equation The mass transfer of cerium from a loaded Ce-DEHPA-heptane solvent phase to a sulfuric acid receiver phase was measured with the membrane permeation cell as described above. The effect of DEHPA, MEHPA, acidity and Ce DEHPA complex on the cerium flux was determined, and is presented in log plots in Figures 5.19-5.22. Prior to the start of each experiment, a solvent mixture was loaded with around 0.2-3.6 g L-1 Ce, with the DEHPA/Ce ratio being high enough to ensure mainly the presence of the species CeR4 in the solvent, as suggested by Equation 5.38. The amount of free DEHPA was calculated by deducting four times the organic cerium concentration from the initial monomer DEHPA concentration. Experimental details are given in Chapter 2. In order to eliminate the effect of diffusion on the apparent rate equation, interfacial concentrations were estimated for the reacting species using Equations 5.68-5.70. The individual aqueous and organic diffusion layer mass transfer coefficients were calculated as explained previously, and are presented in Table 5.6 The diffusivities of [H+] in 2 M sulphuric and DEHPA and Ce DEHPA complex in heptane were estimated using the Stokes-Einstein correlation. No significant differences between bulk and interfacial concentrations were registered. The differences averaged 1.6%, 0.05% and 0.7% for [CeR4]j, [H+lom and free [(HRht, respectively. These results indicate that diffusion processes do not play a significant role in the overall transport and the stripping process is mainly controlled by chemical reaction. Figures 5.19 - 5.20 show that the mass transfer rate is linearly dependent on [CeR4], [H+] and free [(HRh] concentrations. The free DEHPA concentration was calculated by taking into account the DEHPA molecules complexed with the cerium loaded in the organic. 158 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction -~------x 1 c 2 A3 +4 o 5 20 ~ 18 16 14 - E c. 12 c. -...,. 10 'i' 8 ~ 6 4 2 0 0 5 10 15 20 25 30 Time (minutes) ------~ ------ Figure 5.17 Back-extraction of Ce in membrane permeation cell: Increase of Ce in receiver solution as a function of time. 3 1. 185 mol m-3 DEHPA, 2.0 mol m· MEHPA, 4000 mol m·3 H2S04, [CeRJ]=26 mol m·3 3 2. 193 mol m· DEHPA, 1.7 mol m-3 MEHPA, 4000 mol m·3 H2S04, [CeRJ]=18 mol m-3 3 3 3. 104 mol m· DEHPA, 3.3 mol m· MEHPA, 4000 mol m·3 H2S04, [CeRJ]=7 mol m·3 3 3 4. 153 mol m· DEHPA, 3.5 mol m· MEHPA, 4000 mol m·3 H2S04, [CeRJ]=7 mol m·3 3 3 5. 197 mol m· DEHPA, 1.8 mol m· MEHPA, 8000 mol m·3 H2S04, [CeRJ]=15 mol m-3 Hydrophilic Aqueous Membrane Organic .. .. -----I~ [Ce}¼] [Ce] [CeJ¼]i a/o interface+ Figure 5.18 Concentration profiles for reacting species in the back extraction process of Ce4•, from DEHPA/n-heptane solvent phase to a sulfuric acid aqueous phase, through a hydrophilic membrane. 159 Chapters The role of reaction kinetics in membrane assisted solvent extraction 1.E-05 ~-- ·------~ ';"II)-- 1.E-05 '1'e in 0 "I .§.1.E-06 E >< 0 ::I .§. ii: ::::,>< C) 1.E-06 .9 ii: 1.E-07 ~.~-.~~~~~ 1.E-07 '---~-----...... a 10 100 1000 10000 [Hji,om (mol m-3) Log [CeR.J (mol m·3) Figure 5.19 Effect of [CeR.i] on Figure 5.20 Effect of acidity on back extraction flux. back extraction flux. 3 [DEHPA] = 190 mol m·3, [DEHPA] = 190 mol m· , [H2S04] = 4000 mol m·3 [CeR4] = 17 mol m·3 1.E-04 ------ --II) ';" 1.E-05 -II) c:iE N 'E E 1. E-06 e ->< ::I ->< ii: ::I I ii: C) C) 1.E-06 ~. .2 .2 [ 10 100 10 100 1000 log [MEHPA] (mol m·3) log [(HR)z]; (mol m·3) Figure 5.22 Effect of [MEHPA] on Figure 5.21 Effect of free[(HR)2] on back extraction flux. back extraction flux. [CeR.i] = 7.1 mol m·3, [CeR.i] = 7.1 mol m·3, [H2S04] = 4000 mol m·3 [H2S04] = 4000 mol m·3 160 Chapters The role of reaction kinetics in membrane assisted solvent extraction TABLE 5.6 Calculated Mass Transfer Coefficients for Ce, H+, CeR4 and (HRh for Back- extraction with Hydrophilic Membrane Mass Transfer Viscosity Kinematic System Coefficient Diffusivity solvent Viscosity ms-1 m2s-1 cP m2s-1 ka (Ce/2 M sulphuric) 8.22E-06 4.76E-10 11.45 11.30E-06 km (Ce/2 M sulphuric) 9.46E-07 ka (W/2 M sulphuric) 3.75E-05 4.65E-09 km (W/2 M sulphuric) 9.24E-06 k0 ((HR)2'heptane) 2.33E-05 1.01 E-09 0.40 5.58E-07 k0 (CeR4/heptane) 1.78E-05 6.70E-10 1(R.C.Weast (Ed.) 1981) The flux increases with increasing cerium concentration in the solvent and hydrogen ion concentration in the aqueous as is expected from the back extraction reaction represented by 5.72 Conversely, the reverse equilibrium reaction also explains the inverse dependency of the flux on the free extractant or DEHPA concentration. In contrast to that observed for extraction, the MEHPA concentration was found to have no significant effect on the back-extraction mass transfer rate at low MEHPA concentrations of 1.7 - 10 mol m-3• At this level, MEHPA represents less than 5% of the total extractant (MEHPA + DEHPA) concentration. As this ratio is increased, the MEHPA then becomes a significant extractant affecting the equilibrium of the reaction, which does not favor stripping and therefore will tend to reduce mass transfer as discussed in Chapter 4. This effect can be seen by a drop in flux at 15 mol m·3 MEHPA shown in Figure 5.22. The initial stripping or back extraction rate can thus be represented by Equation 5.73 with an average error of 25%. 5.73 161 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction The power dependencies of the reverse rate equation are related to the slow step of the overall reaction mechanism and therefore are not necessarily the same as those arising from stoichiometric considerations of the equilibrium. In this work, as for the extraction reaction, the focus is to use the experimentally determined reverse rate equation to quantify the relative contributions of chemical reaction with respect to diffusion processes. 5.3.6.2 Transport mechanism of membrane liquid-liquid back-extraction process Similarly to the methodology used for describing the transport mechanism for the liquid-liquid membrane extraction process, the back extraction process can be described in terms of four main resistances: 1. Resistance due to diffusion through organic stationary layer. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: 5.74 2. Resistance due to chemical reaction. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: 5.75 3. Resistance due to diffusion through the membrane pores. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: Nh,m = k,,,(ce-2Msulfuric)[Ce];,om k,,,(Ce-2 M sulfuric) [ CeR4 lb = 5.76 (Kd + 1) K _ [CeR4]b -[Cel,om where d - [C ]. e ,,0111 162 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction 4. Resistance due to diffusion through the stationary aqueous layer. Under conditions where this resistance controls the flux of cerium, the mass transfer can be described as: N,,a = ka(Ce-2Msulfuric)[Ce ]i,am _ ka(Ce-2Msulfuric)[CeR4]b 5.77 (Kd + 1) The fractional resistance due to aqueous layer diffusion (8r,a), chemical reaction (8r,c), membrane diffusion (8r,m) and organic layer diffusion (8r,o), can be quantified by Equations 5.74-5.77 with a method analogous to that used for extraction. These have been calculated for the various experimental conditions tested and the results are presented in Table 5. 7. It is evident from the results that the main resistance to mass transfer is due to chemical reaction, with the fractional resistance averaging around 83% of the overall resistance for the back stripping of cerium. Under the experimental conditions studied, membrane diffusion contributes on average 14.3% to the overall resistance, with aqueous and organic film diffusion layer contributing 1.6 and 0.7%, respectively. The relative importance of membrane and chemical reaction to the overall transport is dependent on the distribution coefficient. In this set of results the greatest changes in the distribution coefficient occur with changes in acidity. Kd in this instance decreases by 1.5 orders of magnitude as one moves from 2200 to 8000 mol m-3 hydrogen ion concentration. Thus at the higher acidities where stripping is favoured, the reverse chemical reaction is relatively fast and the membrane contribution accounts for 40% of the resistance. This is important from a practical point of view because these are likely operating conditions in a process. These results show that in the case of stripping of cerium the chemical reaction kinetics play an important role and should be taken into account in the design of equipment. For instance, the slowness of the stripping reaction maybe counteracted by increased surface area made available in the stripping module, when compared to extraction. 163 Chapters The role of reaction kinetics in membrane assisted solvent extraction TABLE 5.7 Fractional Resistances of Aqueous and Organic Diffusion Layers, Diffusion in Membrane Pores and Chemical Reaction for the Stripping Reaction of Cerium [DEHPA] [MEHPA] [CeR4] [HJ mo1m-3 mo1m-3 mol m-3 mol m-3 Lir,a{%) Lir,c{%) Lirm{%) Lir,oC%) 181 1.7 18.4 2200 0.4 95 3.9 0.2 193 1.7 17.8 3400 1.0 89 9 0.5 193 1.7 18.4 4000 1.5 85 13 0.7 193 1.7 17.8 5000 2.3 77 20 1.1 193 1.7 18.2 6000 3.2 68 38 1.5 197 1.8 15.5 8000 4.6 53 40 2.1 ------104 3.3 15.7 4000 3.2 68 28 1.5 153 3.5 15.7 4000 1.8 82 15 0.8 203 3.5 15.7 4000 1.2 88 11 0.6 252 3.7 15.7 4000 0.9 90 8 0.4 300 3.4 15.7 4000 0.8 92 7 0.3 ------185 1.7 2.1 4000 1.1 89 9.3 0.5 185 1.7 4.6 4000 1.1 89 9.8 0.5 190 1.7 9.6 4000 1.2 88 11 0.6 185 1.7 16.1 4000 1.5 85 13 0.7 193 1.7 18.4 4000 1.5 85 13 0.7 185 2.0 25.8 4000 2.0 80 18 0.9 ------201 2.3 15.7 4000 1.3 87 11 0.6 200 10 15.7 4000 1.1 89 10 0.5 201 15 15.7 4000 1.1 89 9 0.5 5.4 Conclusions The mass transfer rate of membrane assisted extraction of cerium into a solvent phase containing di-2-ethylhexyl phosphoric acid and its back extraction into an acidic aqueous phase were investigated with a modified Lewis-type membrane permeation cell. The hydrodynamics of the membrane permeation cell were characterised by study of simple mass transfer with no chemical reaction of iodine and acetone. This method allowed for the quantitative calculation of the contribution of the 164 Chapters The role of reaction kinetics in membrane assisted solvent extraction various diffusion processes to the overall mass transfer of cerium and the derivation of an empirical rate equation for the chemical reaction. It was found that for the range of experimental conditions studied, the liquid - liquid extraction process for cerium is mainly controlled by membrane diffusion. On average, contributions to the overall resistance are distributed as follows: diffusion through the membrane pores contributes 63%, chemical reaction contributes 20%, diffusion through the aqueous layer contributes 11 % and diffusion through the organic layer contributes 6%. Reducing the thickness of the membrane will therefore significantly lower the resistance to mass transfer and increase overall flux. The rate of chemical reaction can contribute up to 36% of the overall resistance, reflecting the finite interfacial capacity for extractant molecules at the interface. The accelerating effect of MEHPA on cerium extraction discussed in Chapter 4 has been quantified. Without the presence of low concentrations of MEHPA, the contribution of chemical resistance would be greater. The back extraction reaction was also studied using a similar method, and in this case it was determined that the chemical resistance, on average, accounted for 82% of the total resistance to mass transfer. In this case, membrane diffusion contributed around 16%. However, when the distribution coefficient is very low (eg, high acid concentrations), which is optimum for stripping, the membrane diffusion contribution can be up to 40%. These results highlight the importance of understanding the fundamental processes occurring in a particular extraction system, so that the appropriate design can be applied to a potential process. In the present cerium(IV)/sulfate/DEHPA/MEHPA system, it has been shown that the most important parameter for extraction is the thickness of the membrane, whilst for stripping, it is important to increase the surface area available for back extraction relative to the surface area available for extraction. Furthermore, in the modeling of membrane assisted extraction, it has often been assumed that instantaneous chemical equilibrium occurs at the interface. This work has shown that this assumption is not valid for the system studied. 165 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction 5.5 Nomenclature Symbol A membrane area (m 2) D diffusivity of a species in a bulk liquid phase (m 2 s-1) 0 D} relative density at 20°c (g cm- 3) Ka overall aqueous mass transfer coefficient (m s-1) Kd equilibrium distribution coefficient 1 K0 overall organic mass transfer coefficient (m s- ) ka aqueous film mass transfer coefficient (m s- 1) 1 k0 organic film mass transfer coefficient (m s- ) km mass transfer coefficient trough the membrane (m s-1) N Flux or mass transfer rate (mol m-2 s-1) R Rate of forward and reverse chemical reaction (mol m-2 s-1) t time (s) V Total volume of aqueous or organic (m 3) VA Molar volume of solute at its normal boiling point (cm-3 mol-1) Greek letters gm thickness of the membrane (m) s porosity of the membrane T/e viscosity of the solvent (cP) r kinematic viscosity of the solvent (m 2 s-1) , tortuosity factor of the membrane ro rotational speed (s- 1) Subscripts a aqueous b bulk c chemical reaction 166 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction f forward interface m membrane o organic r reverse reaction 0 initial concentration Acronyms and abbreviations [~ Molecular iodine concentration (mol m·3) [Ac] Acetone concentration (mol m·3) [DEHPA] di-2-ethylhexyl phosphoric acid concentration (mol m·3) 3 [(HR) 2 ] dimeric concentration of DEHPA (mol m· ) [MEHPA] mono-2- ethylhexyl phosphoric acid (mol m·3) 5.6 Reference List 1. Ajawin L. A., Demetriou J., Perez de Ortiz E.S. and Sawistowski H. 1985. I. Chem.E. Sym. Series 88: 183. 2. Ajawin I. A., Perez de Ortiz E.S. and Sawistowski H. 1983. Extraction of zinc by di(2-ethylhexyl) phosphoric acid. Chem. Eng. Res. Des 61: 62- 66. 3. Albery W. John and Fisk Peter R. 1981. The kinetics of extraction of copper with Acorga P50 studied by a diffusion cell. Hydrometallurgy '81 , Proc. Soc. Chem. Ind. Symp. F5-1-F5/15, Soc. Chem. Ind., London, UK. 4. Amankwa L. and Cantwell F. F. 1989. Investigation of fast mass transfer kinetics in solvent extraction using rapid stirring and a porous membrane phase separator. Anal. Chem. 61, no. 9: 1036-40. 5. Aparicio J. and Muhammed M. 1989. Extraction kinetics of zinc from aqueous perchlorate solution by D2EHPA dissolved in lsopar-H. Hydrometallurgy 21: 385-99. 167 Chapters The role of reaction kinetics in membrane assisted solvent extraction 6. Awakura Y., Doi T. and Majima H/. 1988. Determination of the diffusion coefficients of CuS04, ZnS04 and NiS04 in aqueous solution. Metal/. Trans. B. 198: 5-12. 7. Chen C. Q. and Zhu T. 1994. The kinetics of cobalt(II) extraction with EHEHPA in heptane from acetate system using an improved Lewis cell technique. Solvent Extraction and Ion Exchange 12, no. 5: 1013-32. 8. Cianetti C. and Danesi P.R. 1983. Solv. Ext. Ion Exchange 1: 9. 9. Coleman C. F. and Roddy J. W. 1971. Kinetics of metal extraction by organophosphorus acids. Solvent Extraction Reviews, ed. Marcus Y. NY: Marcel Dekker.Inc. 10. Cotton F. A. and Wilkinson G. 1980 Fourth ed. New York: John Wiley & Sons. 11. Danesi P. R. 1992, Vol. Chapter 5. Principles and Practices of Solvent Extraction, eds. Rydberg J., Musikas C., and Choppin G.R. New York: Marcel Dekker Inc. 12. Danesi P. R. 1984. The relative importance of diffusion and chemical reactions in liquid-liquid extraction kinetics. Solvent Extr. Ion Exch. 2: 29- 44. 13. Danesi P. R. and Chiarizia R. 1980. The kinetics of metal solvent extraction. CRC Critical Reviews in Analytical Chemistry 10: 1. 14. Danesi P. R., Vandegrift G.F. and Horwitz E.P. 1980. Simulation of two step consecutive reactions by diffusion in the mass-transfer kinetics of liquid-liquid extraction of metal cations. J. Phys. Chem. 84: 3582-87. 15. Danesi P.R. and Vandergrift G. F. 1981. Kinetics and mechanism of the interfacial mass transfer of Eu3+ and Am3+ in the system bis(2- ethylhexyl)phosphate-n-dodecane-NaCI-HCl-water. Journal of Physical Chemistry 85: 3646-51. 168 Chapters The role of reaction kinetics in membrane assisted solvent extraction 16. Danesi P. R., Vandergrift G.F., Horwitz E.P. and Chiarizia R. 1980. J. Phys. Chem. 84: 3582. 17. Daoud J. A., Gasser M. and Aly H. F. 1995. Extraction mechanism of Eu(III) by octylphenyl acid phosphate and di-(2-ethylhexyl) phosphoric acid in benzene from phosphate medium. Radiochimica Acta 69, no. 3: 197-200. 18. Dreisinger D. B. and Cooper W. C. 1989. The kinetics of zinc, cobalt and nickel extraction in the D2EHPA-heptane-HCI04 system using the rotating diffusion cell technique. Solvent Extraction & Ion Exchange 7, no. 2: 335-60. 19. Golding J. A. and Pushparajah. 1985. Hydrometallurgy 14, no. 3: 295. 20. Harada M. and Miyake Y. 1986. Formulation of metal extraction rates in solvent extraction with chelating agents. J. Chem. Eng. Jpn. 19: 196- 207. 21. Huang T. C. and Huang C.T. 1988. Kinetics of the extraction of uranium(VI) from nitric acid solutions with bis(2-ethylhexyl)phosphoric acid. Ind. Eng. Chem. Res. 27, no. 9: 1675-80. 22. Hughes M. A. and Biswas R.K. 1991. The kinetics of vanadium(IV) extraction inthe acidic-sulphate-D2EHPA-n-hexane system using the rotating diffusion cell technique. Hydrometallurgy 26: 281-97. 23. Hughes M. A. and Rod V. 1984. On the use of the constant interface stirred cell for kinetic studies. Hydrometallurgy 12: 267-73. 24. Ihm S. K., Lee H.Y. and Lee D.H. 1988. Kinetic study of the extraction of copper(II) by Di(2-ethylhexyl)phosphoric acid in a Lewis type cell. J. Membr. Sci. 37: 181-91. 