1

Electrochemical Studies of Catalysed Aqueous Sulphide Oxidation

A thesis submitted for the degree of Doctor of Philosophy of the University of London and The Diploma of Imperial College

by

Ian Thompson

Department of Mineral Resources Engineering Sept. 1987 Imperial College of Science and Technology University of London LONDON SW7 2BP But it's all right now, In fact it's a Gas Gas Gas!

Mick Jagger (1968) Abstract 3

Abstract

Electrochemical Studies of Catalysed Aqueous Sulphide Oxidation This thesis concerns the mechanism of oxidation of aqueous sulphide solutions in the British Gas Stretford Process, which uses atmospheric oxygen to achieve the partial oxidation of hydrogen sulphide, producing elemental sulphur and water. Hydrogen sulphide is absorbed in an alkaline solution (pH 8.5-9.5) containing vanadium (V) salts and anthraquinone derivatives which act as oxidation catalysts.

The important methods of removing hydrogen sulphide from fuel gases were reviewed, and a detailed description of the Stretford Process was provided.

The thermodynamic data on sulphur species were presented in the form of Eh/pH diagrams, and the literature relating to the oxidation of sulphide solutions was surveyed. The redox behaviour of sulphide and polysulphide solutions were investigated using electrochemical techniques such as cyclic and pulse voltammetry at gold ring-disc electrodes. It was shown that polysulphide species were important intermediates in the oxidation of HS" ions.

The aqueous chemistry of vanadium was described and the electrochemical behaviour of vanadium (V) and (IV) solutions at pH 9.2 were investigated at mercury, carbon and gold electrodes. The electrochemical reduction of vanadium (V) was shown to be irreversible and to lead to vanadium oxide films, rather than to solution species.

The redox chemistry of anthraquinone was reviewed and electrochemical studies were made of the compound anthraquinone 2,7-disulphonate. The reduced species and intermediates were identified using UV-visible spectrophotometry and ESR spectroscopy.

The reduction pathways of oxygen in alkaline solution were reviewed, and the role of hydrogen peroxide as a possible reactive intermediate was investigated.

The Stretford Process chemistry was examined using stopped-flow spectroscopic methods; these enabled the courses of the redox reactions between sulphide solutions and solutions containing the catalysts to be followed.

A mechanism was proposed for the Stretford Process, and possible process improvements were discussed. Acknowledgements 4

Acknowledgements

I would like to thank Dr. G. H. Kelsall for his supervision during the course of this work. Thanks must also go to Dr. T. Ritter from the British Gas London Research Station and to the other staff there who have given me help; Dr. D. Keene, Dr R. Mounce, Dr R. Gibbons, Roy Lowry, Lucien Anthony, and Susan Mahony.

From Imperial College I would particularly like to thank Gordon "the glass" as well as the other technical and academic staff in the Mineral Resources Engineering and the Chemistry Departments. Research is not carried out alone. The present and past members of the research group have both aided my studies and, through their company, made them more enjoyable. Thank you.

I would also like to acknowledge the financial help of British Gas, the Science and Engineering Research Council and last but definitely not least, Corina Thompson. Contents 5

Contents

Abstract 3

Acknowledgements 4

Contents 5

List of Figures 9

List of Tables 12

1. Introduction: The Importance of Sulphide Oxidation 14 1.1 The Removal of Hydrogen Sulphide 14 1.1.1 Absorption by liquids 15 1.1.2 Adsorption by solids 16 1.1.3 Electrochemical oxidation of hydrogen sulphide 17 1.1.4 Aqueous oxidation of hydrogen sulphide 18 1.2 The Stretford Process 19 1.2.1 Historical development 2 0 1.2.2 Operational Problems 21 1.2.3 Mechanistic studies 23 1.3 Objectives of the Present Study 25 1.3.1. Research Approach 2 5

2. Review of Sulphide Oxidation 2 6 2 .1 The Oxidation States of Sulphur 2 9 2.1.1 Sulphide (-II) 2 9 2.1.2 Poly sulphides (-1 to 0) 3 0 2.1.3 Elemental Sulphur (0) 3 2 2.1.4 Polythionates (0 to IV) 3 3 2.1.5 Thiosulphate (II) 33 2.1.6 Sulphite (IV) 3 4 2.1.7 Sulphate (VI) 3 4 2.2 Electrochemical Studies of Sulphide Oxidation 3 4 2.3 Chemical Oxidation of Sulphide using Oxygen 3 7 2.3.1 Rate of Reaction of Sulphide Solutions with Oxygen 3 8 2.3.2 Effect of Temperature and pH on Reaction Rate 3 9 2.3.3 Catalysis of Sulphide Oxidation 3 9 2.3.4 B acterial Action in S ulphide Oxidation 4 0 2.4 The Production of Elemental Sulphur 4 0 Contents 6

3. Sulphide Electrochemistry 41 3.1 Thermodynamic Calculations 43 3.2 Experimental 44 3.2.1 Solution Preparation 45 3.2.2 Electrochemical Instrumentation 46 3.2.3 Electrode Pretreatment 47 3.2.4 Experimental: Ion chromatography 4 8 3.3 Sulphide Voltammetry: Results and Discussion 5 0 3.4 Thiosulphate Voltammetry: Results and Discussion 51 3.5 Polysulphide Voltammetry: Results and Discussion 53 3.6 Ring-Disc Studies: Results and Discussion 56 3.7 Calculated Polysulphide Concentrations vs. Potential 62 3.8 Detection of Polysulphides Using Ion Chromatography 64 3.8.1 Ion Chromatography: Results and Discussion 6 4 3.9 Summary 65

4. Vanadium 66 4.1 Vanadium (V) 67 4.2 Vanadium (IV) 7 0 4.3 Vanadium (V)/(IV) Compounds 72 4.4 Vanadium (III) 74 4.5 Vanadium (II) 7 4 4.6 Vanadium Electrochemistry 75 4.6.1 The Vanadium (V)/(IV) Couple 7 5 4.6.2 Vanadium (TV) reduction 76 4.6.3 The Vanadium (ffl)/(II) Couple 76 4.7 Oxidation of Vanadium (IV) solutions using Oxygen 76 4.8 Vanadium Sulphides 7 7 4.8.1 V3S,V5S4,VS 77 4.8.2 V2S5 78 4.8.3 VS2andVS4 78 4.9 Vanadium-Sulphur Complexes 7 8 4.10 Summary 80

5. Vanadium Electrochemistry 82 5.1 Vanadium Electrochemistry: Experimental 82 5.1.1 Solution Preparation 83 5.2 Vanadium (V) Voltammetry: Results and Discussion 84 5.3 Summary 91 Contents 7

6. Review of Anthraquinone Redox Chemistry 9 2 6.1 Anthraquinone Reduction 9 2 6.1.1 Substituent effects 9 4 6.1.2 Photo-reduction 9 5 6.2 Anthraquinones in the Production of Hydrogen Peroxide 9 6

7. Redox Chemistry of Anthraquinone 2,7-disulphonate 9 7 7.1 Purification of Anthraquinone 2,7-disulphonate 9 7 7.1.1 Analysis of the Purified Material 9 7 7.2 Experimental: Voltammetry 9 8 7.3 Experimental: Exhaustive Electrolysis 9 9 7.3.1 Calculations: Exhaustive Electrolysis 100 7.3.2 Calibration of Exhaustive Electrolysis Apparatus 101 7.4 Voltammetry: Results and Discussion 103 7.5 Exhaustive Electrolysis: Results and Discussion 108 7.6 UV-Visible Spectrophotometry: Experimental 111 7.7 Results and Discussion: UV-Visible Spectrophotometry 113 7.7.1 Spectral Assignments 114 7 .8 ESR Spectroscopy: Experimental 116 7 .9 ESR Spectroscopy: Results and Discussion 117 7 .1 0 ESR Spectral Structure 119 7.11 Summary 120

8. Oxygen Reduction 122 8.1 The Oxygen / Water Couple 124 8.1.1 The Evolution of Oxygen 126 8.2 Hydrogen Peroxide 127 8.3 Superoxides 129 8.4 Experimental 129 8.5 Oxygen Reduction: Results and Discussion 130 8.6 Summary 132 Contents 8

9. The Redox Chemistry of the Stretford Process 133 9.1 Experimental 133 9.1.1 S topped Flow Apparatus 134 9.1.2 Experimental: Measurement of Solution Potential 135 9.1.3 Experimental: Preparation o f51V NMR Samples 136 9.2 Reaction of AQ27DS and HS“: Stopped Flow Results 137 9.2.1 Rate Studies 138 9.2.2 S olution Potential Measurements 140 9.3 Reaction of V(V) and HS“: Stopped Flow Results 142 9.3.1 Vanadium (V) Reduction 144 9.4 Interaction of AQ27DH" ions with Oxygen 147 9.5 Stretford Solution Chemistry: Electrochemical Results 148 9.6 The Stretford Process: Possible Mechanism 150

10. Conclusions 152 10.1 The S(-II)/S(0) Redox Couple 152 10.2 The V(V)/V(IV) Redox Couple 152 10.3 The Anthraquinone/Anthraquinol Redox Couple 153 10.4 The C^/OH- Redox Couple 154 10.5 The Redox Chemistry of the Stretford Process 154 10.6 The Mechanism of the Stretford Process 155 10.7 Concluding Remarks 156

Appendix: Thermodynamic Data Used in Eh-pH Diagrams 158

References 161 Figures 9

List of Figures

1 .1 The Stretford Process. 19

2 .1 Possible valence states of sulphur in aqueous media. 2 6 2 .2 Efo-pH diagram for the sulphur/water system at 298 K. 27 2 .3 . Efo-pH diagram for metastable sulphur system at 298 K. 2 8 2 .4 E^-pH diagram for the sulphur/water system at 298 K. 2 9 (Oxy-sulphur anions not considered.)

3 .1 Eh-pH Diagram for the Au/Cl/S System. 4 2 3 .2 Eh-pH diagram for the sulphur/water system at 298 K. 43 3 .3 Metastable Eh-pH diagram for the S/H20 system at 298 K. 44 3 .4 A Rotating Ring Disc Electrode. 4 4 3 .5 Ion Chromatography Apparatus. 4 9 3 .6 Voltammogram of HS~ on Gold Plated Disc Electrode. 5 0 ([HS_] = 10 mol m-3, pH = 9.2, nth. cycle, 20 mV s_1.) 3 .7 Cyclic Voltammograms of Sodium Thiosulphate. 5 2 ([Na2S2C>3] = 10 mol m~3. 1st Scans. 100 mV s-1. pH = 8.2.) 3 .8 Voltammograms of Polysulphide Solution at a Gold Disc. 5 3 ([Sx] = 1 mol m-3. xav = 2. pH = 8.2. Scan rate 50 mV s-1.) 3 .9 E^-pH Diagram of the Sulphide/Polysulphide System. 5 4 3 .1 0 Voltammograms of Poly sulphide Solution at a Gold Disc. 5 5 ([Sx2~] = 1 mol m~3. xav = 2. pH = 8.2. Scan rate 50 mV s-1.) 3 .1 1 Ring-Disc Voltammetry of Sulphide Solution at Au RRDE. 5 7 ([HS~] = 10 mol m~3. co = 9 Hz. Scan rate = 100 mV s_1.) 3 .1 2 Ring-Disc Voltammetry of Sulphide Solution at Au RRDE. 6 0 ([NaOH] = 1 kmol m“3, [HS"] = 1 kmol m"3, co = 4 Hz) 3 .1 3 Ring-Disc Potential Pulse Study. Au RRDE. 61 ([HS“] = 10 mol n r3, co = 9 Hz. pH 9.3.) 3 .1 4 Poly sulphide Distribution vs. Potential (pH =14) 62 3 .1 5 Polysulphide Distribution vs. Potential (pH = 9) 63 3 .1 6 Ion Chromatography Results. 6 4

4 .1 E^-pH Diagram for the Vanadium-Water System. 67 4 .2 Structure of the V10O286" i°n- 6 9 4 .3 Vanadium (V) Speciation. 7 0 4 .4 Structure of V1804212’- 71 4 .5 Vanadium (IV) Speciation in Solution. 7 2 4 .6 Vanadium (IE) Speciation in Solution. 7 4 Figures 10

5 .1 Hanging Mercury Drop Electrode. 8 2 5 .2 Voltammogram of Vanadium (V) in Borate Buffer at pH 9.2. 8 5 (First Scan, commenced at 0.35 V vs. SHE. 50 mV s_1.) 5 .3 Efo-pH Diagram for the V-H 2O System at 298 K. 8 6 5 .4 Voltammogram of Vanadium (V) in Carbonate Buffer at HMDE. 8 7 (1st. Scan, 50 mV s'1, pH 9.3, [V(V)] = 10 mol n r3, T = 40 °C.) 5 .5 Cyclic Voltammogram of Vanadium (V) on a Gold Electrode. 8 8 (1st. Scan, 50 mV s’1, pH 9.3, [V(V)] = 10 mol n r3, T = 19 °C.) 5 .6 Cyclic Voltammogram of V(V) on a Vitreous Carbon Electrode. 8 9 (1st. Scan, 50 mV s’1, pH 9.3, [V(V)] = 10 mol n r3, T = 20 °C.) 5 .7 Cyclic Voltammogram of VS 43", HS' on a Gold Disc. 9 0 ([VS43-] = 0.2 mol n r3, [HS-] = 0.36 kmol n r3.)

6 .1 9,10-Anthraquinone. 9 2 6.2 Possible Intermediates in the Reduction of Anthraquinones. 9 2 6.3 Anthraquinone 2,7-disulphonate (AQ27DS). 9 4 6.4 Na4 NN'-disulphomethylanthraquinone-2,6-disulphonamide. 9 5

7 .1 Electrochemical Cell Design for Voltammetry Experiments. 9 8 7 .2 Exhaustive Electrolysis Apparatus. 9 9 7 .3 Plot of Log it vs t during the reduction of Fe(CN)g3". 102 7 .4 Cyclic Voltammogram of AQ27DS. 103 ([AQ27DS] = 1 mol m“3, Sweep Rate 5 mVs"1, pH 9.3.) 7 .5 Cyclic voltammetry of AQ27DS at a rotated gold disc electrode. 106 (Scan rate = 20 mVs"1. pH = 9.23. C0 = 0.357 mol m-3.) 7 .6 Plot of i vs. co 1/2 for reduction of AQ27DS. 107 7 .7 Plot of Reduction Potential vs. pH for AQ27DS. 108 7 .8 Plot of Charge vs. Time During the Electrolysis of AQ27DS. 109 (Electrolysis potential = -0.6 V vs. SHE. pH = 9.3.) 7 .9 Plot of Current vs. Time for Electrolysis of AQ27DS. 11 0 (Electrolysis potential = -0.6 V vs. SHE. pH = 9.3.) 7 .1 0 Electrolysis with Linked UV-Visible Spectrophotometry. I l l 7 .1 1 Spectra at 15 C Charge Intervals during AQ27DS Reduction. 113 7 .1 2 Absorbance(330 nm) vs Charge during AQ27DS Reduction. 114 7 .1 3 Electrochemical ESR Apparatus. 116 7 .1 4 Flow Profile accross a Tube. 117 7 .1 5 Normalised ESR signal (S/iiim) vs. V f2^3. 118 7 .1 6 Structure of AQ27DS-" 119 7 .1 7 Actual and Simulated ESR Spectra of AQ27DS-" 120 Figures 11

8.1 Efo-pH diagram of the O 2/H2O System at 298 K. 124 8.2 The Structure of Hydrogen Peroxide. 127 8 .3 E^-pH Diagram for the H20 2/H20 System at 298 K. 128 8.4 Cyclic Voltammograms Showing Oxygen Reduction 130 8.5 Experimental and Calculated O 2 Reduction Currents at a RDE. 132

9 .1 Stopped Flow Apparatus. 134 9 .2 Gold Indicator Electrode for Measuring the Solution Potential. 135 9 .3 Spectra Taken During Reaction between AQ27DS and HS“. 137 9 .4 UV-visible Spectrum of Sodium Polysulphide. pH 9.3. 138 9 .5 Plot of ln(Abs 330 nm) vs. Time During Reduction of AQ27DS. 139 9 .6 Evans Diagram Showing the establishment of a mixed potential. 140 9 .7 Measured and Theoretical Solution Potentials vs Time. 141 9 .8 Spectra Taken During Reaction betweenV(V) and HS~. 143 9 .9 Spectra of 10 mol V(IV) m"3, before and after aeration. 145 9 .1 0 E^-pH Diagram for the V-S-H 2O System at 298 K. 146 9 .1 1 Voltammetry of Stretford Solution During Reduction. 149 Tables 12

List of Tables

1.1 Typical Stretford Solution Composition. 2 1

2.1 The pKa values of Polysulphides. 3 2

4 .1 51V NMR Chemical Shifts of V(V) Species. 6 S 4 .2 Some Known Vanadium Sulphides. 7 7 4 .3 Spectral Summary of Thiovanadates. 8 0

7 .1 UV-Visible Spectral Summary of AQ27DS Reduction. 115

A .l AGf° Values for Vanadium Compounds at 298 K. 15 8 A .2 AGf° Values for Sulphur Compounds at 298 K. 15 9 A .3 AGf° Values for Vanadium Sulphides at 298 K. 160

/ 13

Electrochemical Studies of Catalysed Aqueous Sulphide Oxidation Introduction 14

1. The Importance of Sulphide Oxidation The oxidation of aqueous sulphide species (H 2S, HS', S ") is of considerable technological importance, and sulphide oxidation processes have been devised for many applications. The gas industry has developed methods of removing hydrogen sulphide from fuel gases based on dissolving the gas in an aqueous solution and then oxidising this solution by aeration. Pollution control and effluent treatment processes must decrease the concentration of aqueous sulphide species, which would otherwise cause a loss of the dissolved oxygen in rivers and lakes. Many industries evolve hydrogen sulphide, controlled oxidation of which can produce elemental sulphur; in this way a toxic pollutant can be converted into a saleable by-product.

1.1 The Removal of Hydrogen Sulphide from Fuel Gases When the North Sea gas fields start to become depleted, coal gasification processes to produce methane (Synthetic Natural Gas or SNG) will need to be developed in the United Kingdom. All coal deposits contain some sulphur, present in organic and inorganic forms. Organic sulphur occurs within the coal matrix, in organic compounds such as thiols. Inorganic sulphur occurs predominantly as inclusions of the mineral pyrite (FeS2). Although improved mineral processing techniques can reduce the amount of pyritic sulphur, it is impossible to remove the organic sulphur by physical processing. Attempts have been made to extract the organic sulphur using chemical methods; these methods were reviewed by Eliot [1] who concluded that they have only been partially successful at removing the organic sulphur content and are unlikely to be implemented on an industrial scale in the near future.

Though coal deposits vary greatly, organic sulphur commonly comprises from 30 to 70 % of the coal's total sulphur content. Thus all coals, and gases derived from coals, are likely to contain sulphur compounds for the foreseeable future.

In America the oil price rise of the mid-seventies increased interest in developing natural gas fields containing a high proportion of hydrogen sulphide - sour fields. This led to developments in gas desulphurisation processes.

In 1984 the Gas Research Institute (Chicago, Illinois) initiated an investigation into aqueous sulphide oxidation processes that are used to purify both sour natural gases, SNG and other fuel gases produced from coal. America has large reserves of coal, but some deposits, especially those in the South West, have a high sulphur content. Coal gasification with subsequent gas desulphurisation [ 2] is seen as a means of avoiding the atmospheric pollution that would otherwise result from direct combustion of these high sulphur coals; recently the KILnGAS process for producing clean gas from high sulphur coal has been demonstrated on an industrial scale [3]. The resulting fuel gas could be used, for example, for electricity generation. Introduction 15

Combined cycle electricity generation schemes have been proposed. Coal is gasified and the resulting gas is purified. This gas is then burnt in a gas turbine which generates electricity, and the exhaust gas temperature is still high enough to raise steam to drive a conventional steam turbine. As well as offering reduced atmospheric pollution resulting from sulphur removal, this scheme offers increased generating efficiencies [4].

During coal gasification, the sulphur in coal is converted into hydrogen sulphide. If methane is to be produced, this has to be removed, since it would cause poisoning of the methanation catalysts that are used later in the process. Even if sulphur resistant catalysts were to be developed, its toxicity, the objectionable odour of hydrogen sulphide, and its detrimental effect on steel pipelines [5] would still necessitate its complete removal.

Although other methods have been suggested, such as selectively permeable membranes [6,7], there are at present three basic ways of removing hydrogen sulphide from fuel gases: 1. Absorbing the gas in a liquid. 2. Adsorbing the gas on the surface of a solid. 3. Chemically converting the gas to a less toxic product.

1.1.1 Gas Absorption by a Liquid Gas absorption processes are usually followed by regeneration of the absorbing solution and release of the hydrogen sulphide, which still requires further processing. For example, alkanolamines are widely used for absorbing the 'acid gases' hydrogen sulphide and carbon dioxide [ 8]. Simplified reactions are: HORNH2 + H2S -> HORNH3+ + HS- ( 1.1) HORNH2 + C 02 + H20 HORNH3+ + HC03- (1.2) Heating the solutions reverses the above reactions and regenerates the hydrogen sulphide and carbon dioxide.

The many other liquid absorption processes, which remove hydrogen sulphide and carbon dioxide from gases, have been reviewed recently [9]. The Rectisol process absorbs hydrogen sulphide into cooled methanol under high pressure, and then releases the gas when the pressure is decreased [10], and the Potash Vacuum and Benfield ‘ processes rely on the absorption and desorption of hydrogen sulphide by aqueous solutions of potassium carbonate [ 10,11].

All the above processes suffer from the disadvantage that the acid gases are only concentrated, and not converted into non-toxic compounds. Although carbon dioxide can be vented safely to the atmosphere, hydrogen sulphide cannot, and so it must undergo further chemical treatment. The most common processes involve partial oxidation with atmospheric oxygen to produce elemental sulphur. Introduction 16

There are various ways of achieving this oxidation, the earliest of which was developed last century by Claus, and is still in use today. Hydrogen sulphide is split into two streams. One stream, consisting of a third of the gas, is combusted with air to produce sulphur dioxide: H2S +3/2 02 -> S02 + H20 (1.3) In a subsequent reaction chamber the sulphur dioxide that is produced acts as an oxidising agent for the remaining hydrogen sulphide, and forms elemental sulphur: 2H 2S + S 02 <-> 3/2 S2 + 2 H20 (1.4) Reaction (1.4) is an equilibrium; at the temperature occurring in the hydrogen sulphide combustion furnace, the equilibrium lies to the left. The combustion products must be cooled to about 650 °C in order to shift the equilibrium to the right.