25. lmato S. K., Ogawa H., Morooka S. and Kato Y. 1981. Transport of copper through supported liquid membranes containing LIX65N . J. Chem. Eng. Jpn. 14, no. 4: 289-95. 169 Chapter5 The role of reaction kinetics in membrane assisted solvent extraction 26. lpinmoroti K. 0. and Hughes M.A. 1990. The mechanism of V(IV) extraction in a chemical kinetic controlled regime. Hydrometallurgy 24: 255-62. 27. lrabien A. and Ortiz I. 1990. Kinetics of metal extraction: Model discrimination and parameter estimation. Chem. Eng. Process. 27: 13- 18. 28. Islam F. and Biswas R.K. 1980. Kinetics of solvent extraction of metal ions with HDEHP-11. Kinetics and mechanism of solvent extraction of V(IV) from acidic aqueous solutions with bis(2-ethylhexyl)-phosphoric acid in benzene. J. lnorg. Nucl. Chem. 42: 421-29. 29. Juang R. S. and Chang Y.T. 1993. Kinetics and mechanism for copper(II) extraction from sulphate solution with Di(2- ethylhexyl)phosphoric acid dissolved in kerosene. J. Chem. Eng. Jpn 26, no. 2: 219-22. 30. Juang R. S. and Huang R. H. 1997. Kinetic studies on lactic acid extraction with amine using a microporous membrane-based stirred cell. Journal of Membrane Science 129, no. 2: 185-96. 31. Juang R. S. and Lin Y. S. 1998. Investigation of interfacial reactyion kinetics of penicillin G and Amberlite LA-2 from membrane flux measurements. Journal of Membrane Science 141, no. 1: 19-30. 32. Juang R. S and Lo R.H. 1994. Mass transfer characterictics of membrane permeation cell and its application to the kinetic studies of solvent extraction. Ind.Eng.Chem.Res. 33: 1001-10. 33. Juang Ruey-Shin and Lo Ru-Huey. 1994. Kinetics of the Coupled Transport of Vanadium(IV) from Sulfate Solutions through Supported Liquid Membranes. Ind. Eng. Chem. Res. 33, no. 4: 1011-16. 34. Kiani A., Bhave R. R. and Sirkar K. K. 1984. Solvent extraction with immobilized interfaces in a microporous hydrophobic membrane. J. Membr. Sci. 20, no. 2: 125-45. 170 Chapters The role of reaction kinetics in membrane assisted solvent extraction 35. Komasawa I. and Otake T. 1983. Kinetic studies of the extraction of divalent metals from nitrate media with bis(2-ethylhexyl)phosphoric acids. Ind. Eng. Chem. Fundam. 22: 367-71. 36. Komasawa I., Otake T. and Yamada A. 1980. Diffusional resistance in extraction rate of copper with hydroxyoxime extractant. Journal of Chemical Engineering of Japan 13, no. 3: 209-13. 37. Lazarova Z. 1995. Study on the kinetics of copper/LIX54 system using a rotating diffusion cell. Solvent Extr. Ion Exch. 13, no. 3: 525-40. 38. Levich V. G. 1962. Physicochemical Hydrodynamics. 60-72. NJ: Prentice-Hall. 39. Lewis J.B. 1954. Chem. Eng. Sci. 3: 248-59. 40. Lim, T. M. 1996. "Kinetic studies of solvent extraction of rare earths into di-2-ethyl hexyl phosphoric acid (DEHPA)." PhD Thesis, The University of New South Wales. 41. Lim T. M., Levins D. M., Wiblin W. A. and Tran T. 1996. Kinetic studies of solvent extraction of rare earths into DEHPA. Value adding trough solvent extraction, Proceedings to the lnernational Solvent Extraction Conference (ISEC'96)R. Paimin & L. M. Prvcic D.C. Shallcross, Vol1 (451-456), The University of Mlbourne. 42. Ma E. and Jiang P. 1991. Kinetics and mechanism of yttrium extraction with three acidic phosphorus extractants. J. Chem. Tech. Biotechnol. 51: 315-21. 43. Mendes-Tatsis M. A. and Perez de Ortiz E.S. 1986. Mathematical models for the rate of metal extraction: which one describes your system? International Solvent Extraction Conference /SEC 1986 , 2, Frankfurt-on-Main: DACHEMA. 44. Meng Michael X., Yu Shuqiu and Chen Jiayong. 1996. Kinetics of iron(III) extraction with primary amine and TBP using a modified rotating diffusion cell. Hydrometallurgy 41, no. 1: 55-70. 171 Chapters The role of reaction kinetics in membrane assisted solvent extraction 45. Miyake Y., Matsuyama H., Nishida M., Nakai M., Nagase N. and Teramoto M. 1990. Kinetics and mechanism of metal extraction with acidic organophosphorus extractants (I): Extraction rate limited by diffusion process. Hydrometallurgy 23: 19-35. 46. Mori Y., Ohya H., Koresawa E. and Eguchi W. 1986. Extraction equilibrium and kinetics of some lanthanoid with acidic organophosporus extractants. Proceedings to the Symposyum on Solvent Extraction119- 24. 47. Nitsch W. and Kruis B. 1978. J. lnorg. Nucl. Chem. 40: 857. 48. Partridge J. A. and Jensen R.C. 1969. Purification of di-(2- ethylhexyl)phosphoric acid by precipitation of copper(II) di-(2- ethylhexyl)phosphate. J. lnorg. Nucl. Chem. 31: 2587-89. 49. Persaud Gocool, Tian Xiu Min and Cantwell Frederick F. 1987. Behavior of solute adsorbed at the liquid-liquid interface during solvent extraction with porous-membrane phase separators. Anal. Chem. 59, no. 1: 2-7. 50. Prasad R., Frank G. T. and Sirkar K. K. 1988. Nondispersive solvent extraction using microporous membranes. AIChE Symp. Ser. 84: 42-53. 51. Prasad R., Kiani A., Bhave R. R. and Sirkar K. K. 1986. Further studies on solvent extraction with immobilized interfaces in a microporous hydrophobic membrane. J. Membr. Sci. 26, no. 1: 79-97. 52. Prasad R. and Sirkar K. K. 1988. Dispersion-free solvent extraction with microporous hollow-fiber modules. AIChE J. 34, no. 2: 177-88. 53. Prasad R. and Sirkar K.K. 1992. Membrane Handbook., eds. W.S.W.Ho, and K.K.Sirkar. New York: Van Nostrand Reinhold. 54. Prasad R. and Sirkar K. K. 1987. Solvent Extraction with Microporous Hydrophilic and Composite Membranes. AIChE Journal 33, no. 7: 1057- 66. 172 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction 55. Qiang Y., Jingfen Y., Jiufang L. and Teng T. 1989. lnterfacial characters of organophosporic and phosphonic acid extraction systems. Rare Metals 8, no. 2: 5-10. 56. R.C.Weast (Ed.). 1981. Handbook of Chemistry & Physics 61st Edition.CRC Press. 57. Reid, R. C., Prausnitz J.M. and Poling B.E. 1987. The properties of gases and liquids. McGraw-Hill Inc. 58. Rydberg J. W., Reinhardt H. and Liljenzin J.O. 1973. Ion Exch. Solvent Extr. 3: 111. 59. Saad E. A. 1998. Kinetic studies of extraction for a mixture of some europium phosphonate complexes and 2-benzoyl or 2-anilinopyridine in n-hexane. Microchemical Journal 58, no. 1: 6-12. 60. Sato T., Yoshino T., Nakamura T. and Kudo T. 1978. The kinetics of Aluminium(III) extraction fromsulphuric acid solutions by di-(2- ethylhexyl)-phosphoric acid. J.lnorg.Nucl. Chem. 40: 1571-74. 61. Sato Y. Kondo K. and Nakashio F. 1989. Extraction kinetics of zinc with 2-ethylhexylphosphonic acid mono-2-ethylhexyl ester using a hollow fiber membrane extractor. J. Chem. Eng. Jpn. 22, no. 6: 686-9. 62. Sato Y., Akiyoshi Y., Kondo K. and Nakashio F. 1988. Extraction kinetics of copper with 2-ethylhexylphosphonic acid mono-2-ethylhexyl ester. J. Chem. Eng. Jpn. 22, no. 2: 182-9. 63. Simonin J. P., Turq P. and Musikas C. 1991. J.Chem.Soc, Faraday Trans. 87, no. 17: 2715-21. 64. Svendsen H. F., Schei G. and Osman M. 1990. Kinetics of extraction of zinc by di(2-ethylhexyl)phosphoric acid in cumene. Hydrometallurgy 25: 197-212. 65. Tarasov V. V., Yagodin G.A. and Yurtov E.V. 1976. Chimiia I CHimiceskaia Tech. 19, no. 2: 245. 173 Chapter 5 The role of reaction kinetics in membrane assisted solvent extraction 66. Teramoto M., Matsuyama H., Yamashiro T. and Okamoto S. 1989. Separation of ethylene from ethane by a flowing liquid membrane using silver nitrate as a carrier. Journal of Membrane Science 45: 115-36. 67. Vandegrift G. F. and Horwitz E. P. 1980. lnterfacial activity of liquid-liquid extraction reagents - I. J. /nor. Nuc/. Chem. 42: 119-25. 68. Vandegrift G. F. and Horwitz E. P. 1977. The mechanism of interfacial mass transfer of calcium in the system: di(2-ethylhexyl)phosphoric acid in dodecane-dilute nitric acid. J. lnorg. Nucl. Chem. 39: 1425-32. 69. Yoshizuka K., Kondo K. and Nakashio F. 1986. Effect of hydrophobicity of the extractant on extraction kinetics of copper with N-8- q uinolylsulfonamide. J. Chem. Eng. Jpn. 19, no. 5: 396-400. 70. Yoshizuka K., Kondo K. and Nakashio F. 1986. Effect of interfacial reaction on rates of extraction and stripping in membrane extractor using a hollow fiber. J. Chem. Eng. Jp 19, no. 4: 312-18. 71. Yoshizuka K., Sakamoto Y., Baba Y., Inoue K. and Nakashio F. 1992. Extraction Kinetics of Holmium(III) and Yttrium(III) with D2EHPA in Membrane Extractor using a hollow Fiber. in Solvent Extraction 1990T. Sekine805-10, Elsevier. 72. Yoshizuka K., Sakamoto Y., Baba Y., Inoue K. and Nakashio F. 1992. Solvent Extraction of Holmium and Yttrium with Bis(2- ethylhexyl)phosphoric Acid. Industrial and Engineering Chemistry Research 31, no. 5: 1372-78. 174 Chapter 6 Purification of cerium with hollow fibre contactors Summary This chapter investigates the use of hollow fibre contactors for the purification of cerium from other rare earths and explores some of the practical aspects of the technology. The continuous operation ofa two module hollow fibre contactor rig for 65 hours is described. The operation was successful, achieving good extraction and selectivity with respect to cerium(IV). Low solvent losses to the feed stream and low carry over offeed to strip solution were measured. Chapter6 Purification of cerium with hollow fibre contactors 6.1 Introduction In Chapter 5, it has been shown that non-dispersive solvent extraction using flat sheet membranes is applicable to both the extraction and back extraction of cerium in sulfate media. Chapter 4 explored this system in a flat sheet bulk liquid membrane configuration. The same concept can be applied to hollow fibre membrane contactors using one module for extraction and a second module for stripping, with the solvent circulating between the two, acting as a bulk liquid membrane. There are quite a few studies in the literature dealing with the application of hollow fibre contactors to the liquid-liquid extraction of metal ions such as Zn, Cr and Cd. These publications, whilst important in establishing the applicability of the technology, do not deal with the more practical aspects that need to be addressed to further develop the technology towards industrial application. The current chapter investigates the use of hollow fibre contactors for the purification of cerium from other rare earths, exploring some of the practical aspects of the technology. 6.2 Background Mass transfer in liquid - liquid systems with hollow fibre membrane contactors occurs without dispersion of one phase into another. This technique offers a number of advantages over conventional contactors such as packed columns and mixer-settlers, which include; known surface area, no emulsification, independent fluid flows, a density difference is not required between the two liquids, no moving parts and straightforward scale-up (Dahuron 1988), (Prasad 1990), (Prasad 1988b). Membrane contactors have been applied to a wide range of separations. A recent review on the applications of hollow fibre membrane contactors has been published (Gabelman 1999). A hollow fibre membrane contactor consists of a bundle of hollow fibres encased in a tubular cartridge. A schematic diagram of a hollow fibre contactor is shown in Figure 6.1. In the most common design, the tube and shell side flows are parallel. Hoechst Celanese have developed a more intricate baffled module, known as Liqui-Cel® Extra-Flow, which minimises shell side bypassing and improves mass transfer (Sirkar 1997). 175 Chapter 6 Purification of cerium with hollow fibre contactors / Hollow Fibres Tube Tube Side Side Inlet Outlet Shell Side Shell Side Inlet Outlet Figure 6.1 Schematic diagram of hollow fibre contactor A) B) v C Figure 6.2 Photograph of cross sectional area of hollow fibre contactors A - Extraction module: ID= 15 mm, hydrophobic fibres= 0.6 mm ID B - Stripping module: ID = 35 mm, hydrophilic membrane capillaries = 1.5 mm ID 176 Chapter6 Purification of cerium with hollow fibre contactors Hollow fibre membrane contactors can be used for membrane assisted solvent extraction and back-extraction. Examples of such studies are presented in Section 6.2.1. Two other important techniques that also make use of hollow fibre contactors are hollow fibre supported liquid membranes (HFSLM) and hollow fibre contained liquid membranes (HFCLM). These two are also briefly discussed. 6.2.1 Liquid-liquid extraction with hollow fibre membrane contactors Dispersion free liquid-liquid extraction has been widely applied in the laboratory to many different systems. The extraction of organic acids such as lactic acid (Hano 1996), (Tong 1998), (Tong 1999), (Coelhoso 1997) and mevinolenic acid (Prasad 1989) from fermentation broths is of particular interest for the production of pharmaceutical products, or as a substitute of petroleum derivative precursors for organic synthesis. It also has a use in the food industry, for example, in the extraction and purification of the sweetener L-phenylalanine (Escalante 1998). This technique has been applied to the removal of organic pollutants (Yun 1992), (Yunfeng 1999) and toxic heavy metals (Yun 1993) from waste waster. Extraction and recycling of chromium in electroplating baths (Alonso 1993), (Alonso 1994), recovery of Cd and Ni from spent batteries (Soler 1996), (Galan 1998) and the potential recovery of Cu, Ni, Pb and Zn from hydrometallurgical streams has also been explored (Daiminger 1996), (Valenzuela 1997). The effect of different membrane supports with respect to pore size, hydrophibicity and geometry of module has been extensively studied (Prasad 1988a), (Prasad 1990). Equations governing simple mass transfer have been published (Prasad 1988b) with the major area of uncertainty being the characterisation of shell side flow (Gabelman 1999). When mass transfer is accompanied by a chemical reaction, as in the case of metal extraction, description of the mass transfer becomes more complicated. If the chemical reactions are slow with respect to diffusional processes, then they have to be taken into account in the analysis of the overall mass transfer (Huang 1988). However, it has been widely assumed that chemical reactions are sufficiently fast so that only diffusion in the lumen, shell and membrane pore is taken into account in the modelling of mass transfer (Alonso 1996). 177 Chapter6 Purification of cerium with hollow fibre contactors The hollow fibre liquid-liquid extraction process can been coupled with other conventional separation techniques to recover the solvent and separate the product. An example is the study of hexanol extraction from water using octanol as a solvent. Octanol and hexanol were then separated by distillation (Seibert 1993). (Hutter 1994) et al studied the removal of volatile organic compounds from water by membrane assisted extraction into sunflower oil. Membrane assisted distillation at 70-90°C, in a separate hollow fibre module, was used to separate the contaminants. In non-dispersive metal solvent extraction processes, the metal needs to be stripped back into an aqueous phase and this process can also be carried out with a second membrane contactor. In most studies both the extraction and stripping modules are made up of hydrophobic fibres (Ortiz 1996), (de Haan 1989), (Galan 1998). Tubular ceramic modules have also been used for the back-extraction process due to the very high acidity of the stripping solution (Alonso 1997). Although the literature indicates that this technology has many advantages, membrane contactors have only been commercialised in gas-liquid applications such as in carbonation of soft drinks. 6.2.2 Hollow fibre contained liquid membrane technique Hollow fibre contained liquid membrane (HFCLM) is a technique developed in the last decade that combines extraction and back-extraction into a single hollow fibre module (Sengupta 1988). This is achieved by having two sets of fibres, one for the feed solution and the other for the strip solution. The aqueous solutions flow through the hollow fibre lumen and the solvent is stationary on the shell side and acts as a bulk liquid membrane. In this type of dual function module, the solvent in the shell can also be circulated. HFCLM has many of the advantages of liquid membrane techniques, without the problems associated with emulsion and supported liquid membrane techniques (Sirkar 1997). Only a few studies have been published on the application HFCLM to metal extraction (Sato 1990), (Lamb 1990), (Guha 1994). (Yang 1996) et al extended the HFCLM concept further by incorporating three sets of fibres in a single module. They used a mixed extractant to simultaneously remove Cr and Cu from the feed solution and selectively strip the metals into two separate strip streams. 178 Chapter6 Purification of cerium with hollow fibre contactors An interesting aspect is the question of how HFCLM perform against two separate hollow fibre modules with the solvent circulating between the two. In the former case the advantage is lower solvent inventory. Two separate modules, however, are easier to manufacture and pressure control at the membrane support interface, essential in prevention of leakages, is easier to achieve. Harrington investigated this issue with respect to the extraction of Cr with an amine type solvent and concluded that there was no significant difference in performance between two separate modules and a HFCLM set-up with flowing solvent (Harrington 1999). 6.2.3 Hollow fibre supported liquid membrane Hollow fibre membrane contactors have also been used in a supported liquid membrane configuration (HFSLM). The solvent phase is impregnated in the hollow fibre pores and the feed and strip solution are passed through the tube and shell side. With this technique, as is the case for HFCLM, extraction and back-extraction occur simultaneously in a single module. As already discussed in the introductory section to Chapter 4, the major drawback of this technology is the long term stability of the liquid membrane caused by solvent loss to neighbouring aqueous solutions. Nevertheless, hollow fibre supported liquid membranes for metal extraction is a very active field of research. Recent studies include the extraction of copper and its separation from Fe, Mo, Al, Co, Ni and Pb (Breembroek 1998), (Valenzuela 1999), (Campderros 1998), extraction of Zn from waste waters (Lee 1999), (Acosta 1998) and extraction of vanadium (Rosell 1997). The long term stability of this technique as applied to uranium extraction from contaminated Hanford site groundwater has been studied by (Chiarizia 1990) et al. They found that over a period of 1.5 years, the liquid membrane permeability could be sustained, provided the membrane support was re-impregnated every two months. A review on recent publications has been published by (Sastre 1998) et al. To summarise, the literature indicates that non-dispersive metal solvent extraction with hollow fibres has many advantages. This research is relatively recent and compared to supported liquid membranes, which is also a competitive technology to conventional solvent extraction, has been much less studied. Many laboratory studies have been performed on metal extraction systems, but integration into a whole process with extraction and stripping is less well documented. In particular, use has been made of the more commonly 179 Chapter6 Purification of cerium with hollow fibre contactors available modules containing hydrophobic fibres for both extraction and stripping, although in many instances it would appear the hydrophilic fibres in the stripping module enhance mass transfer. No studies have been found on the extraction and purification of cerium with hollow fibres, which is the subject of this chapter. 6.3 Results and discussion 6.3. 1 Experimental set-up and mode of operation The experimental set-up consisted of two hollow fibre modules, one for extraction and another for stripping. A detailed description of the apparatus is presented in Chapter 2, Section 2.7. The hollow fibre modules were manufactured in-house and therefore the opportunity was available to design the modules to suit the process. A summary of the module characteristics is presented in Table 6.1. Hydrophobic fibres were used for the extraction module and hydrophilic fibres for the stripping module. This arrangement maximized the overall mass transfer as discussed in Chapter 4. The internal diameter of the hydrophilic fibres was 1.5 mm. With such a large internal diameter the fibres cannot be classified as hollow fibres but are really capillary membranes. They were used in the stripping module because problems were experienced with the manufacture of a module using the smaller hydrophilic hollow fibres available (0.3 mm internal diameter). With the smaller diameter hydrophilic fibre, it was found that the epoxy resin used for potting travelled up the lumen, blocking a considerable number of the pores. Photographs of the cross sectional areas of the modules are shown in Figure 6.2. The stripping module was designed to have a much larger surface area than the extraction module to compensate for the lower stripping rates compared with extraction rates. The surface area available in each module is dependent on the physical parameters and on which phase is operated in the tube or shell side. For instance, in the extraction module, the membrane pores are wetted by solvent and therefore the surface area available for extraction is larger if the solvent is run on the tube side rather than the shell side. The reverse is true for the hydrophilic module, where the largest surface area is available when the aqueous is operated in the tube side and the solvent in the shell. In these tests, both the aqueous feed and strip solutions were operated in the tube side 180 Chapter6 Purification of cerium with hollow fibre contactors in order to maximize the difference in surface area between the extraction and stripping modules. The solvent was circulated between the two modules on the shell side. TABLE 6.1 Summary of module and fibre characteristics Extraction Stripping Module Module fibre type polypropylene polyethersulfone ID of fibre (mm) 0.60 1.5 fibre wall thickness (µm) 228* 300 Internal surface area (m2) 0.046 0.26 External surface area (m2) 0.085 0.37 packing fraction (%) 62 61 * wall thickness higher than specification of 200 µm due to swelling 6.3.2 Effect of f/owrate on pressure drop In dispersion-free extraction with microporous membranes, the impregnating phase has to be kept at a lower pressure than the non-impregnating phase in order to prevent leakage from one phase to another (Prasad 1988b), (Prasad 1988a). In practical terms, in a hydrophobic hollow fibre contactor, the solvent phase wets the pores of the hydrophobic membrane, and therefore the operating pressure of the aqueous phase has to be higher than that of the solvent phase. Conversely, in a hydrophilic hollow fibre contactor, the solvent pressure has to be higher than the aqueous pressure. However, if the difference in operating pressures between the two phases (dP) goes beyond a critical pressure (dPcr), then the non-impregnating phase will breakthrough the pore. In order to ensure operation of a hollow fibre contactor away from the breakthrough pressure, it is important to understand the relationship between flowrate and the pressure drop across the module. An example of such measurements for water flow in either the tube or the shell side of the extraction module is shown in Figure 6.3. The pressure drop in the lumen side can be predicted by the Hagen-Poiseuille equation describing laminar flow in cylindrical tubes (Equation 6.1 ). 181 Chapter6 Purification of cerium with hollow fibre contactors 4 Q= 1lr; flp 6.1 8µL where the pressure drop (~p) is dependent on the flowrate (Q), viscosity (µ) and internal radius of the fibre (,i). (McCulloch 1999) measured the effect of contact with kerosene on the Akzo propylene fibres and found that some degree of swelling occured, reducing the internal diameter of the fibre by 3%. Although this change is small, it has a significant effect on the predicted pressure drop at high flowrates due to the fourth dependency of this parameter in Equation 6.1. Provided the swelling of the polypropylene fibres is taken into account, the predicted pressure drop in the lumen is in good agreement with the measured data (Figure 6.3). 1+-Lumen inlet pressure ~ Shell inlet pressure l!_Pressure drop in lumen o Pressure drop i~ shell 40 -.------, 35 30 - • i 25 • ~ ~ 20 ;::l ;::l ~ 15 0.. lO Flowrate (mL min- 1) Figure 6.3 Inlet pressure and pressure drop measurements of water in the extraction module Closed symbols: water in lumen, air in shell, fibres wetted with Shellsoll D70 Open symbols: water in shell, air in lumen, fibres wetted with Shellsol D70 Dotted line: Predicted pressure drop in lumen using equation 6.1 Full line: Predicted presure drop in shell using equation 6.2 182 Chapter6 Purification of cerium with hollow fibre contactors The pressure drop on the shell side was also measured with water in the shell side and air in the lumen. The effect of flowrate on shell side pressure drop is adequately predicted by Equation 6.2, which was used by (Serra 1998) and (McCulloch 1999) to describe shell side flow in a hollow fibre module. 8(1 6 ·)2 where g(c) = - 2 is a function of packing fraction 6.3 2c• - Inc· - 0.5& 0 -1.5 In the stripping module, capillary tubes were used instead of hollow fibres. In this case the internal diameter is much larger and therefore there is negligible pressure drop across the module. Similarly, the pressure drop on the shell side is very low, predicted to be < 0.5 kPa at a flowrate of 1400 ml min·1. 6.3.3 Effect of interfacial tension on critical pressure One of the advantages of using a hollow fibre contactor for liquid-liquid extraction is that the system can easily operate with liquids of similar density and having low interfacial tensions (Gabelman 1999). In conventional solvent extraction circuits these conditions can lead to emulsification and/or flooding. The interfacial tension does however effect the critical pressure at which the non-impregnating phase breaks through the membrane pore. In its most simplistic form, and assuming that the membrane pores are cylindrical in shape, the critical breakthrough pressure can be described by the Young-Laplace Equation 6.4. More complicated relationships which take into account the geometry of the pore have been discussed in the literature (Kim 1987), (Zha 1993). However, in all the expressions used, the critical breakthrough pressure is proportional to interfacial tension. It is expected therefore, that at interfacial tensions close to zero, the critical pressure maybe so low as to render the operation of hollow fibre contactors difficult. M' = 2ycos0 er 6.4 rP where r is the interfacial tension, 0 is the contact angle and rP is the pore diameter. 183 Chapter6 Purification of cerium with hollow fibre contactors The reagent chosen for the extraction of cerium (di-2-ethylhexyl phosphoric acid or DEHPA) is interfacially active. Furthermore, as has already been discussed in Chapter 4, small additions of the mono ester (MEHPA) have a beneficial effect on the rate of cerium extraction, but significantly decrease the interfacial tension. At the optimum conditions for extraction and back-extraction of cerium, the interfacial tension of the solvent system and the feed is very low, at 2 mN m-1• It was therefore thought appropriate to investigate the effect of interfacial tension change on the critical breakthrough pressure, in order to determine whether a practical operating pressure range could be obtained. For these sets of tests, only the extraction module containing hydrophobic fibres was used. Aqueous solution (0.015 M H2S04) was pumped through the lumen of the fibres and the solvent was pumped through the shell. Both the aqueous and the solvent were circulated through reservoirs. The solution in the lumen side was kept at a higher pressure than that in the shell. The solvent flowrate was kept constant and the aqueous flowrate was progressively increased, increasing the pressure difference between lumen and shell. At a certain point, the solvent reservoir became cloudy and aqueous accumulated very rapidly at the bottom of the reservoir, indicating that the breakthrough pressure had been reached. The experiment was repeated with various solvent concentrations. The concentration of DEHPA remained constant, but the MEHPA concentration was changed. The effect of MEHPA concentration on the critical pressure is shown in Figure 6.4. It can be seen that the critical pressure decreases with increasing MEHPA concentration, leveling out at around 0.012 M MEHPA. The minimum ~Per is around 60 kPa, which is sufficiently high to ensure a comfortable operating margin. Data obtained on the effect of MEHPA concentration on the interfacial tension are also plotted in the same figure. In both cases the DEHPA concentration was constant, albeit at different concentration levels. The interfacial tension data follow a similar trend, decreasing to a plateau at around 0.008 M MEHPA. The MEHPA concentration at which the plateau occurs is different, but this has been attributed to the rather crude way of determining the critical breakthrough pressure. 184 Chapter6 Purification of cerium with hollow fibre contactors 140 30 120 25 ";"E- '2' z 0.. 100 C, _ 20 s ~ C ;:l 80 0 t/l ·u; t/l • 15 ~ C Q. 60. Q) "3 • m- .g - 10 ·u ·;::: 40 • u ~ 20 5 Q) • .... _. -=- 0 0 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 MEHPA(M) Figure 6.4 Effect of MEHPA concentration on interfacial tension and breakthrough pressure (~Per) open symbols: [DEHPA] = 0.2 M, diluent Shellsol 2046. closed symbols: [DEHPA] = 0.01 M, diluent n-heptane. l.E-05 -II) ,! ! --- -I ~ E I 0s X :::, u:: l.E-06 0 100 200 300 400 Aqueous flowrate (ml min-1) ···---· Figure 6.5 Effect of tube aqueous feed flowrate on the initial extraction flux Extraction conditions: [DEHPA] = 0.2 M, [MEHPA] = 0.001 M, [Ce]o = 0.001 M 185 Chapter6 Purification of cerium with hollow fibre contactors An interesting observation made during the course of these measurements was that at roughly 10 kPa below the critical pressure, some entrainment of solvent in the aqueous feed could be observed, presumably due to surface shear reaching a critical value to promote entrainment. This indicates that there is an operating region where no breakthrough occurs, but losses of solvent to the feed stream could be considerable due to instability at the interface. This may not always be obvious unless careful measurement has been made of the effect of solvent composition on the critical pressure, thus ensuring that the operating pressures are appropriate to the system being studied. 6.3.4 Effect of tube side flowrate on extraction flux The extraction module was operated with the aqueous feed in the tube side and the solvent in the shell side. In order to determine the optimum operating conditions in terms of flowrate, a set of experiments was undertaken where parameters such as solvent and feed compositions and solvent shell flowrate were kept constant, and the tube side flowrate was changed. The effect of tube side flowrate on the measured overall flux is shown in Figure 6.5. It can be seen that the effect is not very marked and that above 200 ml min·1, the overall flux is independent of tube side flowrate. Work presented in Chapter 5 identified that the major resistance to mass transfer is due to diffusion of the extracted complex in the membrane pore. The chemical reaction was calculated to contribute up to 33% to the overall mass transfer. It would therefore be expected that the lumen side flowrate, which affects only the aqueous diffusion layer, not be a major parameter, as shown by the results presented in Figure 6.5. The equation developed by (Prasad 1992) to describe simple mass transfer, (ie assuming very fast chemical kinetics), in hollow fibres, when the aqueous phase is in the tube and the solvent is in the shell of hydrophobic fibres is shown below. 6.5 This expression is similar to that used to describe simple mass transfer in a flat sheet configuration (Equation 5.22, Chapter 5), in that when Kd >> 1, the term containing kat is expected to be the most significant. However, when the 186 Chapter6 Purification of cerium with hollow fibre contactors membrane thickness is sufficiently large, the term containing km becomes important. The hydrophobic membrane used for the flat sheet membrane studies was 125 µm. The wall thickness of the fibres used for the current tests is 200 µm thick. With an increased membrane thickness, the impact of diffusion in the membrane pore will be enhanced. 6.3.5 Preliminary batch permeation test A preliminary batch permeation test, incorporating the hydophilic stripping unit, was performed to confirm that the operating conditions chosen for the extended run would give the expected fluxes of cerium from feed to strip or receiver solution. The conditions and concentration profiles of the feed and strip solutions are presented in Figure 6.6. Both the feed and strip cerium(IV) concentrations were monitored continously by uv-vis spectrophotometry. The sulfuric acid concentration was much lower in the feed than in the strip solution, thus providing a driving force for the mass transfer. Solvent concentrations were chosen according to optimised conditions determined in Chapter 4. The diluent, however, was changed from n-heptane to Shellsol D70 due to the hazard posed by the high volatility and low flash point of n-heptane. Shellsol D70 is a low aromatic, kerosene type diluent used commercially in solvent extraction circuits. It has a relatively high flash point (76°C}. As expected, the cerium was extracted into the solvent in the extraction module. The cerium(IV) feed concentration profile indicates that around 95% of the cerium(IV) was extracted. However, the cerium(IV) concentration in the strip solution was negligibly low for the first 150 minutes and only increased after a second feed batch was put through the circuit. Periodic sampling of the strip reservoir and analysis by ICP showed that the total cerium, (cerium (Ill) plus cerium(IV)), in the strip circuit was much higher than the cerium(IV). These results suggest that there was a chemical reduction of extracted cerium(IV) to cerium(III). Cerium(III) is not detected by the uv-vis method used for continuous monitoring. Furthermore, the reduction must have occurred in the solvent as cerium(III) was also detected in the feed solution (ie around 15% of cerium(IV) in the feed was converted to cerium(III) and remained in the feed). A summary of the results obtained for this preliminary test is shown in Table 6.2 187 Chapter6 Purification of cerium with hollow fibre contactors - Feed - strip Ce(IV) J:,,,. Strip (total Ce) 280 240 200 §_ 160 a. a, 120 S?.. 80 40 o i--.::.::=~tll!!!f!!_~ ~~~ ----'-----1 0 100 200 300 400 Time (min) Figure 6.6 Simultaneous extraction and stripping with two module hollow fibre contactors Feed: [Ce]0 = 0.001 M, 0.2 M H2S04 Solvent: [DEHPA] = 0.2 M, [MEHPA] = 0.0019 M, Shellsol 2046 diluent Strip: 4 M H2S04 6.E-06 ~------. ~ 5.E-06 <:, E 4.E-06 0 .§. 