The sulphur vapour is condensed, and after further purification the sulphur can be sold. However, at 650 °C, the equilibrium mixture still contains appreciable quantities of hydrogen sulphide. Repeating the reaction scheme can reduce this quantity further but the thermodynamics of the process militate against the complete removal of hydrogen sulphide.

Increasingly stringent environmental regulations have meant that conventional Claus processes now require tail gas purification units before the off gases can be vented. This has meant that Claus Units have become more complex and expensive, although they are still widely used [ 10].

1.1.2 Gas Adsorption on a Solid Processes involving gas adsorption on solids (followed by chemical reaction), have been widely used to remove the hydrogen sulphide from coal gases. One historically important method used dry iron (III) oxide; for which the adsorption reaction that is quoted by Kohl and Riesenfeld [10] is: ^ e2^3 3 H2S —> Fe2S3 3 H20 (1.5) However, an iron (1H) sulphide phase has never been identified, and it is likely that the adsorption takes place with simultaneous reduction of the iron(III). Thus the phase Fe2S3 may be better regarded as a mixture of FeS 2 and FeS: Fe20 3 + 3H 2S -» FeS2 + FeS + 3 H20 (1.6) Periodically, air is blown through the sulphidised bed. This oxidises the 'Fe 2S3', producing elemental sulphur and regenerating the iron (III) oxide: F^2^*3 + 3/2 0 2 —> Fe203 3 S (1.7) The sulphur forms around the iron (III) oxide particles and eventually prevents further reaction. Fouled beds contain between 40 and 50 % sulphur, which in principle can be recovered. However, the beds were commonly discarded, or combusted to yield sulphur dioxide for sulphuric acid manufacture. Introduction 17

Zinc oxide filters [12] have also been used to remove trace amounts of hydrogen sulphide: ZnO + H2S -> ZnS + H20 (1.8) The sulphidised bed cannot be easily converted back to zinc oxide, so zinc oxide filters are used only when complete elimination of sulphur is required; they are used, for example, as 'guard tubes' to protect catalysts.

Recently silica gel has been suggested as a selective adsorbent to remove hydrogen sulphide from biogas [13], but the process has not been demonstrated on a large scale. Activated carbon filters [14] have also been proposed; oxidation of the adsorbed species produces elemental sulphur, which eventually deactivates the surface. The carbon can be reactivated by contact with steam, but it is difficult to make the process continuous. Again, no industrial applications of this principle have yet been implemented.

1.1.3 Electrochemical Oxidation of Hydrogen Sulphide Hydrosulphide ions can be oxidised at an anode to form either free sulphur, as in reaction (1.9), or polysulphide ions (S 22',S32_,S42“ and S 52") by reactions such as (1.10): HS‘ —> S + H+ + 2e" (!-9) 2 HS- -> S22‘ + 2 H+ + 2 e- ( 1.10) At the cathode, the reduction of protons produces hydrogen; a by-product which can create income to offset the cost of the electrical energy required. 2 H+ + 2e- -> H2 (1.11) A process for the direct electrolysis has been proposed by Bolmer [15], but the sulphur produced can passivate the anode surface; addition of a sulphur solvent at 85 °C has been suggested to prevent this deactivation. A second problem with direct electrolysis is that polysulphide ions can diffuse to the cathode where they can undergo reduction, thus decreasing the current efficiency.

Dandapani, Sharifker and Bockris [16] showed that the use of elevated temperatures (85 °C) and high sodium hydrosulphide concentrations enabled the electrolysis to proceed without passivation; polysulphide solutions were produced that increased in concentration until elemental sulphur precipitated. Cation exchange membranes prevented the polysulphides from reaching the cathode, and high current efficiencies were reported.

Lim and Winnick investigated the electrolysis of hydrogen sulphide when it was dissolved in a molten potassium sulphide/sodium sulphide mixture [17,18]. The electrolysis was operated at a temperature of around 800 °C, and high current densities were achieved on graphite electrodes. The anode compartment was purged with hydrogen, and the sulphur vapour produced reacted with this to re-form hydrogen sulphide. The cell thus acted as an electrochemical hydrogen sulphide concentration Introduction 18

device. They proposed that this process would be suitable for desulphurising gases that would be subsequently fed to molten carbonate fuel cells, which also operate at high temperatures. However, one problem with the process was the absorption of carbon dioxide by the sulphide electrolyte (forming alkali metal carbonates).

Indirect electrolysis has also been suggested; at the anode an oxidant is generated that is capable of oxidising HS' ions in a subsequent chemical step. Kalina and Maas [19] used electrochemically generated iodine (present as 13“ in the iodide solution) as the oxidant: At Anode: 31- —» I3- + 2e* ( 1.12) At Cathode: 2 H+ + 2 e" h 2 (1.13) In Electrolyte: Is' + H2S —» 2H+ + 31- + S (1.14) Overall reaction: h 2s —^ H2 + S (U 5 )

Olson [20] proposed a similar scheme based on the electrochemical production of an iron (HI) complex; hydrogen sulphide was absorbed and oxidised in one vessel and the iron (II) produced re-converted to iron (III) in an external electrochemical cell.

All electrochemical routes suffer from the cost penalty of utilising electrical energy rather than using oxygen as the oxidant. Efficient electrolysis has been claimed using a cell voltage of only 0.5 V on a laboratory scale [16], but this still represents an energy requirement of 837 kWh per tonne of sulphur produced. To date no electrochemical process for H 2S removal has been operated on an industrial scale.

1.1.4 Aqueous Oxidation of Hydrogen Sulphide Aqueous oxidation processes ensure that the hydrogen sulphide is converted into a non-toxic product, and not merely concentrated. These methods offer the advantage that, unlike the Claus Process, they can remove the hydrogen sulphide completely. In contrast to the solid adsorption processes, they are easily adapted for continuous use, and they can utilise the oxidising power of atmospheric oxygen rather than consuming expensive electrical energy.

Aqueous oxidation processes employ solutions which contain oxidising agents capable of producing elemental sulphur from hydrogen sulphide; the reduced solutions are then re-oxidised with air and recycled, so that the oxidising agents complete a redox cycle and are not consumed. The overall reaction for all these process is given by equation (1.16): H2S + 1/2 0 2 -» S + H20 (1.16) Various oxidising agents have been used to catalyse this reaction: iron (III) salts, with a suitable sequestering agent, are used in the Low-Cat process [21]; arsenic (V) compounds are used in the Thylox and Vetrocoke processes [10]; and organic oxidants such as quinones are used in the Perox and Takahax processes [10]. Introduction 19

Amongst these alternatives, the Stretford Process is one of the most commercially successful; the process uses an absorbing solution containing vanadium(V) salts and soluble anthraquinone derivatives.

1.2 The Stretford Process The Stretford process was developed in the 1960's to oxidise hydrogen sulphide in coal gas to sulphur using an aqueous solution. The process was developed at the North West Gas Research Laboratories at Stretford, near Manchester. The process is shown schematically in Fig. 1.1 There are three main components: the absorber, the reactor, and the oxidiser.

The absorber is a gas/liquid contacting device which ensures that any hydrogen sulphide in the gas stream is absorbed into the solution. Since the pKaj of hydrogen sulphide is about 7, and the absorbing solution is buffered at around pH 8.5, the gas is absorbed according to reaction (1.17): H2S(g) <-» H2S(aq) —» HS" + H+ (1-17) The carbonate/bicarbonate buffer solution prevents the protons released from lowering the pH.

Fig. 1.1 The Stretford Process.

In the reaction tank and the oxidiser, the Stretford process achieves the oxidation of this hydrosulphide ion to sulphur: HS“ +l/2 0 2 -> S + OH- (1.18) Air is blown through the solution in the oxidiser and the sulphur produced, which is naturally hydrophobic, is carried to the surface by the rising air bubbles. Thus, aeration serves the dual purposes of oxidising the solution and carrying the sulphur particles to the surface, where the sulphur-containing froth can be skimmed off and Introduction 20 filtered. However, reaction (1.18) proceeds slowly using atmospheric oxygen without a catalyst, and higher oxidation products (such as thiosulphate, sulphite, and sulphate) tend to be produced. This constitutes an effluent problem which would otherwise be absent.

It was found that using a solution containing vanadium (V) salts and anthraquinone derivatives increased the rate of reaction greatly. It is thought [23] that the vanadium (V) salts are responsible for the hydrosulphide oxidation, according to simplified reactions such as (1.19): HS- + 2V5+ + OH- -> S + 2V4+ + H20 (1.19) If this is the case, then the vanadium(V) salts should not strictly speaking be termed catalysts, since they are consumed stoichiometrically according to equation (1.19); however, they are regenerated in a subsequent aeration tank, reaction ( 1.20), and so take part in a catalytic redox cycle: 2V4+ + 1/2 0 2 + H20 -> 2V5+ + 2 OH- (1.20) The anthraquinone derivatives that are added are said to catalyse the reoxidation of vanadium (IV) to vanadium (V) by atmospheric oxygen [23].

The overall reaction occurring in the process is the same as that given for other aqueous oxidation processes: H2S + 1/2 0 2 -> S + H20 (1.16) Since this reaction does not involve the production or consumption of protons, no permanent pH change will occur. However, since gas absorption, hydrosulphide oxidation and vanadium (V) regeneration occur in different vessels, local changes in pH would be expected.

1.2.1 The Historical Development of the Stretford Process The development of the Stretford process of aqueous sulphide oxidation has been reported in some detail by Vasan [24], Moyes and Wilkinson [23], and Nicklin and Holland [25]. The major stages in the history of the process are as follows:

In 1963 the Stretford Process was developed jointly by the North Western Gas Board and the Clayton Aniline Company Limited; it was intended for use on streams of coke oven gas, and utilised a solution containing sodium anthraquinone disulphonate in an alkaline sodium carbonate/bicarbonate buffer. The original process did not use a solution containing vanadium (V) salts. Three plants were built to this design and worked satisfactorily on a feed of gas containing 0.08 % (by volume) hydrogen sulphide, but the hydrosulphide ion concentration in the working solution could not exceed 1.25 mol m“3.

In an attempt to increase the maximum HS" concentration that could be oxidised, and to reduce the reaction time, which was about half an hour for the first generation of plants, various oxidising agents were investigated for use in the process. Vanadium (V) was Introduction 21 chosen for study since it did not to form a sulphide precipitate under plant operating conditions. It was found that the vanadium (V) was an effective oxidising agent for the hydrosulphide, but that the vanadium (IV) species produced could not easily be re­ oxidised by aeration alone. However, if anthraquinone disulphonate salts were present, regeneration of the vanadium (V) was rapid. Using this system, hydrosulphide solutions of concentration 30 mol m-3 could be oxidised in several minutes. This enabled a much smaller volume of liquid to be re-circulated in order to achieve the same gas throughput rates as obtained in the first generation of plants. In this basic form, the Stretford process has remained in use to the present date and there are now over 100 installations worldwide [26]. The composition of a typical Stretford solution is given in Table 1.1 (The Data is taken from Murin et A1 and Mallot [26,27 ]):

Compound kg n r 3 mol m Sodium anthraquinone 2,7-disulphonate (Na2AQ27DS) 3.00 7.3 Sodium vanadate (NaV03) 1.70 32.5 Sodium citrate (Na 3C 02CH2C (0H )C02CH2C 02) 10.00 52.1 Sodium carbonate (Na 2C 03) 6.25 59.0 Sodium hydrogen carbonate (NaHC03) 18.75 223.0 Sodium thiosulphate (Na 2S20 3) (variable) 89.50 360.0 Sodium sulphate (Na 2SC>4) (variable) 40.2 283.0 Sodium thiocyanate (NaSCN) (dependent on HCN in feed)

Table 1.1 Typical Stretford Solution Composition.

Although the Stretford Process has been one of the most successful methods of hydrogen sulphide removal, several problems still exist.

1.2.2 Operational Problems Experienced by Stretford Plants Sometimes elemental sulphur can form in the absorbing vessel, which can build up so as to restrict, and eventually block the gas flow [26]. If the solution is over-oxidised, the production of soluble sulphoxy compounds, primarily thiosulphate (S 20 32") results. The thiosulphate concentration can build up in the solution, and if left unchecked, sodium thiosulphate would precipitate. To prevent this happening, a portion of the solution must be discarded or treated (for example in a fixed salt recovery unit), and fresh liquor added. Thiosulphate production also consumes hydroxide ions: 2 H2S + 2 0 2 + 2 OH" S20 32" + 3 H20 (1.21) Therefore, alkali has to be added continually to the solution to prevent the pH from decreasing.

Fixed salt recovery units prevent the loss of sodium vanadate; working solution that has been withdrawn from the re-circulating circuit is incinerated in a reducing atmosphere. This converts the thiosulphate and thiocyanate into gaseous hydrogen sulphide, ammonia and carbon monoxide; these gases are fed to the sour gas input. Introduction 22

Solid sodium vanadate and sodium carbonate are also produced; which can be used to make up fresh solution. Sodium citrate and sodium anthraquinone disulphonate (Na2AQDS) are both destroyed by the reductive incineration. The Na 2AQDS is expensive and this loss constitutes one of the main operating costs of the process.

In several plants a black solid, containing vanadium, sulphur and oxygen precipitated from the solution. Chemical analyses revealed that samples originating from different plants had different compositions. It was unclear whether the sulphur was chemically combined in the compound, or whether it was physically entrained in the precipitate. It was claimed that the compound precipitated from solution when the pH was allowed to rise above 9, and that a high concentration of carbonate ions promoted precipitation [26]. One solution to this problem was to increase the flow rate to keep particles suspended until they entered the aeration vessel. It was noted that on prolonged aeration the compound redissolved. Complexing reagents have also been added to a number of Stretford solution in the hope of preventing the vanadium from precipitating. Tartrate ions, citrate ions, and di-sodium ethylenediamine tetraacetate (Na 2EDTA) have all been used industrially. These compounds are claimed to form complexes with the vanadium (V) species, but in a recent study Haley [28] contradicted Malott [27], by asserting that no complex was formed between vanadium (V) and citrate ions, and only a weak complex was formed with tartrate ions.

If hydrogen cyanide is present in the feed gas, it can react with elemental sulphur to form the thiocyanate ion: HCN + S + OH- SCN- + H20 (1.22) This reaction consumes alkali, and the presence of thiocyanate is also thought to increase the conversion of hydrogen sulphide to thiosulphate. To combat this, pre­ washing the gas to absorb hydrogen cyanide (for example by contacting the gas with polysulphide solutions) was introduced.

It was found that bacteria present in the Stretford solutions could affect the plant's performance adversely. Analyses showed that most solutions contained 10 12 bacterial cells m-3. The bacteria included autotrophic sulphur bacteria of the genus thiobacillus; these can oxidise thiosulphate in solution to produce sulphuric acid, which consumes the sodium carbonate and bicarbonate. Bacteria cells are also encapsulated in a slime layer which can break away and cause the solution to foam when it is aerated. These problems caused British Gas to investigate the addition of various biocides, which proved effective at controlling the bacterial population. Microbial problems were not apparent in the early coke oven gas desulphurisation plants; this is thought to be due to the biocidal concentrations of thiocyanate (SCN- ) formed from the hydrogen cyanide in the feed [29]. Introduction 23

1.2.3 Mechanistic Studies Working from free energy of formation data by Israel and Meites [30], the calculated value of Eo' for the vanadium (V)/(IV) couple, at pH 8.5, is +0.05 V vs SHE. At this pH the indicated vanadium (V) species are HV 2O73- ions, and the vanadium (IV) species V4O92" ions. The standard reduction potential of the sulphate/hydrosulphide couple at this same pH is only -0.20 V, [31]. This means that vanadium (V) is thermodynamically capable of oxidising hydrosulphide ions to sulphate.

It is clear from the literature that disagreement exists as to the nature of the vanadium (V) and vanadium (IV) species existing in the Stretford solution. Malott [27] argued that the vanadium (V) species HjjVO ^3'11)-, HnV207(4"n)“, V3093" and V 4O124" may all be present in the solution. Habayeb and Hileman [32] suggested V 2094",V3093‘ and V 4O124" are the major species, whilst many workers have simply assumed that the oxidising species is V5+. Malott suggested that V 2052- may be the predominant vanadium (IV) species, but Pope [33] claims that V jg C ^12" is the major vanadium (TV) form in alkaline solution.

Aeration of alkaline sulphide solutions can result in a variety of reaction products, as Kuhn and Kelsall [34] pointed out in their review. The rate of reaction and the oxidation products are highly dependent on the pH, the solution potential (determined by the dissolved oxygen content) and the temperature. Catalysts can further change the product mixture.

Andrzheevskii [35] undertook a study of the mechanism of the oxidation process occurring in the Stretford Process. He studied the oxidation of sodium sulphide solutions at pH 9.0 using three solutions containing: 0.1 kmol m-3 sodium anthraquinone 2,6-disulphonate (Na2AQ26DS), 0.1 kmol m-3 sodium metavanadate (NaVC>3 ), and a mixture of composition similar to that used in working Stretford plants (12.5 mol Na2AQ26DS m-3, 40 mol NaVC >3 m"3)- A set of experiments were conducted in the absence of air or dissolved oxygen, adding sulphide to the above solutions to make the concentration 50 mol HS_ m-3. The reaction was followed by monitoring the remaining hydrosulphide concentration. He analysed the samples using two methods: a potentiometric titration using mercury (II) nitrate; and a 'chemical' method using cadmium acetate in acetic acid. Unfortunately, no experimental details of either analysis method were given. Andrzheevskii found that the hydrosulphide concentrations as measured by the two methods were identical initially, but as the reaction proceeded, the cadmium acetate method indicated a rapid removal of HS- (decreasing from 50 mol m-3 to zero in 10 minutes), whereas the potentiometric titrations suggested a much slower rate of reaction (after 10 minutes 40 mol HS- m-3 remained). Introduction 24

Andrzheevskii argued that this was evidence for the HS- being present in two forms, one free in solution, the other in a complex with vanadium. He proposed that the lowering of the sample pH during the cadmium acetate analysis caused rapid oxidation of the complexed hydrosulphide, while the potentiometric titration recorded the total [HS-] in both complexed and uncomplexed forms.

Without details of Andrzheevskii's analysis techniques it is difficult to interpret his results. However, Boulege [36] utilised mercury (II) chloride solutions to titrate solutions containing HS-, Sn2_, S 2O32-, and SO 32-. He states that at pH 13.0 the first end point (detected by a sulphide selective electrode) corresponds to the completion of the two reactions: Hg2+ + HS- -> HgS + H+ (1.23) Hg2+ + Sn2- -> HgS + (n-l)S (1.24) This end point is detected by the sharp decrease in potential as the sulphide selective electrode responds to the decrease in [S2-]. If the pH is then adjusted to 7-8, two further end points can be detected corresponding to equations (1.25) and (1.26): Hg2+ + 2S 20 32- -> Hg(S 20 3 )22- (1.25) Hg2+ + 2 S 0 32- -> Hg(S0 3)22- (1.26) These end points are detected because free Hg2+ ions in solution are known to interfere strongly with the response of the electrode, which causes a further decreases in potential.

Andrzheevskii carried out his potentiometric titration at pH 9.0, and so it is not clear which reactions occurred. Certainly the total sulphide and polysulphide concentrations will be recorded, and he may also have titrated dissolved S 2032- and SO32-. His analysis method using acidified cadmium acetate almost certainly involves precipitation of cadmium sulphide. Without experimental details, it is unclear how polysulphide ions, thiosulphate and sulphite ions will react; it may be that they are oxidised during the analysis. Thus these species may account for the discrepancy between his two analyses, and his claim that some of hydrosulphide must be present in the form of a complex must be regarded as speculative; however, Harrison and Howarth have recently obtained 51V NMR evidence for complex formation between HS“ and vanadium (V) [37].

Andrzheevskii's work reveals some interesting observations: oxidation of HS* using stoichiometric amounts of sodium metavanadate or Na2AQ26DS (assuming one electron reduction in both cases) proceeds relatively slowly, in around 90 minutes; a mixed solution (containing a third molar excess oxidising power) oxidised the solution much more rapidly, in about 15 minutes, and when air was admitted oxidation proceeded even more rapidly. If the same solution was used repeatedly to oxidise samples of hydrosulphide solution, with oxygenation used to restore the oxidising power, as in the industrial process, Andrzheevskii noted that the solution lost its ability Introduction 25 to oxidise the HS~ ions in the absence of air. Over 60 minutes, no loss of HS~ was detected (using the potentiometric titration). However, if oxygen was then admitted, complete oxidation was effected in 10-12 minutes. It was stated - without the supporting evidence, that prolonged oxygenation of these solutions for 40 minutes did not regenerate detectable amounts of vanadium (V), although this contradicts other workers [26]. No detailed product analyses were given in Andrzheevskii's work, but the following generalisations were offered: 1. Using solutions of NaV 03 or Na2AQ26DS with oxygen as the oxidising agent, thiosulphate was the main oxidation product. 2. Using a mixed solution with oxygen, elemental sulphur was the main oxidation product.

In 1984 a Research programme was started by the Gas Research Institute (Chicago: USA) into the Stretford Process [26]; they have built a bench-scale circulating flow unit in the hope of determining the optimum operating conditions for a Stretford Plant. They also acknowledged that the complex chemistry of the process is not well understood, and considered that the key to improving the process performance lies in a better understanding of the basic chemistry.

Thus, previous workers have not demonstrated unambiguously the mechanism of oxidation of hydrogen sulphide in the Stretford Process, although many studies of the process have been undertaken. Since 1963, some operating problems have been overcome by using a practical experimental approach, but little of the fundamental chemistry has been elucidated.

1.3 Objectives of the Present Study The aim of the present study is to elucidate the reaction mechanism, in the hope that this can help both to solve operational problems (such as the formation of thiosulphate and the precipitation of vanadium salts) and to point the way towards future process improvements.