3.E-06 • :,>< U::: 2.E-06 1.E-06 ...... ~~~~~-~~~~~~~ 0 200 400 600 800 [Ce] (ppm) Figure 6.8 Initial fluxes for feed batches 3 - 9 Feed: [Ce]0 = 100 - 675 ppm, 0.2 M H2S04 188 Chapter6 Purification of cerium with hollow fibre contactors The reduction of cerium was not detected in the tests undertaken with only the extraction module and therefore it is assumed that the introduction of the stripping circuit was responsible for the chemical reduction of cerium. Since in the second batch preliminary permeation test, cerium(IV) was detected in the strip, it is assumed that the chemical component in the system that was acting as reducing agent, had been consumed. The flux of cerium from feed to solvent was approximately 35% lower than the flux measured under similar chemical conditions using the flat sheet bulk liquid membrane technique. The work presented in Chapter 5 showed that for the extraction reaction, membrane diffusion controls mass transfer from feed to solvent. Since the hollow fibres used for extraction have thicker walls (200 µm ) than the flat sheets used (125 µm ), the reduction in flux registered is acceptable. The overall flux was not calculated because of the reduction problem already discussed. TABLE6.2 Summary of results obtained in preliminary permeation test % Extraction - Cerium(IV} 95% % Reduction of Ce in Feed 15% Ce Flux from Feed to Solvent 8.2 x 1o-e mol m·2 s·1 % Ce permeation from Feed to Strip 80% % Reduction of Ce in Strip (1 st batch} 100% % Reduction of Ce in Strip (2nd batch} 86% 6.3.6 Continuous run The operation of the extraction and stripping modules was tested in a continuous run of 65 hours duration. The main aims of this testwork were to investigate: • Stability of operation; • Flux history; • Solvent losses; and • Degree of purification achievable. 189 Chapter6 Purification of cerium with hollow fibre contactors The preferred mode of continuous operation is with a once through pass of the feed solution. Due to a combination of the size of the extraction module available and the low fluxes measured for the system being studied, it was not practical to operate in this mode. The feed was instead circulated through a reservoir. The feed consisted of a sulfate solution of cerium and other rare earths. When the cerium(IV) concentration in the feed reservoir dropped below 20 ppm (1.4 x 10·4 M), a fresh feed batch with concentration ranging from 100 - 670 ppm (1.4 x 104 M to 4.8 x 10·3 M) was used. The solvent and strip were operated in closed circuits in circulation mode. The solution compositions are shown below: 4 3 Feed : 0.2 M H2S04 , Cerium(IV) 7 x 10 - 4.8 x 10· M Solvent: [DEHPA] = 0.2 M, [MEHPA] = 0.002 M, Shellsol 070 diluent Strip: 2 M HCI, 0.1 M H20 2 Chloride media was used in the strip circuit in order to quantify the amount of sulfate transferred from feed to strip. Hydrogen peroxide was added as a reducing agent to effect the back-extraction of cerium(IV) from the solvent, which otherwise is not stripped by hydrochloric acid. In this test, therefore, only the feed cerium(IV) concentration was monitored continuously as the cerium in the strip solution was present in the reduced trivalent state. 6. 3. 6. 1 Concentration profiles All together ten feed batches were run through the rig at four different cerium feed concentrations. The cerium(IV) concentration in the feed was monitored every 0.5 minute. The complete feed concentration profile history is shown in Figure 6.7. Periodic sampling of the strip and solvent reservoir was also undertaken. The equivalent volume of fresh solution was added to the reservoirs in order to keep the total volume constant throughout the run. The strip solution was analysed by ICP after appropriate dilution. The solvent sample was stripped into an aqueous phase, which was also analysed by ICP. The cerium strip and the solvent concentration profiles are also presented. 190 700 650 600 550 "E 500 Q_ Q_ 450 "--' "O 400 (l) (l) 350 '+- .£; 300 © 250 ~ 200 150 100 + \ \ \ I \ I \ I \. • 50 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 2000 Ia Strip reservoir I [1500 8 ~ CIJaCIJ CD aaaaa a:P aCIJcjJ 0CDaaaa:DaaaaaD CJDlCIDCIJaCD mace [[]CD ·5 1000 0 500 0.. j ~ ~aaaaaaaa _...... ,._1 I I I I I I I I I I I I I I I I I I I I I O ell , 1 1 1 ' 1 • • 1 1 1 1 1 --;I~...... _...... ,,~ 1;-...... """-+...... ""'"-+ ...... ~ , I ,\ I I I 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Running time (h) Figure 6.7 - 1st Graph: Concentration profiles of cerium(IV) in feed and solvent reservoirs(•) 2nd Graph: Concentration profile of cerium(III) in strip reservoir Note: Feed concentrations shown as continuous lines in different colours, reflecting periodic saving of data Chapter6 Purification of cerium with hollow fibre contactors The drawback of batch feed operation is that the system is never at steady state and this is reflected by the humps in the solvent concentration profile. The solvent concentration reached a maximum of 140 ppm, but averaged 58 ppm. No steady accumulation of cerium was found in the solvent. This indicates that the stripping module coped well with the rate of cerium extraction. In other words, with the higher surface area available for stripping compared to extraction, the permeation rate was controlled by flux from feed to solvent. A summary of the percent extraction for every batch and permeation from feed to strip solution is shown in Table 6.3. On average, 94% extraction of cerium(IV) was achieved (calculated be difference of cerium(IV) concentration in the feed and raffinate) The result for the amount of cerium that permeated to the strip solution is much lower at 67%, due to cerium chemical reduction. Cerium reduced to the trivalent state, which reports to the feed, cannot be extracted by DEHPA under the operating conditions chosen for this test. Compared to the preliminary batch test, the amount of cerium reduction in the feed increased from 15% to 25%, which caused a corresponding reduction in the overall cerium permeation from 80% to 67%. This is a direct result of less cerium being available for extraction. The reason for the change in oxidation state of cerium when the stripping circuit was introduced is not clear. Possible explanations are given later in the text, in section 6.4. TABLE 6.3 Summary of results obtained in continuous permeation test % Extraction - % Reduction of % Ce permeation Cerium(IV) Ce in Feed from Feed to Strip Batch 1 96 26 67 Batch 2 84 14 69 Batch 3 96 23 71 Batch 4 94 29 63 Batch 5 95 26 67 Batch 6 93 24 66 Batch 7 97 27 68 Batch 8 97 29 66 Average 94 25 67 192 Chapter6 Purification of cerium with hollow fibre contactors 6. 3. 6. 2 Flux change with time Operation of the two hollow fibre contactors was smooth throughout the 65 hours of operation. The initial pressure settings proved to be easily maintained. Slight problems were experienced with feed batch changes, but this would not be an issue in a continuous operation. The initial fluxes for cerium permeation from feed to solvent have been calculated for each batch using the method of initial slopes discussed in Chapter 5, Section 5.3.2. The fluxes are plotted as a function of initial cerium concentration (Figure 6.8) and the results are approximately 3 x lower than comparative batch tests run in a flat sheet configuration. Part of this decrease can be attributed to the increase in membrane thickness from 125 to 200 µm, but it is likely that there is another factor contributing to the low fluxes observed. A plot of the measured overall aqueous mass transfer coefficient as a function of batch number (Figure 6.9) shows that there was a significant decrease in mass transfer coefficient in the first two batches. There was no further change in subsequent batches. Past batch tests have indicated that the only parameter in the system that has the ability to have an effect of such magnitude is the MEHPA concentration. The MEHPA concentration in the samples taken from the reservoir was accordingly determined and the results are shown in Figure 6.10. It is apparent that the MEHPA concentration in the solvent has been reduced by 50% of its initial value and this decrease has occurred during operation of the first few batches. Thus the trend for MEHPA decrease is similar to that for overall mass transfer coefficient decrease. It seems therefore that a reduction in MEHPA concentration in the solvent phase was responsible for decreased fluxes. The possible reasons for decreased MEHPA concentration are discussed in section 6.4. 193 Chapter6 Purification of cerium with hollow fibre contactors 1--2.SE-06 ,------~ ! 2.0E-06 1;j ~ 1.SE-06 • • • • • 1.0E-06 +------1 0 2 4 6 8 10 Batch Number Figure 6.9 - Aqueous mass transfer constant for batches 1- 9 ---~·-----· 3.0E-02 •• 2.SE-02 :E - 2.0E-02 1.0E-02 O l 2 4 6 8 10 Batch Number Figure 6.10 - MEHPA concentration in the solvent reservoir 194 Chapter6 Purification of cerium with hollow fibre contactors 6. 3. 6. 3 Purification of cerium with respect to other rare earth elements The feed solutions for the continuous test were prepared by dissolution of a cerium depleted mixed rare earth carbonate product derived from the processing of monazite. The mixed rare earth carbonate was dissolved in sulfuric acid and the solution doped with cerium(IV) sulfate. The compositions of the various feed batches are shown in Table 6.4. TABLE 6.4 Purification of strip solution with respect to cerium Batch Running &ce La Ce Pr Nd Sm Eu Gd y Time(h) % mgL·1 Feed* 1,2,3 58.2 289 433 60 222 16.7 3.1 7.1 1.18 4,5,6 47.2 303 291 62 234 17.7 3.3 7.2 1.22 7,8,10 66.8 293 636 63 225 16.8 3.1 7.3 1.23 9 23.2 284 92 56 220 16.8 3.1 6.7 1.14 Strip# 0.6 80.7 0.4 37 6.5 1.5 0.3 0.07 0.4 0.03 Batch 1 4.3 96.6 0.4 248 6.5 1.5 0.3 0.07 0.4 0.03 Batch 2 11.2 98.4 0.4 540 6.5 1.5 0.3 0.07 0.4 0.03 Batch 3 18.3 98.9 0.4 790 6.5 1.5 0.3 0.07 0.4 0.12 Batch 4 25.2 99.1 0.4 972 6.5 1.5 0.3 0.07 0.4 0.35 Batch 5 31.5 99.1 0.4 1015 6.5 1.5 0.3 0.07 0.4 0.44 Batch 6 37.6 99.2 0.4 1227 6.5 1.5 0.3 0.07 0.4 0.74 Batch 7 48.4 99.3 0.4 1439 6.5 1.5 0.3 0.07 0.4 0.94 Batch 8 58.4 99.4 0.4 1705 6.5 1.5 0.3 0.07 0.4 1.11 Batch 9 60.4 99.4 0.4 1744 6.5 1.5 0.3 0.07 0.4 1.11 Batch 10 64.2 99.5 0.4 1805 6.5 1.5 0.3 0.07 0.4 1.17 & Ce % calculated as percentage of total cerium over total rare earths • Cerium in the feed is the Ce4+ state while in the strip it is in the Ce3+ state # Detection limits: La (0.4), Pr (6.5), Nd (1.5), Sm (0.3), Eu (0.07), Gd (0.4), Y (0.03) 195 Chapter6 Purification of cerium with hollow fibre contactors The percent cerium in the feed ranged fron:, 23 - 67%. The ICP analyses of the strip solutions at running times corresponding to the termination of a feed batch are also presented. It is evident that, due to the circulation of the strip solution in a closed loop, the cerium concentration increased with time. The concentrations of La, Pr, Nd, Sm, Eu and Gd in the strip solutions remained at the detection level for this analysis technique for all samples taken during the 65 hour run. The Y concentration increased with time as expected from the trend in the lanthanide extraction with organophosphorous acids, namely, increased extraction with decreasing ionic radius, which translates into higher extraction of heavy rare earths compared to the light rare earths (Sato 1989). These results demonstrate that the solvent system is very selective for cerium(IV), with purity steadily increasing from 80.7% to 99.5% as the cerium concentration is increased. Since the cerium in the stripping circuit is reduced to cerium(III), the cerium concentration in the strip solution is not expected to adversely affect the overall flux of cerium(IV). The upper limit of cerium in the strip solution is therefore only restricted by its solubility, which is in the 100 g L-1 range. It is therefore predicted that this process is capable of producing very high purity cerium chloride solutions. 6.3.6.4 Transfer of su/fate In these tests, extraction was operated in sulfate media and stripping in chloride media. The extraction of cerium into the DEHPA solvent in the extraction module and its subsequent back extraction in the stripping module can be described by Equations 6.6 and 6.7, as discussed in Chapters 3 & 4. - K -- Ce4+ + 2(HR)2 ~ 0·0> >CeR 4 + 4H+ 6.6 6.7 It has also been shown that at high solvent loading, when there is a scarcity of extractant molecules, formation of CeS04R2(HR) 2 occurs according to Equation 6.8. 6.8 196 Chapter6 Purification of cerium with hollow fibre contactors The current test was operated with chloride media in the strip to determine whether sulfate transfer from feed to strip solution occurs in a hollow fibre process. The chemical analyses of S and Ce in the strip solutions are shown in Table 6.5. The results show that the level of sulfate in the strip solution remained constant throughout the run, indicating that no sulfate was being transferred from feed to strip between the first and subsequent 64 hours of operation. The presence of 120 ppm of sulfate in the first sample taken could be due to entrainment associated with the start up period. For the whole duration of the run, the solvent reservoir was very clear, indicating that entrained carry over of feed to solvent did not occur during normal operation. TABLE 6.5 Cerium and sulfur concentrations in the strip reservoir as determined by ICP analysis Running hours Ce s ppm ppm 0.67 37 123 10.3 524 128 20.3 830 128 30.3 1094 142 40.2 1273 127 50.4 1493 124 64.2 1505 118 In conventional solvent extraction, the solvent is operated as close to its full capacity as practicable, due to economic factors, and therefore it is expected that sulfur will be co-extracted into the organic phase as part of the extracted complex. The extraction of cerium(IV) with DEHPA was tested continuously in mini-plant trials using mixer-settlers (Soldenhoff 1999). A copy of this paper is given in Appendix 5. Results from these tests (unpublished) show that sulfate transfer does indeed occur. In these runs the Ce:S molar ratio in the loaded organic phase was measured at approximately 1. If a similar process was occuring in the hollow fibre tests, the S concentration in the strip solution would be expected to increase from 123 to 460 ppm as the 197 Chapter6 Purification of cerium with hollow fibre contactors cerium concentration increased from 37 to 1505 ppm. This is clearly not the case. From the above discussion it can be concluded that no sulfate is chemically transferred from feed to strip in a hollow fibre process, and this is attributed to low solvent loading. 6.3.6.5 Solvent loss to raffinate Solvent losses to the aqueous phases, and in particular to the raffinate solutions, are a very important issue, which is crucial for the future development of hollow fibre technology. In solvent extraction processes, solvent losses can be classified into three categories: • Solvent loss to the aqueous solution due to solubility. This loss is dependent on the type of extractant and diluent used. It is the same for solvent extraction and liquid membrane processes. The only way it can be reduced is by changing the chemical structure of the extractant (for example increasing the length of the alkyl chains). In this regard, hollow fibre contactors may have an advantage, in that a wider range of extractants is available, because there are no constraints due to density differences. • Solvent loss due to evaporation. This is a function of the type of contacting equipment used. It is a particular problem in some type of mixer-settlers, but it is not an issue in packed or pulsed columns. Similarly, the hollow fibre contactors are a closed system, and therefore solvent loss by evaporation is not an issue. • Solvent loss by entrainment. In solvent extraction systems, poor phase separation, emulsification and crud formation are responsible for high solvent loss by entrainment. Liquid - liquid extraction with hollow fibres should considerably reduce solvent losses because the interface is controlled and there is no intimate mixing of the two phases. However, entrainment could be a function of surface shear. The focus of this section is to quantify these entrained losses. The best way to measure entrainment is to take a sample during continuous once-through operation. However, the feed in this run was operated in batches 198 Chapter6 Purification of cerium with hollow fibre contactors and, in this case, the whole batch has to be treated in order for meaningful results to be obtained. The following method was used to quantify the amount of solvent entrained. A known amount of clean diluent was added to all spent feed batches, which were stirred gently and left to stand for one week. Any solvent entrained in the aqueous phase will make its way into the diluent. Soluble extractant, however, will remain in the aqueous phase and will therefore not be included in the measurement. A sample of the equilibrated diluent was then digested with acid and analysed for phosphorous, which is present in the DEHPA and MEHPA. The solvent loss was calculated with the assumption that the diluent loss was proportional to the original ratio of DEHPA/diluent. During proper operation of the rig, there was no visible solvent entrainment. Twice, while swapping feed batches, there was air in the feed line and the feed pump stopped pumping. This resulted in a decrease in the inlet feed pressure less than that of the shell side. The result was solvent breakthrough to the feed. When this occurs, the solvent literally flows into the feed until the proper pressure is restored. The entrainment figures obtained for the various feed batches are presented in Table 6.6. TABLE 6.6 Measurement of entrained phosphorous in spent feed batches (as determined by ICP analysis) Batch No mg P per litre volume of solvent solvent of spent feed entrained per litre of entrainment spent feed (ml) ppm Batch 1 0.2 0.032 32 0.045 45 Batch 3 0.28 <0.019 <19 Batch 4 <0.12 0.071 71 Batch 6 0.44 <0.019 <19 Batch 7 <0.12 <0.12 <0.019 <19 Batch 8 <0.019 <19 Batch 9 <0.12 <0.019 <19 Batch 10 <0.12 199 Chapter6 Purification of cerium with hollow fibre contactors For over 60% of the time, the entrainment was less than that detected by this method (< 19 ppm). At other times, entrainment as high as 71 ppm was registered. These figures are low when compared to the best operated mixer settler solvent extraction circuits, which run at losses of 100-150 ppm, but are not negligible as has been claimed in some literature. Of concern are the periods when problems occur with pressure controls, which can have a significant effect on solvent losses, but these should be controllable. Further studies are required with operation in once-through mode. In addition, better detection limts on analysis of entrained solvent are required in order to better quantify this parameter. 