1.3.1. Research Approach The approach taken was to investigate the redox couples involved in the process separately and then to consider the interactions between these couples. The couples; S(-II)/S(0), V(V)/V(IV), anthraquinone/anthraquinol and 0 2/H20 were studied using electrochemical techniques such as cyclic and pulse voltammetry. Products were identified using UV-visible and esr spectroscopy. Finally, the redox chemistry of the process was investigated by using stopped flow spectrophotometry and conducting small scale batch experiments on solutions containing two or more of the redox couples. Review of Sulphide Oxidation 26

2. Review of sulphide Oxidation Sulphur has a range of oxidation states from -II to VI:

-II 0 II IV VI h 2s s S20 32- S 032- S042- HS- h s o 3" h s o 4_

Fig. 2.1 Possible valence states of sulphur in aqueous media, ('per' compounds are omitted)

Sulphide solutions represent the lowest oxidation state of sulphur, and in theory they can be oxidised to any of the higher states. However, only the -II, 0, and +VI states are thermodynamically stable in aqueous solution at normal temperatures and pressures. Balanced redox reactions between the various species can involve the production or consumption of protons, e.g. H2S S + 2H+ + 2 e - (2 . 1) Thus at high pH's the forward reaction is favoured, i.e. the oxidation can be achieved be using a relatively low potential. Certain acid-base equilibria may also^mportant, e.g. H2S(aq) <-> HS- + H+ (2.2) HS- <-> S2- + H + (2.3) The values for the equilibrium constants of these equations are not always known with great certainty. The second acid dissociation constant K 2 for equation (2.3) has values reported as far apart as 10-13 and 10 -19 [38]; however, at a pH of around 9 there is no doubt that the predominant solution sulphide species are hydrosulphide ions.

The available thermodynamic information can be summarised in the form of an E^-pH diagram, which can be used to predict the most stable species at any given E^ and pH. The solution potential, E^, can be applied electronically at an electrode surface, or by adding a redox couple to the solution. In the latter case the potential at equilibrium will be give by the familiar Nemst equation, and can be measured with a suitable indicator electrode (e.g. a platinum wire): E = E° + RT ln (a 0 ) (2.4) zF ar E = reversible potential vs. SHE / V. E° = standard reduction potential vs. SHE / V. z = number of electrons transferred; F= Faraday's constant / C (mol electrons )-1 R = Gas Constant / J mol -1 K_1; T = Temp. / K. ar, a0 = activities of reduced and oxidised species respectively. Review of Sulphide Oxidation 27

The thermodynamics of the sulphur - water system were first summarised in the form of an E^-pH diagram by Valensi [39]. His diagram is shown in Fig. 2.2. Notice that there are only three stable oxidation states.

Fig. 2.2 Eh-pH diagram for the sulphur/water system at 298 K. Sulphur Species present at unit activity [39].

Oxidation of hydrosulphide at pH 9.0 from Eh = -0.4 V to +0.1 V would be predicted to form sulphate. This prediction that sulphate will be the main oxidation product is in clear disagreement with the findings of numerous workers who have studied the oxidation of sulphide solutions in these potential and pH regions.

The reason for this apparent contradiction is that E^-pH diagrams are based solely upon the assumption that equilibration between species is taking place under thermodynamic control, and no account is taken of the rate of the possible reactions. Reactions producing sulphate ions, for instance, are known to proceed at a very slow rate during the atmospheric oxidation of sulphide solutions. Some account of these kinetic factors can be made by excluding from the diagram species which are known to form very slowly. This can be considered equivalent to adding to their free energy of formations in order to compensate for their large activation energies. Peters [40] produced a diagram from which sulphate species are excluded, Fig. 2.3 (overleaf). Review of Sulphide Oxidation 28

Fig. 2.3. Eh-pH diagram for metastable sulphur system at 298 K. [40].

A potential change from = -0.4 V to -0.1 V, again at pH 9, would now predict that sulphide will be oxidised to thiosulphate, and that a further increase to + 0.1 V would yield sulphite: 2HS- + 8 OH- -+ S20 32- + 5H 20 + 8 e- (2 .5 ) S20 32- + 6 OH- -> 2S0 32- + 3 H20 + 4e- (2.6)

Hamilton and Woods [41,42] studied the electrochemical oxidation of sulphide solutions at a gold electrode, and found that under mildly oxidising potentials (+0.2 V vs. SHE at pH 9.2) the rate of production of all sulphoxy compounds was negligible. They produced an E^-pH diagram with all such species excluded, Fig. 2.4 (overleaf), and used it to explain their experimentally observed oxidation products: polysulphide ions and elemental sulphur. Review of Sulphide Oxidation 29

Fig. 2.4 Eh-pH diagram for the sulphur/water system at 298 K Oxy-sulphur anions not considered [42].

Thus, three Eh-pH diagrams can be drawn for the sulphur/water system: one which considers all thermodynamically stable sulphur species; one which considers metastable sulphur and sulphoxy species; and one which only considers metastable sulphur species. These diagrams would predict the sulphide oxidation product to be sulphate, thiosulphate and sulphite, or sulphur respectively.

2.1 The Oxidation states of Sulphur The redox chemistry of sulphur is a rich and complex area of study, and complete textbooks have been written on this subject alone [43]. This review will be confined to the aqueous sulphur redox chemistry, and will concentrate on those studies which have been carried out in alkaline solutions.

2.1.1 Sulphide (-II) Below pH 7, sulphide (-II) exists as the species H2S(aq), which is moderately soluble: H2S(g) H2S(aq) K = 0-101 (2.7) This means that under a partial pressure of hydrogen sulphide of one atmosphere, aqueous concentrations of up to 100 mol n r 3 can be achieved. Above pH 7 the dominant species are the HS' ions: ^ 2^(aq) ^ HS“ + H+ pK ai = 6.99 (2.8) Review of Sulphide Oxidation 30

Using free energy data from Zhdanov [31], the [HS-] in solution can be calculated according to equation (2.9): H2S(g) + OH- <-> HS- + H20 log[HS-] = pH - 7.995 + logP^s (2.9) For instance at pH 8.5 a partial pressure of hydrogen sulphide of only 0.01 atm. would be in equilibrium with a solution containing 30 mol HS- m"3.

The HS- ion can deprotonate further to form the sulphide ion: HS- <-> S2- + H+ p K ^ = 13-19 (2.10) Until recently the value for pK a2 was accepted at around 13, implying that strongly alkaline sulphide solutions contained the S2_ species. However, evidence has now accumulated to suggest a value for pK a2 of 19 ± 2 [38, 44-46]. An early calculation of pK a2 [47] relied on UV-visible spectrophotometric measurements; assigning an absorbance band at 230 nm to the HS" ion and a band at 360 nm to the S2_ ion. Giggenbach [45] pointed out that the absorbance band at 230 nm due to HS" was subject to a blue shift on increasing the hydroxide ion concentration, which would produce an apparently falling absorbance value if measurements were made at a single wavelength. He suggested that the weak absorbance at 360 nm was due to the presence of polysulphide ions, which had been formed through the partial oxidation of the sulphide solution. He failed to observe this band, previously assigned to the S2_ ion, when oxygen was completely excluded, and from his own measurements suggested a value of pK ^ of around 17.1. This result was largely ignored for the next decade, but Meyer [38] later confirmed the higher value using IR Raman spectroscopy. He identified the presence of the HS" ion by its peak at 2570 cm"1, and found that it was present even in 16.9 kmol m -3 NaOH.

It now seems that a high value for p K ^ will have to be accepted, which has many implications [44]. The S2_ ion will not be a predominant species in aqueous solutions, even in strongly alkaline media. The value for AGf° for S2_ must be revised, together with any associated thermodynamic values calculated from this, which means that the solubility products of many metal sulphides will be even lower than previously thought. The formation of mercury sulphide has limited the value of polarographic investigations into the oxidation of sulphide solutions; HgS is formed at lower potentials than those at which the sulphide is oxidised [48].

2.1.2 Polysulphides (-1 to 0) Polysulphides are the low chain length sulphur di-anions; S22", S 32-, S42-, S52-. They have formal oxidation states intermediate between -I and 0. Alkali metal polysulphides (e.g. Na 2S4) can be prepared [49,50] and they will dissolve readily in water to form bright yellow solutions. Review of Sulphide Oxidation 31

Polysulphides can also be formed by the dissolution of elemental sulphur in alkaline sulphide solution: (n-l)S + HS- + OH- -> Sn2' + H20 (2.11) Polysulphides are more easily oxidised by atmospheric oxygen than sulphide species [51], forming S, S 2C>32-, SO32- and SC> 42_. Polysulphides are also formed as intermediates during the oxidation of sulphide solutions, especially around pH 7. Partially oxidised solutions develop a yellow-green colouration which is due to polysulphides.

It has been known for a long time that polysulphides in solution are always present in equilibrium with each other, even when solutions are prepared from solids with a fixed stoichiometry [50,52-53]. Giggenbach [53], working at an ionic strength of 2, derived a set of equilibrium constants for the reactions: 3 S 52- + HS- + OH- —^ 4 s42' + h 2o Kcq == 2 x 10-4 (2.12) 2 S42' + HS- + OH- —^ 3S 32- + h 20 Keq == 1.8 x 10-2 (2.13) S32’ + HS- + OH- —) 2 S22- + h 20 Keq == 4 x 105 (2.14) These equilibrium constants can be used to determine the concentration of an individual polysulphide ion in a solution of known sulphur to sulphide ratio and pH (see section 3.5.1). Nevertheless Power and Richie [49] ignored the above equilibria and simply assumed that a solution made up from Na 2S4 contained only the anion S 42-.

Using these equilibrium constants enables the concentrations of individual polysulphide ions to be calculated; the results show that tetrasulphide is usually the major species in aqueous solution and that HS" ions are present at a comparable concentration. Schwarzenbach and Fischer's earlier study is in agreement with these findings [50]. The average chain length in poly sulphide solutions that are saturated with sulphur is never greater than five.

Since polysulphides are thermodynamically unstable in alkaline solution, there exists the possibility of spontaneous disproportionation to produce thiosulphate: 4 S42' + 8 OH" + H20 <-» 3 S20 32- + 10 HS- (2.15) The equilibrium constant and the rate of this reaction have been measured [54]; the forward rate is slow, but increases significantly with temperature. However, polysulphide solutions at room temperature show no noticeable change in their UV- visible spectra even after months of storage.

Polysulphides protonate to form the polysulphanes, which are reported to be yellow solids. There is an experimental difficulty in determining the first and second dissociation constants of the polysulphanes, since in the course of an acid titration polysulphide species can disproportionate to form HS" and elemental sulphur: Sn2- + H+ -> (n-l)S + HS- (2.16) Review of Sulphide Oxidation 32

Schwarzenbach and Fischer [50] used a continuous flow technique to achieve the mixing of acid and polysulphide solutions, and determined the pH downstream from the mixing vessel, after the poly sulphide solution had only been acidified for 10 ms. They claimed that this time was too short for the polysulphide to disproportionate, and from the resulting titration curves determined the following pKa values:

Species P K a l P K a2 H2S2 5 .0 9 .7 h 2s 3 4.2 7 .5 h 2s4 3.8 6.3 h 2s 5 3 .5 5 .7

Table 2.1 The pKa values of Polysulphides [50].

The above values imply that at pH 8.5 all the polysulphides would be present in aqueous solution as their dianions except the disulphide, which would be present as HS2~ Disulphides normally constitute only a minor component in a polysulphide mixture.

Lessner et al [55] studied the electrochemical redox behaviour of polysulphide solutions at pH 12. They found that using slow sweep voltammetry at platinum and cobalt electrodes, sulphur was deposited during the positive going scans. This current peak and the associated electrode passivation masked the oxidation reactions that produce higher polysulphides, eg: 5S 42- -» 4 S 52- + 2e- (2.17) Upon negative going potential scans they found that hydrogen was evolved before a well defined diffusion limited current peak due to polysulphide reduction was observed.

Using voltage pulse methods, the same authors found that the resulting currents were much smaller than those expected from the calculated concentrations of tetrasulphide. They concluded that the electroactive species were a minor component in the equilibrium mixture, and suggested that they were supersulphide ions, S 2", which are known to be produced from tetrasulphide ions at elevated temperatures [54]: S42- 2 S2‘ (2.18) Using the equilibrium constant of this reaction Lessner et al [55] predicted supersulphide concentrations that were consistent with their observed currents over the temperature range 25-80 °C.

2.1.3 Elemental Sulphur (0) Pure sulphur exists at room temperature in the crystalline orthorhombic form, consisting of stacked layers of puckered Sg rings. Above 368 K it transforms into the monoclinic phase, which is stable up to the melting point of 392 K. Review of Sulphide Oxidation 33

When molten sulphur is heated to 457 K, the Sg rings break and chain molecules result; if this liquid phase is then quenched rapidly amorphous sulphur is formed. This form of sulphur deforms plastically and can be stretched to several times its original length. It has also been suggested that amorphous sulphur can also result from the electrochemical oxidation of metal sulphides [56]. Sulphur is insoluble in water and has a high electrical resistance (resistivity 1.9 x 1015 Q m); sulphur coatings therefore passivate electrodes. Colloidal sulphur is a poorly defined material [43]; it can contain polythionates of the type SO3" -Sn-S03~, where n has a value of 10 to 20.

It is clear that all reactions of Sg must first require ring scission, which demands a considerable activation energy; the S-S single bond strength is 226 kJ mol -1 [57]. Therefore, elemental sulphur is resistant to further oxidation, and in many systems the formation of sulphur is irreversible. Habashi and Bauer [58] found that elevated temperatures and a high pressure of pure oxygen were required to effect the complete oxidation to sulphate.

2.1.4 Polythionates (0 to IV) Polythionates have the general formula (OgS-Sn-SOg)2-, the best characterised ions are those having n = 1-4. They can be prepared by reducing sulphurous acid with hydrogen sulphide, a process which produces a complex mixture: Wackenroder’s solution. Tetrathionate is produced quantitatively by the oxidation of thiosulphate with iodine: 2 S2032" + I2 —> 2 T + S40 62- (2.19) In acid solution the polythionates disproportionate to give S, SO 2 and SO 42'.

2.1.5 Thiosulphate (II) The thiosulphate ion has the structure S-SOg2- [59], so the two sulphur atoms have differing chemical environments. Thiosulphate solutions disproportionate in acid solution to give elemental sulphur and sulphur dioxide: S2032- + 2 H+ -> H20 + S + S 02 (2.20) In alkaline solution the reverse reaction can occur, and thiosulphate can be prepared by heating sulphur with sulphite solution. Many metals form soluble complexes with thiosulphate, particularly silver and mercury. Thiosulphate solutions are used to dissolve the light sensitive silver bromide in photographic emulsions to 'fix' the image.

The fact that mercury forms a complex ion means that the oxidation of thiosulphate cannot be studied using polarography, as the the mercury surface is oxidised preferentially [48]: Hg + 2 S2032- —> Hg(S 203)22“ + 2 e" (2.21) Thiosulphate is also difficult to reduce, and no reduction waves are observed at a dropping mercury electrode. At a platinum electrode, potentials o f-1.75 V vs SHE are Review of Sulphide Oxidation 34

required before HS" is produced. It is this kinetic inertness that has led to thiosulphate being named a metastable sulphide oxidation product. Though not thermodynamically stable, thiosulphate solutions can nevertheless be kept for weeks without appreciable disproportionation or oxidation.

2.1.6 Sulphite (IV) Although aqueous solutions of sulphur dioxide have been termed sulphurous acid, it has now been established that the free acid does not exist, and aqueous solutions contain SO 2 (aq). Aqueous sulphur dioxide solutions can give rise to two series of salts, the sulphites containing S 032‘, and the bisulphites containing HSO 3". The bisulphite ion can react with itself to form the metabisulphite ion S 2O52": 2H S 03- <-» S2052- + H20 (2.22) S2O52" ions exist in dehydrated solid salts and in concentrated aqueous solutions.

Sulphite ions can be oxidised to sulphate, and in alkaline solution they will act as reducing agents, for instance slowly removing dissolved oxygen. Samec and Weber [60] studied the electrochemical oxidation at a gold electrode. They found that the oxidation rate was much slower than that expected for a diffusion controlled process, and was characteristic of an adsorption process followed by an irreversible two electron transfer to yield sulphate.

2.1.7 Sulphate (VI) Sulphuric acid, and its two series of salts, the sulphates and bisulphates, represent sulphur in its highest normal oxidation state. They are thermodynamically stable in aerated aqueous solutions at all pH s, and are the ultimate oxidation product of all other sulphur salts. Theoretically, the sulphates may be reduced to sulphur (0) or sulphide (—II), (see Fig. 2.2). In practice these reactions are highly irreversible; sulphates are not normally reduced in aqueous media even in the presence of powerful reducing agents. In fact, sulphates are so resistant to reduction that they are commonly used as background electrolytes in electrochemical studies. However, sulphate can be reduced to HS" by the bacteria vibrio desulphuricans which can achieve this in cold aqueous solution [61].

2.2. Electrochemical Investigations into Sulphide Oxidation In an investigation of the oxidation and reduction of hydrosulphide ions on a rotating gold electrode at pH 6.8 and 9.2, Hamilton and Woods [41] concluded that at low potentials, first sub-monolayers, then multilayers of sulphur were produced. They found the ratio of anodic charge to cathodic charge was greater than 1, and furthermore that this charge imbalance increased with increasing rotation rate. They concluded that this must be due to a soluble intermediate which was dispersed at high rotation rates. Polysulphides are known to be soluble, giving yellow-green solutions, and so they postulated that the reaction proceeded through a poly sulphide intermediate. Sulphur (0) is known to exist in polymeric form; rhombic sulphur consists of stacked Sg rings. Review of Sulphide Oxidation 35

Therefore, polysulphide ions are reasonable intermediates to propose for the oxidation of sulphide to sulphur. Indeed, Allen and Hickling suggested a similar mechanism in 1957 [62].

In a later paper, Buckley, Hamilton and Woods [42] showed that the initial sub­ monolayer coverage, which formed in the underpotential region (~ -0.2 V vs. SHE at pH 9.2), showed an X-ray photoelectron spectrum that was consistent with a gold sulphide type structure. If this potential was maintained for extended periods (~ 10 mins.), then multilayers of sulphur were formed which passivated the electrode surface. They confirmed that the oxidation and reduction of the sulphur proceeded via soluble polysulphide species by detecting them using a rotating ring-disc electrode (RRDE) [63].

The above authors [42] first plated the disc with sulphur in a positive going potential scan, then reduced the adsorbed sulphur in a negative going scan; they found that polysulphide species were produced: nS + 2 e- -> Sn2- (2.23) By holding the ring at a highly negative potential (-0.92 V vs. SHE) the polysulphide ions were further reduced to HS~ according to equation (2.24): Sn2- + 2(n-l)e- + nH+ -> n HS‘ (2.24) where n = 2,3,4 or 5. Polysulphides can also be produced by the chemical dissolution of sulphur in hydrosulphide solutions: (n-l)S + HS- -> Sn2- + H+ (2.25) By comparing the charges passed due to equations (2.23) and (2.24), and allowing for the chemical production of polysulphides via chemical dissolution (2.25) they calculated the mean chain length in the polysulphide intermediates, according to equation (2.26): n = 1 + (Qr -qr)/N Q d (2.26) where Qr = charge due to polysulphide reduction at the ring / C. Qd = charge due to sulphur reduction at the disc / C. N = the RRDE collection efficiency, a constant for a given geometry. qr = charge due to the reduction of chemically produced polysulphide / C. They found that at pH 9.2, n = 3.3, indicating that a mixture of different polysulphides was produced.

At potentials above +0.25 V, the adsorbed sulphur layer can be oxidised to form sulphate (2.27). Sulphate can also be formed directly through the oxidation of hydrosulphide to sulphate, equation (2.28); this reaction can occur in parallel with sulphur formation. S + 8 OH- -» S042- + 4 H20 + 6 e- (2.27) HS- + 9 OH- -> S 042- + 5 H20 + 8 e- (2.28) Review of Sulphide Oxidation 36

In a comprehensive review series, Zhdanov [48] reported that in alkaline solution under mildly oxidising potentials, sulphide was oxidised to yield polysulphide ions, which he ascribed to the dissolution of an initial deposit of sulphur in the sulphide solution, equation (2.25). At much higher potentials, 1.0-1.7 V vs. SHE, thiosulphate, and some sulphate, were formed in addition, and it was not until a potential of above 1.7 V was applied that the oxidation product was predominantly sulphate.

Moscardo-Leveist and Plichon [64] studied the electrochemical oxidation of sodium sulphide in an equimolar sodium hydroxide/water melt at 100 °C and again found that two oxidation steps were involved; the first step yielded elemental sulphur and di- and tri-sulphides and the second, at higher anodic potentials, produced sulphite ions.

Remick and Camara [65] studied the electrochemistry of the sulphide/polysulphide couple. They prepared their polysulphide solutions by dissolving elemental sulphur in alkaline sulphide solutions according to equation (2.25). Polysulphide solutions prepared in this way contain a number of polysulphide species; the average length of the polysulphides being determined by the ratio of sulphur(O) to hydrosulphide(-II) used. Giggenbach [53] studied such solutions by UY-visible spectroscopy and determined the concentration of each polysulphide species present from their absorbances. The analysis is complicated by the fact that the separate polysulphide ions show absorbance maxima at very similar wavelengths. Nevertheless, he calculated the absorbance maxima and the extinction coefficients for a range of polysulphides.

Remick and Camera [65] confirmed Allen and Hicklin’s earlier mechanism [62] concerning the oxidation of polysulphide solutions. Their results were consistent with the following scheme: 1. Adsorption of polysulphide ion onto the metal (M) surface: Sn2- + M —> M~Sn2- (2.29) 2. Oxidation of the adsorbed poly sulphide by solution polysulphide: M—Sn2- + Sn2- —> M—Sn.j + Sn.j2- + 2 e- (2.30) 3. The adsorbed layer of polysulphide can then be regenerated by reaction with solution hydrosulphide ions: M—Sn_! + HS- + OH- -> M~Sn2- + H20 (2.31) In this way, a surface layer of polysulphide acts as an electrocatalyst for the oxidation of polysulphide solutions. This mechanism explains the intermediate formation of poly sulphides (e.g. S 42-) observed by Hamilton and Woods [42]. The first step may be oxidation of hydrosulphide to form tetrasulphide ions: 4HS- + 4 OH- -> S42- + 4H 20 + 6 e- (2.32) These can then adsorb onto the electrode surface: M + S42- —> M—S42- (2.33) and there undergo oxidation to produce adsorbed sulphur: M—S42- + HS- + OH- —> M—S + S42- + H20 + 2 e- (2.34) Review of Sulphide Oxidation 37

The polysulphide produced in equation (2.34) could re-adsorb and repeat the reaction scheme. In this way a sulphur monolayer could be built up, with polysulphide ions produced as intermediates close to the electrode surface. These could either adsorb onto the electrode or, if the electrode is rotated, be dispersed into solution.