6.4 General discussion The one aspect in the performance of the hollow fibre contactors that remains unexplained is the reason for the decrease in overall mass transfer coefficient in the first few batches of operation, as shown in Figure 6.9. In order to help rationalise such a decrease a few observations are listed: i) The decrease in mass transfer was accompanied by a decrease in MEHPA concentration in the solvent phase (Figure 6.10). ii) In the preliminary batch tests, where sulfuric acid was used as a stripping solution, a chemical reduction of cerium(IV) to cerium(III) was detected. Most of the reduced cerium (85%) reported to the strip solution, and 15% of the reduced cerium reported back to the feed. Since the solvent will only extract cerium(IV), it follows that any cerium reporting to the strip must have first been extracted by the solvent as cerium(IV). Thus reduction is taking place in the solvent phase, or at the interface between solvent and strip. iii) The reduction of cerium in the preliminary tests was only detected after the introduction of the stripping circuit. When only the extraction module was tested, cerium(IV) remained loaded in that state in the solvent. 200 Chapter6 Purification of cerium with hollow fibre contactors iv) In the extended run, the strip solution contained HCI and peroxide, which acts as a reducing agent for cerium. For these conditions, no comments can be made about the oxidation state of cerium in the strip solution. However, the amount of reduced cerium reporting to the feed solution averaged 25% of the total cerium in the feed. v) The total amount of MEHPA lost from the organic in the prolonged run was comparable to the total amount of cerium in the feed reduced to cerium(III). This molar ratio was calculated at 0.96. From the above observations, it would seem that cerium (IV) is oxidising MEHPA and being reduced in the process. Such a reduction was measured with cerium (IV) in 0.75 M H2S04 solution in contact with 0.6 M DEHPA (Chapter 3, Section 3.3.1.3). The cerium (IV) was loaded onto the organic at a concentration of - 10 g L-1 cerium(IV), and it took 25 days for 50% of the loaded cerium to be reduced. The reduction rate of cerium(IV) in contact with MEHPA was not measured, but it is not unreasonable to assume that it would be of a similar order of magnitude. However, the rate of reduction observed in the preliminary batch experiments was very much faster, with an almost instantaneous reduction occurring. Since the reduction of cerium became apparent only when the stripping circuit was introduced, it could be concluded that MEHPA oxidation by cerium(IV) was being catalysed either by acid or some other unknown substance being leached from the fibres. Further testing in this area is required to provide a satisfactory explanation. Regardless of the reason, MEHPA loss can be counteracted by addition of MEHPA to the solvent reservoir to keep the MEHPA concentration constant. This would ensure that the flux was maintained at the highest possible level. A further consequence of MEHPA oxidation is the reduction of an equivalent amount of cerium. The reduced cerium reports to the feed stream and is lost to the raffinate. This loss can be prevented by addition of a strong chemical oxidant, such as potassium permanganate. Both these measures however, would increase the overall processing costs. 201 Chapter6 Purification of cerium with hollow fibre contactors Comparison of performance of conventional solvent extraction and hollow fibre extraction A similar process to that run in the hollow fibres was tested in a conventional mixer-settler circuit. The solvent extraction process development work was undertaken as part of a commercial project. Results were presented at an international conference on rare earths and a copy of the paper is shown in Appendix 5. It is therefore opportune to make a brief comparison between the two techniques. The feed solution to solvent extraction (SX) was 42 g L·1 Ce with a cerium(IV) to total rare earth ratio of 68%. In the hollow fibre (HF) process, the cerium averaged at 0.42 g L·1 Ce with a cerium(IV) to total rare earth ratio of 54%. The feed concentrations in the two techniques differed by a factor of 100. In the SX process, 97% of cerium (IV) was extracted in two stages. In the HF process only 94% cerium (IV) extraction was achieved because the feed batches were changed when the concentration dropped to 20-30 ppm. Lower raffinate concentrations (ie higher recoveries) are achievable by increasing batch recirculation times. The effectiveness of the two extraction circuits can be compared in terms of the number of moles of cerium removed from the feed per unit volume of the extraction circuit. In the extraction mixer-settler circuit, two stages of 120 ml mixer and 480 ml settler cope with 42 g L·1 Ce feed at a feed flowrate of 2.2 L hr1• This translates into a mass transfer rate per unit volume of extracting equipment of 0.15 mol m·3 s·1. The comparative figure for the hollow fibre module (35 ml) with an average mass transfer coefficient of 1.6 x 10-6 m s·1, is only 6 x 10-3 mol m·3 s·1• In this case the SX is more efficient by a factor of 25. However, this ratio can be improved by the following: • Increasing mass transfer coefficient. This can be done by chemical means and by improving the characteristics of the contacting equipment. In the current tests, the average mass transfer coefficient was very low due to the loss of MEHPA. As already discussed this can be counteracted by replenishing MEHPA levels in the solvent. Another factor that can be used to increase mass transfer is to reduce the hollow fibre membrane wall thickness. Commercial 202 Chapter6 Purification of cerium with hollow fibre contactors modules are available with wall thickness of 30 µm, compared to the 200 µm used for these tests. • Increasing the overall efficiency of the hollow fibre process, which can be achieved by increased surface area to volume ratio of the hollow fibre contactors. Commercial hollow fibre modules (with hydrophobic fibres) are available, that have approximately double the surface area per unit volume. If these factors are taken into account then SX and HF mass transfer rates per unit volume of equipment become comparable. An important difference between a solvent extraction process and a liquid membrane process is that in the later process the solvent is run at low loading capacity, because simultaneous extraction and stripping occurs. In fact the solvent loading in the HF tests was less than 0.2 g L-1 during the 64 hours of operation whilst its loading capacity was calculated at 7 g L- 1 (ie less than 3% of its full capacity). By comparison in the SX process the solvent loading was 20.2 g L-1 which translates to 70% of its full capacity. This had important ramifications for sulfate transfer as already discussed. In the SX process sulfate transfer cannot be prevented, whilst it does not occur in the HF process. This work has also highlighted another important difference in the SX and HF cerium processes, which is related to kinetic factors. It has been shown that the presence of MEHPA is essential to improve reaction kinetic and therefore increase overall mass with hollow fibres. However, in the SX process the presence of MEHPA is not a pre-requisite as much higher contact times are available in the mixers. Therefore the problem of MEHPA loss experienced in the HF process is not an issue in the SX process. In terms of purification, both SX and HF easily produced solutions containing Ce/rare earth ratios of >99% with potential for higher purities to be achieved in both cases. The results obtained for the cerium system would indicate that the HF process would have difficulty in competing with an SX process for recovery and purification of concentrated solutions. The reasons for this are system specific, namely the low mass transfer rates measured for this system and the requirement of MEHPA presence which was found to cause reduction of part of 203 Chapter6 Purification of cerium with hollow fibre contactors the cerium feed. A HF process would be more appropriate for treating dilute cerium solutions, where the advantages associated with low solvent losses could outweigh the disadvantages mentioned above. 6.5 Conclusions This work has shown that it is feasible to use hollow fibre contactors for the liquid-liquid extraction and back extraction of cerium from rare earth sulfate solution. The continuous operation of a two module hollow fibre contactor rig for 65 hours highlighted the positive aspects of this technology: • very steady operation, easy to control and with no evidence of membrane fouling; • the use of a stripping module with large surface area relative to extraction prevented accumulation of cerium in the solvent; • low measured entrainment of solvent in the aqueous feed stream; • no sustained carry over of sulfate, either chemically or entrained, from feed to strip solution; • good extractions of cerium(IV) (94%); and • good product purity, with an upgrade of cerium with respect to other rare earths from 23 to 99.5 %. Problems experienced were: • selective loss of MEHPA resulting in decreased flux; and • cerium reduction, possibly associated with the MEHPA loss. The problems experienced could be overcome by addition of MEHPA to the solvent reservoir to keep a constant MEHPA concentration and therefore prevent mass transfer flux reduction. Cerium reduction in the feed can also be reversed by chemical oxidation, such as addition of KMn04 to the feed solution. Both these measures, however, would significantly increase processing costs. 204 Chapter6 Purification of cerium with hollow fibre contactors Data for the same cerium process obtained from continuous operation with mixer-settlers was compared with the hollow fibre data presented in this chapter. The major conclusions from such a comparison are: • The solvent extraction process was more efficient in terms of mass transfer rate per unit volume of contacting equipment. This has been attributed to mass transfer rates being too low (for this specific cerium system) and the physical characteristics of the contacting equipment used for this test. • It is concluded that modules with increased surface area to volume ratio and thinner walled membranes are required for practical application. • For the cerium process discussed, solvent extraction is expected to be more cost effective in treating concentrated solutions, whilst a hollow fibre process becomes more attractive in dealing with feed solutions containing low cerium concentrations. 6.6 Nomenclature Symbol external module diameter (m) hydraulic module diameter (m) internal module diameter (m) internal tube side hollow fibre diameter (m) logarithmic mean diameter of hollow fibre (m) dimeric form of DEHPA overall aqueous mass transfer coefficient (m s-1) equilibrium distribution coefficient aqueous film, tube side mass transfer coefficient (m s-1) organic film, shell side mass transfer coefficient (m s-1) mass transfer coefficient through the membrane (m s-1) L length of fibre (m) ~p difference in operating pressures between tube and shell side (kPa) 205 Chapter 6 Purification of cerium with hollow fibre contactors 8Per critical pressure (kPa) 8p Pressure drop (kPa) Q flowrate (m3 s-1) ,; internal radius of fibre (m) rP membrane pore diameter (m) Greek letters e packing fraction of hollow fibre module y interfacial tension (mN m-1) µ viscosity (kg m-1 s-1) 0 contact angle Acronyms and abbreviations [DEHPA] di-2-ethylhexyl phosphoric acid concentration (mol m-3) HFCLM Hollow fibre contained liquid membrane HFSLM Hollow fibre supported liquid membrane [MEHPA] mono-2- ethylhexyl phosphoric acid (mol m-3) 6. 7 References 1. Acosta A., Valenzuela F., Basualto C., Marchese J. and Campderros M. 1998. Transport of Zn(II), Cu(II), and Ni(II) on hollow fibers-type solid supported liquid membranes. Boletin De La Sociedad Chilena De Quimica 43, no. 4: 401-9. 2. Alonso A. I., Ortiz M. I., Galan B. and lrabien A. 1993. Viability and Stability study of the recovery of Cr(IV) with Hollow Fiber Supported Liquid Membranes. Latin American Applied Research 23: 179-84. 3. Alonso A. I. and Pantelides C. C. 1996. Modelling and simulation of integrated membrane processes for recovery of Cr(VI) with Aliquat 336. J. Membr. Sci. 110, no. 2: 151-67. 206 Chapter6 Purification of cerium with hollow fibre contactors 4. Alonso A. I., Urtiaga A. M., lrabien A. and Ortiz M. I. 1994. Extraction of Cr(IV) with Aliquat 336 in hollow fiber contactors: mass transfer analysis and modelling. Chem. Eng. Sci. 49, no. 6: 901-9. 5. Alonso A. I., Urtiaga A. M., Zamacona S., lrabien A. and Ortiz I. 1997. Kinetic modelling of cadmium removal from phosphoric acid by non dispersive solvent extraction. J. Membrane Sci. 130: 193-203. 6. Breembroek G. R. M., Vanstraalen A., Witkamp G. J. and Vanrosmalen G. M. 1998. Extraction of cadmium and copper using hollow fibre supported liquid membranes. Journal of Membrane Science 146, no. 2: 185-95. 7. Campderros M. E., Acosta A. and Marchese J. 1998. Selective separation of copper with LIX 864 in a hollow fibre module. Talanta 47, no. 1: 19-24. 8. Chiarizia R., Horwitz E. P., Rickert P. G. and Hodgson K. M. 1990. Application of supported liquid membranes for removal of uranium from groundwater. Sep. Sci. Technol. 25, no. 13-15: 1571-86. 9. Coelhoso I. M., Silvestre P., Viegas R. M. C., Crespo J. P. S. G. and Carrondo M. J. T. 1997. Membrane-based solvent extraction and stripping of lactate in hollow-fibre contactors. J. Membrane Sci. 134: 19-32. 10. Dahuron L. and Gussler. E.L. 1988. Protein Extraction with Hollow fibers. A/ChE Journal 34, no. 1: 130-136. 11. Daiminger U., Nitsch W., Plucinski P. and Geist A. 1996. Nondispersive chemical extraction in hollow fiber modules. Value Adding Through Solvent Extraction, Proceedings of /SEC '960. C. Shallcross, R. Paimin and L. M. Prvcic, 2, 1161-66, Melbourne: The University of Melbourne. 12. de Haan A. B., Bartels P. V. and de Graauw J. 1989. Extraction of Metal ions from Waste Water. Modelling of the Mass Transfer in a Supported Liquid Membrane. J. Membrane Sci. 45: 281-97. 13. Escalante H., Alonso A. I., Ortiz I. and lrabien A. 1998. Separation of L phenylalanine by nondispersive extraction and backextraction. Equilibrium and Kinetic Parameters. Sep. Sci. Tech. 33, no. 1: 119-39. 207 Chapter6 Purification of cerium with hollow fibre contactors 14. Gabelman A. and Hwang S. T. 1999. Hollow fibre membrane contactors. Journal of Membrane Science 159, no. 1-2: 61-106. 15. Galan B., Sanroman F., lrabien A. and Ortiz I. 1998. Viability of the separation of Cd from highly concentrated Ni-Cd mixtures by non dispersive solvent extraction. Chemical Engineering Journal 70, no. 3: 237-43. 16. Guha A. K., Yun C. H., Basu R. and Sirkar K. K. 1994. Heavy metal removal and recovery by contained liquid membrane permeator. AIChE Journal 40, no. 7: 1223-37. 17. Hano T., Matsumoto M., Hirata M., Tohda T., Kubata F., Goto M. and Nakashio F. 1996. Extraction of fermentation organic acids with hollow fiber membranes. Value Adding Solvent Extr., [Pap. ISEC'96] 2: 1387-92. 18. Harrington, P. J. 1999. "The removal and recycling of heavy metals using hollow fibre liquid membranes." University of Melbourne. 19. Huang T. C. and Huang C.T. 1988. Kinetics of the extraction of uranium(VI) from nitric acid solutions with bis(2-ethylhexyl)phosphoric acid. Ind. Eng. Chem. Res. 27, no. 9: 1675-80. 20. Hutter J.C., Vandegrift G. F., Nunez L. and Redfield D. H. 1994. Removal of VOCs from groundwater using membrane-assisted solvent extraction. AIChE Journal 40, no. 1: 166-77. 21. Kim B-S. and Harriott P. 1987. Critical entry Pressure for Liquids in Hydrophobic Membranes. J. Coll. lnt. Sci. 115, no. 1: 1-8. 22. Lamb John D., Bruening Ronald L., Linsley David A., Smith Cheri and Izatt Reed M. 1990. Characterization of a macrocycle-mediated dual module hollow fiber membrane contractor for making cation separations. Sep. Sci. Technol. 25, no. 13-15: 1407-19. 208 Chapter6 Purification of cerium with hollow fibre contactors 23. Lee Jae-Chun, Jeong Jinki, Park Jin Tae, Youn In Ju and Chung Hun Saeng. 1999. Selective and simultaneous extractions of Zn and Cu ions by hollow fiber SLM modules containing HEH(EHP) and LIX84. Sep. Sci. Techno/. 34, no. 8: 1689-701. 24. McCulloch J. K., Macnaughton S. and Soldenhoff K. 1999. Hydrodynamic characterisation of hollow fibre modules for liquid-liquid extraction. CHEMECA 99 Conference Proceedings, September 1999, Newcastle Australia, The Institution of Engineers Australia Publications. 25. Ortiz M. I., Galan B., Alonso A. I. and lradien J. A. 1996. Simultaneous Extraction and Back Extraction of Cr(IV) in Hollow Fibre Modules. Value Adding Through Solvent Extraction, Proceedings of /SEC '96D. C. Shallcross, R. Paimin and L. M. Prvcic, 2, 905-10, Melbourne: The University of Melbourne. 26. Prasad R., Frank G. T. and Sirkar K. K. 1988a. Nondispersive solvent Extraction using Microporous Membranes. New Membrane Materials and Processes for SeparationK. K. Sirkar and D. R. Lloyd, 261, 84, AIChE Symposium Series. 27. Prasad R. and Sirkar K. K. 1988b. Dispersion-free Solvent Extraction with Microporous Hollow-Fiber Modules. A/ChE Journal 34, no. 2: 177-88. 28. Prasad R. and Sirkar K. K. 1989. Hollow fiber solvent extraction of pharmaceutical products: a case study. J. Membrane Sci. 47: 235-59. 29. Prasad R. and Sirkar K. K. 1990. Hollow fiber solvent extraction: performances and design. J. Membrane Sci. 50: 153-75. 