Remick and Camera [65] studied the electrocatalytic activity of several electrode materials. They found that platinum was a poor electrocatalyst, especially for the cathodic reduction of polysulphide to sulphide. They attributed this to the removal of the adsorbed polysulphide layer on cathodic sweeps. As expected, conducting metal sulphides were found to be better electrocatalysts for both the oxidation and reduction reactions. Their work, and the study by Hodes and Joost [ 66], concluded that CoS, NiS, and M 0S2 were the most effective electrocatalysts. Platinum and carbon were less effective.

2.3 Chemical Oxidation of Sulphide Solutions using Oxygen Many workers have studied the air oxidation of sulphide solutions, and recently a review of the topic was published by Kuhn, Kelsall and Chana [34]. When the oxidation is achieved electrochemically, the applied potential governs the extent of oxidation. In the same way, with sulphide solutions which are chemically oxidised, the extent of oxidation is largely determined by the molar ratio of dissolved O 2 to HS'. Studzinska [67] reviewed the reaction and reported that a low ratio of O 2 to HS- favoured sulphur production, whereas a high ratio resulted in S 2O32-, SO32-, and SO42- formation.

The solution pH also has an important role to play; sulphur is thermodynamically stable only in acidic or neutral solutions. Therefore, it would be expected that under mild oxidation in this pH region, sulphur would be the predominant oxidation product, and sulphoxy species would predominate at a higher pH. A substantial pH change can also occur on oxidation, and it must be ensured that the solution is adequately buffered to prevent this altering the reaction course. Alferova and Titova [ 68] carried out the oxidation of sulphide solutions at various pH values by aerating them for 24 hours. At pH 7, they found that most of the starting sulphide was converted into elemental sulphur. As the pH was increased, the proportion of sulphide that was converted into thiosulphate and sulphite rose, until at pH 15 conversion to thiosulphate was almost complete. O'Brien and Birkner [22] listed the reaction products from the results of a number of studies, including their own. In alkaline solutions, thiosulphate and sulphite were the main products, although polysulphides, sulphur, and sulphate were also reported.

It is well known that, in elemental sulphur, Sg rings can be broken by ultra-violet light. Since polysulphide ions absorb in the near ultra-violet wavelength regions, they too are likely to be decomposed by strong light sources, forming reactive free radicals. Therefore, it is possible that ambient light conditions can affect the reaction rate of Review of Sulphide Oxidation 38 aqueous sulphide oxidation and alter the reaction products. Cox and Sandalls [69] showed that light of wavelength 300 to 400 nm caused photo-oxidation of gaseous mixtures containing oxygen and traces of hydrogen sulphide. Sulphur dioxide was the major product, and this was thought to be produced from the reaction between HfjS and O* or OH* free radicals. Pelizetti [70] showed that hydrogen sulphide can be cleaved in aqueous solution by visible light, producing hydrogen and elemental sulphur. He added colloidal cadmium sulphide particles, which acted as photocatalysts: They absorbed photons to generate electron-hole pairs, and the HS“ ions then acted as hole scavengers, so becoming oxidised to form sulphur (or polysulphides).

Many workers studying the oxidation of aqueous sulphide solutions did not monitor the product distribution, following the reaction instead by the decrease in the sulphide concentration. In many industrial applications, e.g. waste water treatment, it is relatively unimportant to determine this distribution (providing the products are not toxic). The possibility of using as a measure of the extent of oxidation was ignored, and no attempts were even made to measure it with a suitable indicator electrode.

2.3.1 Rate of Reaction of Sulphide Solutions with Oxygen The reaction rate is usually defined as the rate of loss of sulphide ions, taking no account of the products. Thus plots of 'rate' against pH can be misleading, as various reactions are known to be predominate in the different pH regions.

It is clear that the uncatalysed oxidation of sulphide solutions using oxygen alone proceeds slowly. At 25 °C, ti/ 2, the time taken for the sulphide concentration to reach half its initial value, is several hours. Many workers noted that the reaction was preceded by an induction period varying from 15 minutes to two hours [71-74]. Such an induction period is characteristic of an autocatalytic reaction. Bowers [72] suggested that the catalytic products were polysulphide ions, whilst Cline and Richards [75] proposed that the catalysts were free radicals. They found that a high surface area to volume ratio in their glass reaction vessels reduced the induction time.

Bhaskarwar and Kumar [76] studied the oxidation of hydrosulphide solutions using a foam bed contactor operating at 75 °C. They suggested that the reaction proceeded through a Sg2- intermediate which could either undergo mild oxidation and ring closure to form elemental Sg, or further oxidation to form S 2O32-. The foam was stabilized using dodecyl sulphate or octyl phenoxy polyethoxyethanol surfactants.

Several workers [22,73,77] found an approximate first order dependence of the reaction rate on sulphide concentration. O’Brien and Birkner [22] determined that the reaction was first order with respect to the oxygen partial pressure, although Chen and Morris [77] quote this order to be 0.56. Review of Sulphide Oxidation 39

2.3.2 Effect of Temperature and pH on Reaction Rate Selmeczi [78] reported that the reaction rate increased substantially with temperature, and Bowers [72] showed an Arrhenius plot of reaction rate vs. temperature in the range 20 °C to 50 °C. The change in rate over this range was approximately 20 fold.

The pH was reported by all authors to have a distinct effect, which is not surprising in view of the varying reactions occurring in differing pH regions. Chen and Morris reported two rate maxima at pH values of 8.5 and 11.5 [77]. Snavely and Blount [74] found a substantial rate increase at pH 11.5, and Alferova and Titova [ 68] reported maximum rates at extremes of pH.

2.3.3 Catalysis of Sulphide Oxidation Snavely and Blount found that just 5 mg Co2+ dm-3 effected complete oxidation of aerated 6 mol m-3 sulphide solution in only 60 seconds [74]. Interestingly, they also noted that their own early results on uncatalysed systems were in error because a chromium plated oxygen probe had catalysed the reaction. It may be that other studies purporting to be on the uncatalysed system have also been affected by trace contaminants (e.g.,in the chemicals used). This may explain the varying rates and reaction products observed by different workers under apparently similar experimental conditions.

Transition metals, noble metals, activated carbon, and organic compounds are all known to catalyse the oxidation. Studzinska claimed that the most effective catalysts were the transition metals [67] and she ranked their effectiveness in the order: Ni2+ > Co2+ > Mn2+ > Cu2+ > Fe2+ Organic catalyts, such as phenols and hyroquinones, were ranked less effective still, but still increased the rate of oxygen uptake 10 to 20 fold. Activated carbon is also known to catalyse hydrogen sulphide oxidation both in the gas phase [79], and in aqueous solution [73]. Oeste [80] proposed that the hydrogen sulphide oxidation over activated carbon took place via an electrochemical mechanism; Kuhn and Kelsall made a similar suggestion about the catalytic effect of the transition metals [34].

Metal sulphides have extremely low solubility products. For example, for copper(II) sulphide: CuS + H20 = Cu2+ + OH' + HS- (2.35) K sp = [Cu2+] [OH-] [HS-] = 6 x 10-37 This means that heterogeneous metal sulphide particles are usually produced when solutions containing transition metal catalysts are added to aqueous sulphide solutions. Surprisingly, workers have consistently ignored the presence of these colloidal particles, which is unusual in view of their possible role as redox catalysts. Review of Sulphide Oxidation 40

Since many of these metal sulphides are known to be electrically conducting and to catalyse oxygen reduction [81], Kuhn and Kelsall proposed the following mechanism for transition metal catalysis of sulphide oxidation [34]. They suggested that the overall equation (2.38) was split into two half reactions, equations (2.36) and (2.37), each of which could occur on the metal sulphide surface: HS- + 9 OH- -> S 042- + 5H 20 + 8 e- (2.36) 4 H20 + 2 0 2 + 8e- -» 8 OH' (2.37) HS- + 2 0 2 + OH" —» S 042- + H20 (2.38) A similar mechanism has been proposed to explain the catalytic activity of colloidal gold particles [82]. Since CoS, NiS, and M 0S2 are now known to be electro-catalytically active for sulphide oxidation [65], this mechanism explains their effectiveness as catalyts. However, an eight electron transfer is unlikely to occur in one step, and the reaction is likely to proceed via a series of intermediates, such as soluble polysulphides. Electrochemical hydrosulphide oxidation at the mineral sulphide surfaces, galena (PbS) and pyrite (FeS 2), has shown that polysulphide ions can be formed [56]. These metal sulphides were found to be more effective electrocatalysts than the noble metals, platinum and gold.

Vanadium does not precipitate a solid sulphide phase when vanadium (V) salts are added to hydrosulphide solutions. Vanadium (V) is a moderately powerful oxidising agent and is capable of directly oxidising the hydrosulphide to produce sulphur [83]. This can produce solid vanadium oxide phases, which are noted for their catalytic abilities during gas phase oxidations at high temperatures; for instance in the well known contact process which oxidises SO 2 to SO 3 [84]. Recently 51V NMR studies [37] have shown that vanadium (V) can also form oxy-sulphur complexes in solution, which may assist the catalytic action of vanadium(V).

2.3.4 Bacterial Action in Sulphide Oxidation Experience in working gas desulphurisation plants shows that bacterial oxidation can change the product distribution of the sulphide oxidation [29]. However, a culture time of several days is required before the bacterial population can significantly change the oxidation pathway. Therefore, it is unlikely that experiments on the 'chemical' oxidation of sulphide solutions have been affected by bacterial action, despite the fact that no special precautions appear to have been taken to exclude bacteria.

2.4 The Production of Elemental Sulphur It is clear from the above discussion that elemental sulphur is not a thermodynamically stable oxidation product, except under conditions of mild oxidation in acidic solutions. Although sulphate (oxidation state +VI) is the preferred product, it is observed only after prolonged oxidation, and more commonly metastable sulphur species are formed; these include sulphur (0), thiosulphate (II), and sulphite (IV). Concentrations of intermediates such as polysulphides and polythionates can also accumulate. Many workers have observed that the available reaction pathways are followed in parallel, Review of Sulphide Oxidation 41 resulting in the simultaneous production of thiosulphate and sulphur from hydrosulphide solutions for instance.

Sulphur is the desired product from many industrial oxidation processes because it is non-toxic, easily handled and a saleable by-product. Complete conversion to sulphur can be anticipated only under oxidation conditions where it is thermodynamically stable; i.e.,low solution potentials of around 0.0 V vs SHE and a pH of around 5. These conditions can be provided if the oxidation is carried out electrochemically, and industrial processes based on this principle have been proposed [15,16].

Operating at an acidic pH retards the absorption of hydrogen sulphide into the aqueous phase. Industrial processes operating at this pH must compensate by ensuring the rapid oxidation of the H 2S once it is in solution. Although there is an industrial process which uses an acidic solution containing an Fe (III) complex [21], most processes have utilised alkaline working solutions. Vanadium (V), iron (III), and arsenic (V) have all been used industrially as the oxidising agents [29,85,10]. These processes all produce a range of higher oxidation state sulphur products.