30. Prasad R. and Sirkar K. K. 1992. Membrane Based Solvent Extraction in Membrane Handbook. Eds W. S. W. Ho and K. K. Sirkar, 727-63. New York: Van Nostrand Reinhold. 31. Rosell A., Palet C. and Valiente M. 1997. Selective separation and concentration of vanadium(V) by a chemical pumping hollow-fiber supported liquid membrane. Anal. Chim. Acta 349, no. 1-3: 171-78. 209 Chapter6 Purification of cerium with hollow fibre contactors 32. Sastre A. M., Kumar A., Shukla J. P. and Singh R. K. 1998. Improved techniques in liquid membrane separations: An overview. Separation and Purification Methods 27, no. 2: 213-98. 33. Sato T. 1989. Liquid-liquid extraction of rare earth elements from aqueous acid solutions by acid organophosphorus compounds. Hydrometallurgy 22: 121-40. 34. Sato Y., Kondo K. and Nakashio F. 1990. A novel membrane extractor using hollow fibers for separation and enrichment of metal. Journal of Chemical Engineering of Japan 23, no. 1: 23-29. 35. Seibert A. F., Py X., Mshewa M. and Fair J. R. 1993. Hydraulics and mass transfer efficiency of a commercial-scale membrane extractor. Sep. Sci. Technol. 28, no. 1-3: 343-59. 36. Sengupta A., Basu R., Prasad R. and Sirkar K. K. 1988. Separation of liquid solutions by contained liquid membranes. Sep. Sci. Tech. 23, no. 12&13: 1735-51. 37. Serra Christophe, Clifton Michael J., Moulin Philippe, Rouch Jean Christophe and Aptel Philippe. 1998. Dead-end ultrafiltration in hollow fiber modules: Module design and process simulation. J. Membr. Sci. 145, no. 2: 159-72. 38. Sirkar K. K. 1997. Membrane Separation Technologies: current developments. Chem. Eng. Comm. 157: 145-84. 39. Soldenhoff K. H., Wilkins D. and Ring R. 1999. The solvent extraction of cerium from sulphate solutions: Mini-plant trials. in Rare Earths '98, Ed. R. C. Woodward, 315-317, pp290-296, Switzerland: Trans Tech Publications. 40. Soler J., Urtiaga A. M., lrabien A. and Ortiz I. 1996. Hollow Fiber Non Dispersive Solvent Extraction for the Recovery of Nickel and Casdmium from Spent Batteries. Value Adding Through Solvent Extraction, Proceedings of /SEC '96D. C. Shallcross, R. Paimin and L. M. Prvcic, 2, 851-56, Melbourne: The University of Melbourne. 210 Chapter6 Purification of cerium with hollow fibre contactors · 41. Tong Yanping, Hirata Makoto, Takanashi Hirokazu and Hano Tadashi. 1999. Back extraction of lactic acid with microporous hollow fiber membrane. J. Membr. Sci. 157, no. 2: 189-98. 42. Tong Yanping, Hirata Makoto, Takanashi Hirokazu, Hano Tadashi, Kubota Fukiko, Goto Masahiro, Nakashio Fumiyuki and Matsumoto Michiaki. 1998. Extraction of lactic acid from fermented broth with microporous hollow fiber membranes. J. Membr. Sci. 143, no. 1-2: 81-91. 43. Valenzuela F., Basualto C., Tapia C. and Sapag J. 1999. Application of hollow-fiber supported liquid membranes technique to the selective recovery of a low-content of copper from a Chilean mine water. J. Membr. Sci. 155: 163-68. 44. Valenzuela F. R., Basualto C., Sapag J. and Tapia C. 1997. Technical note membrane transport of copper with LIX-860 from acid leach waste solutions. Minerals Engineering 10, no. 12: 1421-27. 45. Yang Zhi-Fa, Guha Asim K. and Sirkar Kamalesh K. 1996. Simultaneous and Synergistic Extraction of Cationic and Anionic Heavy Metallic Species by a Mixed Solvent Extraction System and a Novel Contained Liquid Membrane Device. Ind. Eng. Chem. Res. 35, no. 11: 4214-20. 46. Yun C. H., Prasad R., Guha A. K. and Sirkar K. K. 1993. Hollow Fiber Solvent Extraction Removal of Toxic Heavy Metals from Aqueous Watse Streams. Ind. Eng. Chem. Res. 32, no. 6: 1186-95. 47. Yun C. H., Prasad R. and Sirkar K. K. 1992. Membrane Solvent Extraction Removal of Priority Organic Pollutants from Aqueous Waste Streams. Ind. Eng. Chem. Res. 31, no. 7: 1709-17. 48. Yunfeng L., Guangsheng L. and Youyuan D. 1999. Phenol removal from aqueous solution through hollow fibre membrane extraction. Chinese J. of Chem. Eng. 7, no. 2: 104-9. 49. Zha, Fu-Fang. 1993. "Stability and applications of supported liquid membranes." University of New South Wales. 211 Chapter 7 Overall conclusions Summary Overall conclusions of the thesis are presented in this chapter. The potential of hollow fibre contactors for commercial application is discussed. Chapter 7 Overall conclusions 7 .1 Overall conclusions Membrane assisted liquid-liquid extraction of cerium was investigated. Emphasis was placed on the study of the reaction chemistry and the kinetics of non-dispersive solvent extraction and stripping with microporous membranes. A bulk liquid membrane process was developed for the purification of cerium(IV) from sulfate solutions containing other rare earth elements. The cerium process was studied in both a flat sheet contained liquid membrane configuration and with hollow fibre contactors. Di-2-ethylhexyl phosphoric acid (DEHPA) was identified as a suitable extractant for cerium(IV) from sulfuric acid solution, with due consideration of factors such as extraction ability, resistance to degradation, solvent selectivity and potential for sulfate transfer into a strip solution. A detailed study of the extraction of cerium(IV) with DEHPA defined the extraction reaction chemistry of the solvent system. It was shown that there are two distinct regions of cerium extraction. At acidities less than 5 M H2S04, extraction occurs via a cation exchange mechanism, while at higher acidities extraction takes place via a solvating mechanism. In the lower acidity region, formation of two complexes CeR4 and Ce(S04)R2(HRh was proposed, with the former being the dominant species when the distribution coefficient is high. The presence of sulfur in the organic phase was confirmed experimentally, and it was found that the S:Ce ratio increased with increased level of solvent loading. The type of complex formed in the solvent phase was found to be important in practice, and illustrated one of the differences between conventional liquid liquid extraction and bulk liquid membrane technology using hollow fibre contactors. Conventional solvent extraction circuits need to be run at high solvent loadings, which in this case favour the formation of the Ce(S04)R2(HR)2 complex, thus providing a chemical means for sulfate to be transferred from feed to strip or receiver solution. With the same process operated in hollow fibre contactors, it was demonstrated that sulfate transfer does not occur due to the low loading of the solvent. The Ce/DEHPA/sulfate system was also investigated with a flat sheet bulk liquid membrane configuration, using both sulfuric and hydrochloric acids as a 212 Chapter 7 Overall conclusions receiver or stripping solutions. These tests identified that hydrophobic membranes provide better mass transfer for extraction and hydrophilic membranes are better for stripping. However, severe leakage of receiver solution to the solvent phase was observed, due to the lack of pressure control at the membrane interface. Such control was available in the hollow fibre configuration, where no leakages of one phase to another were observed. The presence of an impurity, mono-2-ethylhexyl phosphoric acid (MEHPA), was found to have a dramatic accelerating effect on the rate of the chemical extraction reaction. This was attributed to its higher interfacial activity and population compared to DEHPA, and the fact that MEHPA was also found to be an active carrier for cerium(IV). The low concentration at which MEHPA was effective, together with its interfacial and extractant characteristics, suggests that MEHPA could be acting as a phase transfer catalyst. The mass transfer rate of membrane assisted extraction of cerium with a hydrophobic microporous membrane was studied using a modified Lewis-type cell. The stripping process was also investigated with a hydrophilic membrane. Four main resistances were identified as contributing to the overall resistance to mass transfer for both the extraction and back-extraction. These were: • resistance due to diffusion through the aqueous boundary layer; • resistance due to chemical reaction; • resistance due to diffusion through the membrane pores, and • resistance due to diffusion through the boundary organic layer. The hydrodynamics of the membrane permeation cell were characterised by the study of simple mass transfer of iodine and acetone, which involves no chemical reaction. These data were used to describe the diffusion process occurring in the aqueous and organic boundary layers. This method allowed for the quantitative calculation of the contribution of the various diffusion processes to the overall mass transfer of cerium, and the derivation of an empirical rate equation for the chemical extraction and stripping reactions. It was found that, for the range of experimental conditions studied, and provided MEHPA was present in sufficient quantities, the liquid-liquid extraction process for cerium is mainly controlled by membrane diffusion. In practical 213 Chapter 7 Overall conclusions terms this means that a reduction in membrane thickness is the most effective way of increasing mass transfer from feed to solvent. The accelerating effect of MEHPA was quantified and it was shown that the chemical reaction rate becomes important when the MEHPA concentration is very low. The stripping process was found to be mainly controlled by the chemical reaction rate. However, where the distribution coefficient is low, membrane diffusion was found to contribute up to 40% to the overall resistance. Since efficient stripping requires low distribution coefficients, minimising membrane thickness is also important in the stripping circuit. The work has highlighted the importance of understanding the fundamental processes occurring in a particular extraction system, so that the appropriate design can be applied to a potential process. In the present cerium(IV)/sulfate/DEHPA/MEHPA system, it has been shown that the most important parameter for extraction is the thickness of the membrane, whilst for stripping, it is important to increase the surface area available stripping relative to the surface area available for extraction. Furthermore, in the modelling of membrane assisted extraction, it has often been assumed that instantaneous chemical equilibrium occurs at the interface. This work has shown that this assumption is not valid for the system studied. Finally, two hollow fibre contactors were operated continuously for 65 hours, highlighting the positive aspects of this technology. The configuration of two modules, rather than one, was chosen because of the need for using two different types of fibre (hydrophobic for extraction, hydrophilic for stripping) and the need for larger surface area in stripping relative to extraction. Construction of a single module in a contained liquid membrane configuration in this case was technically complicated. The main findings of the continuous test were: • the use of a stripping module with large surface area relative to extraction prevented accumulation of cerium in the solvent; • low measured entrainment of solvent in the aqueous feed stream; • no sustained carry over of sulfate, either chemically or entrained, from feed to strip solution; • high extractions of cerium(IV) (94%); and 214 Chapter 7 Overall conclusions • good product purity, with an upgrade of cerium with respect to other rare earths from 23 to 99.5%. Problems experienced were: • selective loss of MEHPA resulting in decreased flux; and • cerium reduction, possibly associated with the MEHPA loss. The problems experienced could be overcome by addition of MEHPA to the solvent reservoir to keep a constant MEHPA concentration and therefore prevent mass transfer flux reduction. Cerium reduction in the feed can also be reversed by chemical oxidation, such as addition of KMn04 to the feed solution. Both these measures, however, would significantly increase processing costs. Data for the same cerium process obtained from continuous operation with mixer-settlers were compared with the hollow fibre data presented in this chapter. This analysis showed that the solvent extraction process was more efficient in terms of mass transfer rate per unit volume of contacting equipment. This has been attributed to mass transfer rates being too low (for this specific cerium system) and the physical characteristics of the contacting equipment used for this test. In the case of cerium extraction and purification, it is considered that the major requirement in terms of application of this technology lies in the need for increased flux. This can be provided by a decrease in the wall thickness of the membrane. In addition, increased surface to volume ratio, whilst not increasing flux, does in part compensate for low flux. Commercial modules (hydrophobic fibres) are currently available containing double the surface area to volume ratio and much thinner membrane walls (30 µm compare to 200 µm) than those used for this study. Similar developments in chemical resistant hydrophilic modules are not yet commercially available and would provide a major breakthrough. This work has shown that with the development of hydrophilic modules to match the performance of the best commercial hydrophobic modules, hollow fibre technology becomes an attractive option for cerium extraction and purification, particularly when feed solutions contain relatively low concentrations. 215 Chapter 7 Overall conclusions 7 .2 Recommendations for future studies • In terms of the cerium extraction process, pilot plant studies with commercial hollow fibre modules are recommended, provided equipment is available that meets the requirements discussed in the previous section. • The pilot studies should include the development of a model to predict mass transfer in hollow fibres. The model should include kinetic terms to describe the extraction and stripping reactions, since the stripping reaction kinetics, particularly, have been shown to be an important parameter. Issues arising from the work covered by this thesis that are worth exploring in the context of other metal extraction systems include: • the potential of MEHPA to accelerate extraction kinetics of other metal species, and • systematic analysis of the role of reaction kinetics in mass transfer with hollow fibres. 216 Appendices Appendix 1 Calculation of equilibrium hydrogen ion concentration in a sulfuric acid solution Appendix 2 Calculation of hydrogen ion activity as a function of sulfuric acid concentration Appendix 3 Options for the recovery of cerium by solvent extraction Appendix 4 Examples of calculations of interfacial concentrations Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials Appendix 1 Calculation of equilibrium hydrogen ion concentration in a sulfuric acid solution APPENDIX 1 1.1 Calculation of equilibrium hydrogen ion concentration in a sulfuric acid solution (Skoog D.A. and West D.M. 1976) Following is the procedure used for calculation of the equilibrium hydrogen ion concentration [H+], given a total molar analytical concentration of sulfuric acid as [H 2S04]0 . The first proton in sulfuric acid is totally dissociated according to Equation A 1.1. A1.1 The dissociation of the second proton is governed by Reaction A1.2. A1.2 where [H+] 2nd = [SO;-] The total hydrogen ion concentration is made up of the contribution due to the first and second dissociation as follows: A1.3 The mass balance of S requires that: A1.4 Incorporating A1.3 into A1.4: A1.5 Appendix 1 / Page 1 Appendix 1 Calculation of equilibrium hydrogen ion concentration in a sulfuric acid solution Substituting A 1.3 and A 1.5 into A 1.2 [H+]([H+]-[H2 S~4 ] 0 ) = K 2 A1.6 2[H2 S04 ] 0 -[H ] which rearranges into a quadratic equation: with the solution: 2 + ([H2 S04 ] 0 -K2 )+~K/-2K2 [H2S04 ] 0 +([H2S04 ] 0 ) +8K2 [H2 S04 ] 0 [H ] =-----'------2 A1.8 1.2 References 1. Skoog D.A., and West D.M. 1976. Fundamentals of analytical chemistry. Holt Rinehart and Winston. Appendix 11 Page 2 Appendix 2 Calculation of hydrogen ion activity as a function of sulfuric acid concentration APPEND1X2 1.1 Calculation of hydrogen ion activity as a function of sulfuric acid concentration The aim of this calculation is to estimate the hydrogen ion activity ({H+}), given the total equilibrium sulfuric acid concentration (H 2S04] 0 in molar (M) terms. Published data are available (Harned H.S. and Owen 8.8. 1964) for the molal activity coefficients. The following relationships were used to convert molal activity coefficients to molar activity coefficients (Weast R.C. and Astle M.J. (Eds.) 1981): 1000M m=----- A2.1 1OOO p - 98.08M where m = molality, M = molarity, p = solution density and the molar mass of sulfuric acid = 98.08 g mo1-1• d y(Molar) = (1 + 0.001m98.08)-0 y(molal) A2.2 p where d0 = solvent density {H+} = r(Molar) M A2.3 1.2 References 1. Harned H.S., and Owen 8.8. 1964. The Physical Chemistry of Electrolytic Solutions. New York: Reihold Publishing Company, p746. 2. Weast R.C., and Astle M.J. (Eds.). 1981. CRC Handbook of Chemistry and Physics, 61st Edition. Florida: CRC Press Inc., p F-7 and D-154. Appendix 2 / Page 1 Appendix 3 Options for the recovery of cerium by solvent extraction OPTIONS FOR THE RECOVERY OF CERIUM BY SOLVENT EXTRACTION K.H. Soldenhoff Australian Nuclear Science & Technology Organisation Private Bag 1, Menai, NSW, Australia 2234 Paper presented at the International Conference for Solvent Extraction Melbourne, Australia, March 1996 published in the Proceedings to ISEC'96 Value adding through solvent extraction R. Paimin, L.M. Prvcic and D.C. Shallcross University of Melbourne Appendix 3 / Page 1 Appendix 3 Options for the recovery of cerium by solvent extraction ABSTRACT This paper reports the results of an experimental program to examine the use of various commercial reagents for the extraction of cerium (IV) from sulphate solutions. Extractants tested include organophosphorus esters (TOPO, Cyanex 923 & Cyanex 925), organophosphorus acids (DEHPA, lonquest 801 & Cyanex 272) and high molecular weight amine, Alamine 336. The suitability of reagents is assessed in terms of process relevant criteria such as extraction dependence on acidity, selectivity over other rare earths and thorium, stability of reagent towards oxidation and loading characteristics. INTRODUCTION Processes for the recovery of cerium from rare earth concentrates, such as monazite and bastnasite, commonly rely on the chemical properties of cerium (IV) to effect its separation from the other rare earths. Low-grade cerium products contain different impurities, depending on the composition of the starting material. Thus roasting of bastnasite, followed by HCI leaching, produces an insoluble fraction containing up to 10% F (Kilbourn 1992). This material can be further upgraded to a medium purity product (90-98% Ce02) by dissolution and precipitation. The processing of monazite concentrate has to take into account thorium. Monazite concentrate typically contains 8% Th02 compared with < 0.5% Th02 in bastnasite. Caustic digestion is the most common method of converting the rare earth phosphates into hydroxides which are then leached with acid. Thorium is kept in the residue by pH control. The cerium can be separated from the leach liquors by oxidative precipitation to produce a product of medium purity or by oxidation (electrolytic or chemical) and solvent extraction to produce high purity products (>99% Ce02). Most of the literature on solvent extraction processes for cerium deal with extraction of Ce(IV) from nitrate media. Reviews by Liddell & Bautista (1984) and Ritcey & Pouskouleli (1985) indicate that the process has changed little since the early reports (Korpusov 1962) on the extraction of Ce(IV) with tributyl phosphate and back extraction by reductive stripping with hydrogen peroxide. Korpak (1971) used 40% TBP in n-heptane to extract Ce(IV) at 105 g L-1 from 4.5 M HN03. Hydrogen peroxide was used as a reducing agent. A strip liquor Appendix 3 / Page 2 Appendix 3 Options for the recovery of cerium by solvent extraction of purity 99.95% Ce02 was reported. Hafner (1977) described the recovery of Ce(IV) from nitrate liquor (7.4 M HN03) obtained from the leaching of bastnasite. H3B03 was used to complex fluoride. The strip solution was a mixture of 6 M HCI/H202/H3B03. Saleh (1966) recovered spectrographically pure Ce02 using a similar TBP process, with NaN02 as a reducing agent in the strip circuit. In this last process the starting material was monazite, but other processing steps were included to remove Th, Fe, Si and phosphate. A major variation to the TBP solvent extraction process is the use of electrolytic oxidation and reduction. Electrolytic oxidation of Ce(III) in nitrate media in the presence of a mixture of DEHPA/TBP was investigated by Zhang (1981). A method was reported for the separation of Ce from Pm (tracer concentrations) by electrolytic oxidation and extraction using reductive stripping to recover Ce. Ying-Chu Hoh (1987 & 1988) reported on both electro-oxidative extraction and electro-reductive stripping in the Ce-TBP-HN03 system. Limited information is available for the extraction of Ce(IV) from sulphate solutions. Work at the U.S Bureau of Mines (Douglass & Bauer 1959) showed that bastnasite sulphate leach liquors required addition of 6-8 N HN03 in order to extract Ce(IV) with TBP. Sulphuric acid was used for stripping and the cerium was recovered by oxalate precipitation yielding a 99.6-99.8% Ce02 product. More recently, Rhone-Poulenc filed a patent (Horbez 1989) for a process where cerous sulphate is electrolytically oxidised and simultaneously extracted into an organophosphorus acid. The solvent was stripped with 2.5 - 5 M H2S04. The disadvantages of nitrate solutions are the problems associated with their disposal. In this regard, recovery of cerium from sulphate solution would be preferable. Such solutions can be obtained by dissolution of cerium concentrate from bastnasite processing, dissolution of oxidised conversion cake from monazite processing or dissolution of medium purity cerium products. This paper presents experimental data on the extraction of Ce(IV) from sulphate solutions and assesses some of the commercially available solvents. Appendix 3 I Page 3 Appendix 3 Options for the recovery of cerium by solvent extraction EXPERIMENTAL METHODS Extraction data was obtained by contacting aqueous and organic phases in separating funnels. The temperature was controlled at 22°± 0.5 C. Aqueous Ce, Sm, Y, Th and S concentrations were determined by ICP-OES or ICP-MS. Concentrations in the organic phase were determined by two methods: i) by difference of initial and equilibrium aqueous concentrations and ii) by stripping of solvent with 6M HCI or 0.1 M HCI depending on the organic reagent and element to be stripped. Organic reagents were used as supplied, with the exception of Alamine 336, Cyanex 923 and Cyanex 925 which were pre equilibrated with Na2C03 and H2S04. n-Heptane was used as diluent in all cases, except for the extraction of sulphuric acid with TOPO at 40°C. In this case, Shellsol 2037 was used. The following procedure was used to measure solvent degradation. An aqueous cerium(IV) solution was contacted with a solvent at unit phase ratio so that the equilibrium concentration of cerium in the organic was around 2-3 g L-1. DEHPA and TOPO were not pre-equilibrated. Alamine 336 was tested both with/without pre-equilibration with Na2C03 and H2S04. The two phases were kept in contact with stirring, and equal volume samples were withdrawn from both phases at various time intervals. All samples were analysed for Cetotal by ICP-OES and Ce(IV) by addition of ferrous sulphate and back titration with eerie sulphate. RESULTS & DISCUSSION Acidic Extractants The extraction of cerium over a wide range of acidities with organophosphorus acids (HR) such as DEHPA (di-2-ethylhexyl phosphoric acid), lonquest 801 (2-ethylhexyl 2-ethylhexylphosphonic acid) and Cyanex 272 (bis-2,4,4- trimethylpentyl phosphinic acid) is shown in Figure 1. With the organophosphorus acids studied, the order of extraction follows the pKa order, with the lower pKa acid extracting more cerium(IV) at the higher acidities. i.e. DEHPA > lonquest 801 > Cyanex 272 Appendix 3 / Page 4 Appendix 3 Options for the recovery of cerium by solvent extraction The reverse order occurs for the extraction of sulphuric acid, with the stronger base extracting more acid as shown in Table 1. The sulphur concentration in the organic phase was taken as a measure of sulphuric acid extraction. Significant sulphuric acid extraction takes place only at aqueous concentrations > 10 N H2S04. TABLE 1 Extraction of sulphuric acid at A:0=1 and aqueous [H2S04]=21.6 N Reagent [Reagent], M [S1organic, g L-1 DEHPA 0.36 1.80 lonquest 801 0.29 3.78 Cyanex 272 0.22 10.5 Extraction of cerium(III) with DEHPA decreases with increasing acidity according to the well known reaction (1) for trivalent rare earths. (1) Extraction of Ce(IV) with DEHPA occurs from higher acidities, with two distinct regions of extractions; in the first region (- 0.3-10 N H2S04) extraction decreases with increasing acidity, and in the second region (>10 N H2S04) extraction increases with increasing acidity (see Figure 1). At acidities >10 N H2S04 extraction of cerium(IV) is accompanied by extraction of significant amounts of sulphuric acid. In the first region of extraction, uptake of sulphuric acid by the extractant is not significant. However, S was detected together with Ce in the organic phase. The Ce:S ratio was dependent on both acidity and organic loading. At high solvent loading the Ce:S molar ratio was as high as 2: 1. The presence of sulphur in the organic phase suggests the possibility of formation of complexes of both type CeR4(HR)q and CeS04R2(HR)t, with extraction reactions represented by general equations (2) and (3). Appendix 3 / Page 5 Appendix 3 Options for the recovery of cerium by solvent extraction Of the organophosphorus acids, DEHPA is the most suitable for extraction of Ce(IV) from sulphate solutions. High extractions for cerium with lonquest 801 and Cyanex 272 occur at less acidic regions, where hydrolysis can become a problem. At acidities 0.5-2 N H2S04, extraction of sulphuric acid is not significant. It is important to minimise sulphuric acid extraction in order to prevent sulphate contamination of the strip solution. Maximum solvent loading occur with Ce:S:HR ratios of 1:0.5:3. Thus a 25 vol.% DEHPA solution will have a capacity of 35 g L-1 Ce and will carry up to 8 g L-1 S. o DEHPA 111 DEHPA 80 + Cyanex /:J,. lonquest 5 60 ~ I! 1c 40 w ~ 20 0 ....f.-.-,...... ,...... =---.---.--...... ~__,.--,,-.--,..;::,:i;:i,,c:,-:c...,...... ,...... ,..,..,.,., 0.01 0.1 1 10 100 Acidity N Figure 1 Extraction of cerium from sulphate solution with organophosphorus acids DEHPA (open symbol): Organic - 0.29 M DEHPA in n-heptane, Aqueous - 2 g L-1 Ce (IV) DEHPA (closed symbol): Organic - 0.1 M DEHPA inn-heptane, Aqueous - 0.34 g L-1 Ce (Ill) lonquest: Organic - 0.27 M lonquest 801 inn-heptane, Aqueous - 2 g L-1 Ce (IV) Cyanex: Organic - 0.27 M Cyanex 272 inn-heptane, Aqueous - 2 g L-1 Ce (IV) Solvating Extractants Extractants tested were: tributyl phosphate (TBP), dibutylbutyl phosphonate (DBBP), bis(2-ethylhexyl 2-ethylhexyl phosphonate (BEHEHP), trioctyl phosphine oxide (TOPO or Cyanex 921), Cyanex 923 (a mixture of phosphine oxides with n-octyl and n-hexyl groups) and Cyanex 925 (a mixture of phosphine oxides with n-octyl and 2,4,4-trimethylpentyl groups). Appendix 3 / Page 6 Appendix 3 Options for the recovery of cerium by solvent extraction The extraction of cerium (Ill & IV) with TBP has previously been reported to only take place at concentrations > 10 N H2S04 -Douglass (1959). We found no significant extraction of cerium(IV) with either TBP, DBBP or BEHEHP in the 0.5-7 N sulphuric acid range. Cerium(IV) is well extracted with TOPO. On contact with Cyanex 923 & 925, cerium(IV) was reduced to cerium(III), which was not extracted. Pre-equilibration of the reagents with sodium carbonate eliminated this problem in the case of Cyanex 923 but not in the case of Cyanex 925. Extraction of sulphuric acid by some of these reagents (Figure 2) shows that of the three phosphine oxide reagents, Cyanex 925 extracts the least amount of acid, given similar molar reagent concentrations. However, because the problem of reduction of cerium(IV), no further tests were undertaken with this reagent. 10------r- 9 8 80 % Ce Extraction 7 TOPO (0.18 M) 60 ~6 ----<>-- Cyanex 923 (0.47 M) 5 s ~ .!::! 5 I! C S in the organic 1c "' 4 0e> D Cyanex 925 (0.42 M) 40 ~ 0 ~ Cyanex 923 (0.47 M) ![ 3 ';ft. 0 DBBP (1.14 M) 2 ~ TOPO (0.18 M) 20 V TOP0(0.45 M) 40°C 1 0.1 1 10 Acidity (N) Figure 2 Extraction of sulphuric acid and cerium with solvating extractants diluted in n-heptane. With both TOPO and Cyanex 923, extraction of cerium(IV) was found to be independent of acidity in the 0.5-5 N sulphuric acid range (Figure 2). The minimum molar ratio of sulphur to cerium in the organic phase is around two (Figure 3), indicating extraction of Ce(S04)2. Logarithmical plots of the distribution coefficient versus the molar extractant concentration (Figure 4) yield Appendix 3 / Page 7 Appendix 3 Options for the recovery of cerium by solvent extraction straight lines with slopes close to two, pointing to the extraction reaction outlined in equations 5 and 6. (L = TOPO or Cyanex 923) (5) substituting D and K into (5) log D = log K + 2 log [L] (6) 48------, 1 -.------, <> Cyanex 923 (0.42 M) 44 D Cyanex 923 c TOPO (0.18 M) 40 <> TOPO (Ce]aq i = 0.005 M 0.5 o 36 [Ce]aq i = 0.005 M ~ 32 & o 28 C 0 en u 24 .9 -~ 20 en -0.5 0.. 16 ~ 12 slope 2 ::& 8 -1 4o ...... ______..., 0.1 1 10 -1.5------2.2 -2 -1.8 -1.6 -1.4 -1.2 -1 Acidity (N) Log [Extractant] Fig. 3 Molar ratio of S:Ce in the Fig. 4 Logarithmic plot of the organic phase for the distribution coefficient versus extraction of Ce(IV) extractant concentration at 0.2 M H2S04. Both TOPO and Cyanex 923 are suitable extractants for Ce(IV), but it is important to keep the acidity as low as possible to prevent sulphuric acid extraction. The solubility of TOPO in aliphatic diluents places a restriction on maximum loadings. Thus at 20°c, 100 g L-1 TOPO solution will theoretically load a maximum of 17 g L-1 Ce. With this reagent, Ce:S:L ratios in the organic phase at maximum loading is 1 :2:2. Cyanex 923, being a liquid at room temperature, can be prepared at much higher molar concentrations and will therefore achieve higher solvent loadings. Appendix 3 / Page 8 Appendix 3 Options for the recovery of cerium by solvent extraction Selectivity In the purification of cerium, it is important to take into account the selectivity of the reagent for Ce(IV). Table 2 shows the separation factors obtained for Ce(IV) over Th and some of the trivalent rare earths, with selected reagents. Separation factors (~~~HX~, Ln(III) = La, Gd,Yb) published by Preston (1990) for extraction from 2M NH4N03/1M HN03 solution by TBP are of the order 103. With the exception of DEHPA, selectivity of reagents tested compare well with that of TBP. With DEHPA, selectivity of Ce(IV) over the trivalent rare earths decreases along the series light to heavy. In addition, DEHPA extracts Th preferentially to Ce(IV). However, some opportunity exists for the selective stripping of cerium with dilute acid by reducing it to the trivalent state. TABLE 2 Separation Factors for Extraction of Ce(IV), Ce(III), Y(III), Sm(III) and Th(IV) Reagent DEHPA Alamine 336 Cyanex 923 TOPO 5 vol.% 3 vol.% Ce(IV)/Ce(III) > 103 > 103 > 103 > 103 Ce(IV)/Sm(III) 275 > 103 > 103 > 103 Ce(IV)/Y(III) 2.3 103 > 103 > 103 Ce(IV)/Th(IV) 0.053 >102 8.9 14.3 Initial concentrations = 1OOO ppm, [H2S04] = 1 N Solvent Degradation Cerium(IV), being a strong oxidising agent, is potentially capable of degrading the extractant with prolonged contact. TOPO, DEHPA and Alamine 336 were tested for degradation using the procedure outlined in the experimental section. The degradation of the solvent was not measured directly, but the decrease in cerium organic concentration due to reduction of Ce(IV) to Ce(III) was assumed to be an indication of the rate of degradation of the solvent. The decrease in organic cerium concentration was accompanied by an increase in aqueous cerium concentration. Figure 5 plots the decrease in cerium organic loading (ie [Ce]org / [Ce]org maximum) as a function of time. Data obtained by Warf Appendix 3 / Page 9 Appendix 3 Options for the recovery of cerium by solvent extraction (1947) has also been included to provide a basis for comparison with TBP degradation from a nitrate system. Under the experimental conditions shown, the half-life of the solvents (50% decrease in cerium organic concentration) was 12 hours, 25 days, 32 days and 57 days for Alamine 336, DEHPA, TBP and TOPO respectively. Alamine 336 was also tested after equilibration with Na2C03 to remove possible reducing impurities, but 50% Ce(IV) reduction still occurred within 24 hours. Thus on the basis of degradation alone, Alamine 336 is not suitable for extraction of Ce(IV). The degradation rates of the other three reagents are of the same order of magnitude, with TOPO being the most resistant to oxidation. l100 C) C "C 80 ns 0 ...J .!::? C 60 ns e> 0 o TOPO E 40 :::II 'i: a TBP Cl) 0 20 x DEHPA + Alamine 336 0 0 5 10 15 20 25 30 35 40 Time (days) Figure 5 Reduction of cerium loaded onto the organic as a function of time. Alamine: Organic - 3 vol.% Alamine 336 inn-heptane, Aqueous - 5.5 g L-1 Ce in 0.6 M H2S04 DEHPA: Organic- 20 vol.% DEHPA inn-heptane, Aqueous - 5 g L-1 Ce in 0.7 M H2S04 TOPO: Organic - 36 g L-1 TOPO inn-heptane, Aqueous - 5 g L-1 Ce in 0.7 M H2S04 TBP*: Organic - 100% TBP (vacuum distilled), Aqueous - 1 N HN03/0.5 N (NH4)2Ce(N03)5 *Note: TBP data reproduced from Warf (1947) Appendix 3 I Page 10 Appendix 3 Options for the recovery of cerium by solvent extraction CONCLUSIONS Both neutral and acidic organophosphorus reagents (eg DEHPA & Cyanex 923) are suitable for the extraction of Ce(IV) from acidic sulphate solutions. Extraction of sulphuric acid can be minimised, but co-extraction of sulphate as part of the extraction mechanism cannot be prevented. In both instances, strip solutions will contain considerable amounts of sulphate, leading to possible contamination of solids precipitated directly from strip solutions. The selectivity of TOPO and Cyanex 923 for Ce{IV) over the trivalent rare earths and Th is better than that of DEHPA. Extraction of Ce(IV) with Alamine 336 is not recommended due to the fast rate at which Ce(IV) is reduced when in contact with the reagent. REFERENCES Bauer,D.J., 1959. Report 5536, U.S. Bureau of Mines Douglass,D.A., Bauer,D.J., 1959. Report 5513, U.S. Bureau of Mines Hafner,L., 1977. German Patent 2,633,115, Chem. Abstracts, 1977, vol. 86, 124825t Horbez,D., Stork.A., Grosbois,J., 1989. Australian Patent AU-A-31139/89 Kilbourn,B.T., 1992. in Cerium -A guide to its role in chemical technology, Published by Molycorp, NY, U.S.A. Korpak,W., 1971. Polish Patent 2,633,115, Chem. Abstracts, 1971, vol. 76, 268602 Korpusov,G.V., Levin,V.I., Brezhneva,N.E., Prokhorova,N.P., Eskevich,I.V, Seredenko,P.M., 1962. Russian J. lnorg. Chem.,Z(9), 1167 liddell,K.C., Bautista, R.G., 1984. In Hydrometallurgical Process Fundamentals, Nato Conf. Ser., 6,10, Ed. Bautista R., 429 Preston,J.S., Du Preez,A.C., 1990. Proc.lSEC'90. Kyoto.Japan, 883 Appendix 3 / Page 11 Appendix 3 Options for the recovery of cerium by solvent extraction Ritcey,G.M., Pouskouleli,G., 1985. In Science and Technology of Tributyl Phosphate,vol. 11,71 Saleh,F.A., 1966. Z. anorg. allg. Chem., 343,205 Ying-Chu,H., Tsong-Yang,W., Yuh-Yuan,W., Tai-Ming,C., 1987. Hydrometallurgy, 19,209 Ying-Chu,H., Yuh-Yuan,W., Tai-Ming,C., 1988. in Rare Earths, Extraction, Preparation and Applications, Ed. Bautista R.G. and Wong M.M. Warf,J.C., 1947. Tech. lnf. Br., Oak Ridge. Tenn., AECD-2524 Zhang,S., Deng, D., 1982. He Huaxue Yu Fangshe Huaxue (Journal of Nuclear and Radiochemistry), ~(4),243 Appendix 3 / Page 12 Appendix 4 Examples of calculations of interfacial concentrations APPEND1X4 Examples of calculations of interfacial concentrations In Chapter 5, the flux of cerium across a hydrophobic membrane is described in terms of mass transfer coefficients, bulk and interfacial concentrations of reacting species, as shown in Equations A4.1-A4.4. N J = RJ - R, = kce-0.55Msulfuric ( [Ce ]b -[Ce];) A4.1 = ~ K(HR)z-heptane ( [(HR)2 ]b -[(HR)2 l,am) A4.4 1 1 1 where----= +----- A4.5 K<-HR-h-heptane ko((HRh-heptane) km((HRh-heptane) Following are examples on how the interfacial concentrations were estimated. For a given experiment, R, was experimentally determined, given a set of experimental conditions. 3 For example: at bulk concentrations of [(HR) 2 ] = 90molm- (= 0.18 M DEHPA) [Ce] = 4.45 molm-3 (= 624 mg L-1 Ce) R, was measured at 1.94 x 10-5 mol m-2 s-1. Since the experimental flux was determined in the first stages of the reaction where the reverse reaction is considered negligible, R, = N,. Appendix 4 / Page 1 Appendix 4 Examples of calculations of interfacial concentrations The various mass transfer coefficients, presented in Table 5.4, are then incorporated into Equations A4.1-A4.4. [Ce];can be solved from equation A4.1; [Ce]; = 2.64 mol m-3, and the difference between the bulk and interfacial concentration is 41 %. [H+]; can be solved from Equation A4.2. [H+]; = 1102 mol m-3, and the difference between the bulk and interfacial concentration is 0.15%. [(HR) 2 ]; om can be solved from Equation A4.3; 3 [(HR) 2 ];""' = 88mol m- , and the difference between the bulk and interfacial concentration is 1. 9%. [(HR) 2 L,am can be solved from Equation A4.4; 3 [(HR) 2 ]; 0 ,,, = 73mol m- , and the difference between the bulk and interfacial concentration is 19%. Appendix 4 I Page 2 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials THE SOLVENT EXTRACTION OF CERIUM FROM SULPHATE SOLUTIONS MINI-PLANT TRIALS K.H. Soldenhoff, D. Wilkins, R. Ring Australian Nuclear Science & Technology Organisation Private Bag 1, Menai, NSW, Australia 2234 Paper presented at the International Conference on Rare Earths Fremantle, Western Australia, October 1998 published in Rare Earths '98, Ed R.C. Woodward Materials Science Forum Vols. 315-317 (1999) pp 290-296 ©1999 Trans Tech Publications, Switzerland Appendix 5 / Page 1 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials ABSTRACT A new solvent extraction process for the purification of cerium from sulphate media is presented. This process was developed specifically to recover cerium from a residue produced by the processing of concentrate from the Mt. Weld deposit in Western Australia. The liquor treated by solvent extraction contained 62 g L-1 rare earths and the cerium to total rare earth ratio was 74%. Other impurities, including Fe and Th, totaled 2000 ppm. A solvent mixture of commercially available extractants in a low aromatic content diluent was used to extract Ce4+ selectively over the trivalent rare earths. Partial co-extraction of Fe and Th occurred but it was found that these elements were not easily stripped and therefore selective back extraction of cerium was possible. The cerium was stripped from the organic phase by hydrochloric acid and hydrogen peroxide. In continuous counter-current trials two extraction stages and three strip stages were used. A regeneration process was developed to prevent poisoning of the solvent by impurities. In the continuous counter-current trials, 97% Ce4+ extraction was achieved and the Ce to total rare earth ratio was upgraded to >99%. INTRODUCTION The Mt. Weld deposit in Western Australia has a complex rare earth mineralisation [1-2]. The rare earth phosphate minerals, which include monazite, are amenable to conventional caustic cracking followed by hydrochloric acid dissolution of the trivalent rare earths. The presence of the mineral cerianite in the ore, which is unaffected by the alkali attack, results in rejection of a considerable proportion of the cerium to the acid leach residue. The recovery of cerium from a sulphate solution, resulting from the processing of such a residue, is the subject of the current paper. Appendix 5 / Page 2 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials The purification of cerium from other rare earths is generally carried out in nitrate solution [3-7]. In this case, due to economic and environmental reasons, sulphate was used as the media. There have been some fundamental studies on the extraction of Ce4+ from sulphate solution [8], but no publications have dealt with a sulphate based solvent extraction process for cerium. The solvent extraction process discussed in this paper is a novel sulphate based route for the purification of cerium, that was developed in the context of processing of Mt. Weld ore. It was aimed at the production of a medium purity rare earth cerium product (Ce/ln ratio >95%). EXPERIMENTAL Chemical Reagents The extractant used was commercial grade di-2-ethylhexyl phosphoric acid (DEHPA) supplied by Albright & Wilson (USA). The diluent was Shellsol D70, a low aromatic content diluent supplied by Shell (Australia). Continuous Counter-Current Tests Continuous counter-current tests were run on mixer settler units made of PVC. The size of the extraction and strip mixers was 120 ml and 480 ml respectively. The settler volumes were 500 ml and 1450 ml, respectively. The feed flow rate was 2.2 l h-1, and retention times were of the order of 1 minute for extraction and 5 minutes for stripping. Analysis of aqueous solutions was done by ICP-OES and ICP-MS. Aqueous Ce4+ concentrations were determined by addition of excess ferrous sulphate and back titration with eerie sulphate, using ferroin indicator. Determination of rare earth elements in the solvent was by back extraction with a 2M HCI / 0.35 M H202 solution. Appendix 5 / Page 3 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials RESULTS AND DISCUSSION The feed to solvent extraction was a eerie sulphate solution containing 42.1 g L-1 Ce:4+ and 19.6 g L-1 trivalent lanthanides (Ln) including Ce3+. Other cationic impurities including Na, Ca, Sr, Fe, Cr and Th added to 2.0 g L-1. The rare earth feed composition is shown in Table 1. The ratio of Ce4+/Ln in the feed was 68% and that of Ce/Ln was 74%. The aim of the solvent extraction process was to upgrade the liquor to a Ce/Ln ratio greater than 95%. TABLE 1 Feed Composition Element Concentration Element Concentration mg L-1 mg L-1 La 4458 Tb 47.8 Ce3+ 3700 Dy 223 Ce4+ 42100 Ho 1.9 Pr 1483 y 11.3 Nd 5311 Er 51.7 Sm 750 Tm 0.23 Eu 125 Yb 3.29 Gd 3439 Lu 0.25 Choice of extractant and feed acidity It is possible to extract Ce4+ from sulphate solutions with various commercial extractants and the options available have been discussed elsewhere [9]. In this work, di-2-ethylhexyl phosphoric acid (DEHPA) was used as the extractant, in conjunction with a phase modifier and a low aromatic diluent, Shellsol D70. DEHPA extracts Ce4+ by a cation exchange mechanism, at sulphuric acid concentrations below 1.5 M. The other trivalent rare earths are not well extracted by DEHPA at these acid concentrations. It must be noted that only approximately 92% of the cerium in the feed is in the fourth valent state and available for extraction. The separation factors (J3) for Ce4+ over La and Nd at Appendix 5 / Page 4 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials equilibrium acidities ranging from 0.6 - 1.SM H2S04, and solvent loadings of 6 g L-1 Ce, are shown in Table 2. The selectivity for Ce4+ over the light rare earths is better compared to that over the heavier rare earths. However, the heavier rare earths are present in the feed liquor in much lower concentrations. The feed acidity was therefore optimised with respect to Nd extraction and was chosen as 0.75 M H2S04. Table 2 Separation Factors (P) for Ce4+ over La and Nd Feed Acidity Equilibrium Acidity pCe 4+/La pce4+/Nd H2S04 /M H2S04/M 0.38 0.64 294 207 0.42 0.71 >103 236 0.53 0.78 >103 305 0.62 0.89 >103 414 0.78 1.02 >103 571 1.04 1.25 >103 894 1.29 1.49 >103 1093 Extraction and Stripping of Cerium with DEHPA The loading curve for Ce4+ with DEHPA is shown in Figure 1. The data indicate that loading is favourable, with a two stage extraction at 70% loading predicted to achieve more than 95% extraction. Some reduction of Ce4+ to Ce3+ was detected with contact of feed with the solvent and therefore the number of extraction stages was kept to a minimum to reduce overall contact time. Cerium can be stripped from the organic phase at high sulphuric acid concentrations. However, from a processing point of view, chloride solutions are more amenable for precipitation of cerium products. In order to back extract cerium into an acidic aqueous chloride solution, the cerium needs to be reduced. This can be achieved with hydrogen peroxide in a similar fashion to Appendix 5 / Page 5 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials the conventional TBP/ nitrate process [7]. The amount of H202 required is governed by the equation shown below. 4+ 3+ 2Ce + H 2 O 2 ---), 2Ce + 0 2 + 2H+ ------·--- ~~-----· 35 30 25 .-. ..:i ..._,--tll) ...tll) 0 15 a, !:d.. 10 5 0 0 5 10 15 20 25 30 35 40 45 50 55 [Ce ]aq (g/L) Figure 1 Loading Curve for Ce4+ with 25 vol.% DEHPA This relationship is illustrated in Figure 2 and data presented in Table 3, where the maximum predicted cerium concentrations in the strip solutions are closely matched by experimental results. Hydrochloric acid is used to protonate the extractant back to the acid form and provide sufficient anionic counter ions for cerium. Table 3 Effect of [H202] on Maximum Loaded Strip [Ce] [H202] in the .strip predicted max. [Ce) Measured [Ce] M g L-1 g L-1 g L-1 0.22 7.48 61.6 55 0.26 8.84 73.8 72 0.35 11.9 98.0 92 Appendix 5 / Page 6 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials 100 80 .- ~ 60 0" IOI .. 40 x0.35 M H202 ~ o 0.3 M H202 20 o 0.22M H202 0 0 2 4 6 8 10 12 14 16 18 20 (Ce)org (g/L) Figure 2 Stripping of Cerium Loaded Organic with 1-1.3 M HCI Continuous Counter-Current Tests The continuous circuit tested consisted of two extraction stages and three strip stages. Stripping was done with hydrochloric acid and hydrogen peroxide. The following concept was used for producing two grades of strip liquor. Stripping was divided into two circuits. The first strip circuit consisted of a single stage and the second circuit of two stages. Proportionally more of the trivalent rare earths were stripped in the first circuit. The distribution of cerium between the two circuits was controlled by the amount of hydrogen peroxide added in each circuit. The stripped solvent was then returned to the extraction circuit. A portion of the solvent, representing 2.6% of the organic flowrate was periodically regenerated to prevent build-up of impurities in the solvent. Due to the small flows, the regeneration was not run continuously but was done by batch. The circuit was operated continuously for 250 hours. A schematic diagram of the circuit is shown in Figure 3. Details of concentrations for the feed, raffinate, loaded solvent and the two loaded strip solution streams are shown in Table 4. The two strip circuits were operated so that 73% Ce was stripped in the first stage and the remaining Ce was stripped in the second circuit. This was achieved by controlling the strip circuits operating phase ratios as well as the amount of hydrogen peroxide in the strip liquor. Appendix 5 I Page 7 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials Extraction Circuit Strip Circuit Strip Circuit 2 Proto nation 1 '················)\" organic r,::,l"'\tl"'I,::, ············-> Regeneratio Feed HCI/H202 HCI/H202 Raffinate Loaded Strip Loaded Strip 2 1 Figure 3 Schematic Diagram of Continuous Counter-Current Circuit Table 4 Concentration Profiles for Mini-Plant Trials Element Feed Raffinate Loaded Solvent Strip 1 Strip 2 La 4458 4569 11 53 15 Ce3+ 3700 4700 46200 39000 Ce4+ 42100 1100 20290 Pr 1483 1211 8.8 31 9 Nd 5311 5002 28 107 29 Sm 750 744 10 28 12 Eu 125 134 2.7 7.4 3.9 Gd 3439 3308 10 38 13 Tb 48 46 4.3 8.6 7.8 Dy 223 209 6.4 5.1 2.1 Ho 1.9 1.8 0.3 0.2 0.1 y 11 11 3.3 1.6 1.0 Er 52 53 1.2 1.0 0.35 Tm 0.23 0.21 0.046 0.02 0.01 Yb 3.3 3.2 0.25 0.3 0.25 Lu 0.25 0.19 0.034 0.036 0.031 Total Ln3+ 19605 87.3 46482 39094 other 2023 154 107 138 cations Ce/Ln (%) 74.2 99.57 99.39 99.76 Appendix 5 / Page 8 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials The two stage extraction performed well with over 97% Ce4+ extraction being achieved. The overall cerium extraction was lower because cerium present in the trivalent state in the feed is not well extracted. Some of the Ce4+ in the feed was reduced to Ce3+ in the extraction circuit. This is shown by the increase in Ce3+ from feed to raffinate. The amount of cerium being reduced represents approximately 2% of the Ce4+ present in the feed. The stripping of cerium was also very effective with the recycled solvent containing only 260 ppm Ce. This represents 98.7% overall stripping in the two strip circuits. The selectivity of the solvent for Ce4+ over the other rare earths was very good. The Ce4+/Ln ratio was upgraded from 68% in the feed to 99.57% in the loaded solvent. The results, detailed in Table 4, show that splitting the cerium between the first and second strip circuit in the ratios of 73-27%, produces two loaded strip streams with Ce/Ln ratios of 99.39 and 99.76%, respectively. On average, 82% of the other loaded rare earths reported to the first circuit and the remaining 18% reported to the second circuit. These results show that a product> 99% Ce/Ln ratio can be achieved with a two stage extraction and 3 stage stripping. Under the stripping conditions discussed above, (ie Ce split of 73%-27% in the two strip circuits), two separate strip circuits are not justified, as products of very similar purity are produced. It then becomes simpler to operate a single strip circuit containing three stages. However, having two separate strip streams increases the flexibility of producing two grades of products. For instance, when very little hydrogen peroxide is added to the strip liquor in strip circuit 1, the Ln 3+ distribution is still the same (ie 82-18% split in the first and second strip circuit respectively), but the Ce distribution between the two circuits can easily be controlled at 2-98%. Thus, the Ce/Ln ratio in the first loaded strip stream will be 84%, while that of the second strip stream will be > 99.9%. In this case, the first strip circuit is effectively also acting as a scrub circuit. Appendix 5 / Page 9 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials Some elements present in the feed, are extracted into the solvent and do not significantly report to the loaded strip liquor. These include Cr, Fe, Zr and Th. In order to prevent the concentration of these elements from building up in the solvent, an alkaline regeneration process, followed by acid protonation was used on 2.6% of the total solvent flow. The continuous circuit was run long enough for steady state to be achieved. It was found that at that level of solvent regeneration, the load of impurities carried by the solvent was approximately 15% of the total solvent capacity. The regeneration of the solvent is crucial to the whole process, as without it, the solvent would be rapidly poisoned. CONCLUSIONS This work has shown that the use of di-2-ethylhexyl phosphoric for the extraction of Ce4+ from sulphate solution is feasible. With a two stage extraction circuit, and three overall stages of stripping, 97% extraction and greater than 98% stripping was achieved. It has also been demonstrated that with such a circuit, a feed solution can be upgraded from a 67% Ce4+/Ln ratio in the feed to greater than 99% Ce/Ln in the loaded strip. The circuit also has the flexibility of producing up to 99.9% Ce/Ln loaded strip solution by control of the amount of cerium reporting to the first strip circuit. Other impurities in the feed liquor such as Fe, Cr, Th and Zr, were found to partially load on the solvent but not strip with the hydrochloric acid/ hydrogen peroxide solution used in the strip circuits. Regeneration of a small solvent bleed stream successfully prevented the poisoning of the solvent that would inevitably occur. ACKNOWLEDGEMENTS This work is published with the permission of Ansto and Ashton Mining Limited who sponsored the work. Appendix 5 / Page 10 Appendix 5 The solvent extraction of cerium from sulphate solution - Mini plant trials REFERENCES [1] Duncan, R.K., in Geology of the Deposits of Australia and Papua New Guinea, published by AuslMM, Ed. F.E.Hughes, (1990) pp591 [2] Ring,R., Soldenhoff,K. Bellingham, A. et al.,in Proceedings to the XVIII International Mineral Processing Congress, published by AuslMM, (1993) pp1239 [3] Lindell, K.C., Bautista, R.G., in Hydrometallurgical Process Fundamentals, Nato Conf. Ser., 6-10 (1984) pp 429. [4] Ritcey, G.M., Pouskouleli, G., In Science and Technology of Tributyl Phosphate, vol.11 (1985) pp71. [5] Korpak,W., Polish Patent 2,633,115, Chem. Abstracts,(1971) vol. 76,268602 [6] Hafner,L., German Patent 2,633,115, Chem. Abstracts,(1977) vol.86, 124825t [7] Preston, J.S., Cole,P.M., du Preez, A.C., Fox, M.H. and Flemming, A.M., Hydrometallurgy 41 (1996) pp21. [8] Li Deqian, Wang Zhonghuai, Zeng Gunefu and Xue Zhiying, Journal of the Chinese Rare Earth Society, vol.2, No2 (1984) pp10. [9] Soldenhoff, K. H., in Value Adding Through Solvent Extraction, Proceedings to the International Solvent Extraction Conference (ISEC'96), Ed. D.C. Shallcross, R. Paimin, L.M. Prvcic, published by University of Melbourne, vol. 1 ( 1996) pp 469. Appendix 5 / Page 11