Maximum production of elemental sulphur can be achieved by removing the sulphur from the reaction system as soon as possible. In the Stretford Process the aeration which occurs in the oxidiser also serves to remove the sulphur by froth flotation [35]. It has also been suggested that de-oxygenating the absorbing solution prior to it contacting the gas stream containing the hydrogen sulphide can decrease the rate of production of thiosulphate [ 86]. Sulphide Electrochemistry 42

3. Sulphide Electrochemistry Electrochemical studies of hydrosulphide ions in aqueous solution are made difficult because most metals form their sulphides when oxidising potentials are applied. Of the noble metals, gold is known to form a sulphide coating less readily than platinum [87], and so this metal was chosen for the working electrode material. Gold can dissolve in sulphide solutions to form the gold (I) complex; AuS". Garrels and Christ [ 88] produced an E^-pH diagram for the Au/S/Cl system (Fig. 3.1), which shows this complex to be thermodynamically stable in alkaline solution:

Fig. 3.1 Eh-pH Diagram for the Au/Cl/S System. Dissolved concentrations / kmol n r3: Au as marked; Cl 1; S 0.1.

However, in practice, corrosion of gold in sulphide solutions is not a serious problem; in the present study it was found that several hours continual potential cycling from -0.9 V to +0.3 V vs. SHE. (at 20 mV s-1), in a solution containing 1 kmol Na 2S m"3 at pH 14, was required to strip a gold layer only 1 Jim thick.

Recent evidence from Buckley, Hamilton and Woods suggested that gold electrodes that were immersed in sulphide solutions (at pH 9.2) and held at potentials higher than -0.5 V vs. SHE became coated with a gold sulphide phase [42]. However, such a phase is likely to be semiconducting and has been shown not to passivate the electrode towards further sulphur deposition. Sulphide Electrochemistry 43

3.1 Thermodynamic Calculations Fig. 3.2 shows an Eh-pH diagram for the sulphur/water system at 298 K. It was created using the program PPE produced by Angus [89,90] running on the Apple He microcomputer. The thermodynamic data, in the form of the free energies of formation of the species considered, were taken from a review by Zhdanov [31]. Notice that there are only three stable oxidation states. Oxidation of hydrosulphide at pH 9.0 from Eft = -0.4 V to Ejj = +0.1 V would predict that sulphate would be formed as the predominant product.

Fig. 3.2 Eh-pH diagram for the sulphur/water system at 298 K. [S species] = 10 mol n r 3.

Zhdanov [31] quotes a value of 86.31 kJ mol -1 for the AGf° of S2-, which implies that the value for the pK ^ of H2S is 13. If Zhdanov had accepted the new, higher value for the pKa2 of around 19 (see section 2.1.1) then the value for AGf° (H 2S) would be 119 kJ mol -1 (assuming that AGf° (HS-) = 12.05 kJ mol-1). However, utilising this higher value does not change the above diagram significantly, merely eliminating the area of predominance for S2' ions.

A metastable E^-pH diagram can be produced by eliminating all sulphur (VI) compounds from the calculations (see section 2). Such a diagram is shown in Fig. 3.3. and this indicates that the oxidation of a sulphide solution at pH 9.3 can proceed to form thiosulphate (at E^ > -0.3 V vs. SHE), sulphite (E^ > -0.175 V) or dithionate (E^ > +0.12 V). It is also worth noting that, although the polysulphide species do not appear on the diagram, the area of predominance of the ion S 52" is masked only by that of elemental sulphur. Elemental sulphur can only form when suitable nuclei are available, and often a high degree of supersaturation is required Sulphide Electrochemistry 44 before such nuclei are formed. Thus, in mildly alkaline solution, yellow coloured polysulphide solutions can result from the atmospheric oxidation of hydrosulphide ions, despite the fact that they are not thermodynamically stable.

Fig. 3.3 Metastable Eh-pH diagram for the Sulphur-water system 298 K.S(VI) species excluded. [S species] = 10 mol m“3.

3.2 Experimental Cyclic voltammetry and potential pulse studies were conducted at room temperature (~ 20 °C) using hydrosulphide solutions at pH 9.3 with a rotating ring- disc electrode (RRDE). A schematic diagram of this electrode is shown in Fig. 3.4. crztr^

Fig. 3.4 A Rotating Ring Disc Electrode Sulphide Electrochemistry 45

The ring and disc potentials can be controlled independently using a bipotentiostat, and the potential of the ring can be adjusted, such that any metastable oxidation products passing to the ring become reduced. As the electrode is rotated, these soluble species produced at the disc are spun out to the ring, and the reduction current at the ring can be used to obtain a measure of their concentration in solution. The collection efficiency, N, is defined as that proportion of soluble product, produced at the disc, which is transported to the ring. The collection efficiency is dependent on the electrode geometry and can be predicted theoretically [63], or experimentally checked by monitoring the transport of a material which is known to be oxidised and reduced reversibly (such as Fe(CN)64").

3.2.1 Solution Preparation A carbonate buffer solution of pH 9.3, containing 0.059 kmol Na 2C03 m"3, 0.223 kmol NaHCC >3 m"3 and 0.10 kmol Na 2S04 m-3, was prepared by dissolving the appropriate mass of analytical grade materials (BDH) in triply distilled water. Similarly a borate buffer, having a pH of 9.2, was made up containing 12.5 mol Na2B4O7.10H2O m-3, 0.9 mol NaOH m -3 and 0.1 kmol Na 2S04 m-3. A stock solution containing 0.1 kmol HS" m "3 was prepared by dissolving an accurately weighed amount (about 12 g) of transparent, dried crystals of Analar sodium sulphide (BDH) in 500 cm3 the appropriate deoxygenated buffer solution.

The molarity of this stock solution was checked by conducting an iodate titration: exactly 1 cm3 of the stock solution was taken and mixed with a 15 cm 3 aliquot of potassium iodate solution (0.025 kmol m“3). The mixture was made highly alkaline by adding 10 cm 3 of sodium hydroxide (10 kmol m“3) and boiled for 10 minutes. Under these conditions the sulphide was oxidised into sulphate: 4I03- + 3HS- + 3 0H- 41- + 3 S042- + 3 H20 (3.1) The unused iodate was then back titrated. Excess potassium iodide solution (5 cm 3 of 5 % KI by mass) was added to the cooled solution which had been made acidic by adding 20 cm 3 of H 2SO4 (4 kmol m"3). This converted the unused iodate to iodine: I03- + 51- + 6 H+ -» 3 I2 + 3H20 (3.2) The liberated iodine was then titrated with thiosulphate (0.1 kmol m-3): 6 S20 32" + 3 I2 -> 6 S4062- + 61- (3.3) As the end point neared, the solution became a pale yellow colour and the solution was diluted to 150 cm 3 with distilled water. Several drops of sodium starch glycollate (BDH) were added and the end point was detected when the characteristic deep blue colour of the starch-iodine complex was discharged. The concentration of the original hydrosulphide solution was calculated from equation ( 3.4): C = 7.5 x 105 (3.75 x 10"4 - 16.666 Vt) (3.4) where C = Concentration of stock hydrosulphide solution / mol m “3 Vt = Volume of thiosulphate solution taken / m 3 Sulphide Electrochemistry 46

Freshly opened sodium sulphide was found to contain about 32 % Na 2S, which corresponds to the formula Na 2S.(H20)9 23.

Stock sulphide solutions could be kept for several weeks without degradation in a septum-stoppered bottle with a nitrogen atmosphere over the liquid. Measured volumes were withdrawn using a glass syringe and needle, and injected into a larger volume of the nitrogenated buffer solution, to make solutions containing 10 mol HS" m '3. Electrochemical studies were carried out in a three compartment glass cell of conventional design, (see Chapter 7, Fig. 7.1).

Sodium tetrasulphide (Na 2S4) was prepared according to the method given by Schwarzenbach and Fischer [50]. Under an inert atmosphere 12.588 g of sodium was dissolved in 400 cm3 Qf absolute ethanol, forming sodium ethoxide: Na + C2H5OH C2H5ONa + 1/2 H2 t (3.5) Dry hydrogen sulphide was then bubbled through the solution until it became saturated, whereupon sodium hydrosulphide was formed: C2H5ONa + H2S -> NaHS + C2H5OH (3.6) Excess hydrogen sulphide was absorbed in Dressel bottles containing solutions of NaOH (10 kmol m'3) and CUSO 4 (1 kmol m-3), to prevent its escape into the atmosphere.

To this, 26.386 g of elemental sulphur was added; this dissolved with the evolution of hydrogen sulphide, producing a deep red solution of sodium tetrasulphide: 2 NaHS + 3S -> Na2S4 + H2S T (3.7) To ensure that reaction (3.7) proceeded to completion the solution was refluxed under an inert atmosphere for one hour. It was then cooled to below 40 °C and the solvent evaporated under vacuum so that the solution was reduced to 1/10 of its original volume. A yellow crystalline product was obtained, which was filtered under vacuum in an inert atmosphere and dried for one week over P 2O5. A yield of 33.64 g was obtained (71 %).

Polysulphide solutions were prepared by either dissolving the appropriate mass of Na2S4 in an oxygen-free buffer solution, or by dissolving elemental sulphur in sulphide solution. A stock solution, of average polysulphurisation index of two, was prepared by adding elemental sulphur to a solution of Na 2S.9H20 in the molar ratio 1:1. The total sulphur concentration was 0.1 kmol n r 3 and the sulphur took several days to dissolve, forming a transparent, bright yellow solution. Such a solution will contain not only S 22“, but also HS', S 32", S42" and S 52".

3.2.2 Electrochemical Instrumentation Gold and platinum ring-disc electrodes of were mounted in a motor unit (Oxford Electrodes) which allowed the rotation speed to be continuously varied up to 50 Hz. The disc areas were 0.3848 cm2 and the electrode dimensions were rj = 0.35, Sulphide Electrochemistry 47

T2 = 0.375, 13 = 0.4 cm. This geometry provided a theoretical current collection efficiency of 0.17, which was verified experimentally using a solution containing 1 mol K 4Fe(CN)6 m-3. The hexacyano iron (II) was oxidised at the disc and the product reduced at the ring, the ratio of the two currents was 1: 0.17.

The potentials of the ring and the disc, relative to a saturated calomel reference electrode (EIL), were controlled independently using a bipotentiostat built at Imperial College, based on conventional operational amplifier design. The control potentials were provided by two Hi-Tek PPR1 waveform generators, and bright platinum counter electrodes were used. The ring and disc currents were passed through resistors and the resulting voltages applied to the inputs of separate J J PL4 chart recorders. The thiosulphate cyclic voltammetry was carried out using one channel of the bipotentiostat, and the polysulphide voltammetry was conducted using the Solartron 1286 Electrochemical Interface.

3.2.3 Electrode Pretreatment Initial experiments were carried out using platinum electrodes that had been coated with gold. This coating was achieved by polishing the platinum surface until a mirror finish was obtained, then electroplating the gold from aqueous solution under potentiostatic control, whilst the electrode was rotated at 40 Hz. The electroplating solution contained 1 mol AUCI 4'm -3 and 0.1 kmol HC1 m-3. During the electroplating, the potential was maintained at 0.287 V vs. SHE (0.045 V vs. SCE) which caused a current of approximately 1.3 mA to flow. After 1 minute the electrode was disconnected and the gold surface gently polished with a lint-free tissue to prevent the deposit from becoming dendritic. The electroplating was then continued until the desired thickness of gold had been achieved. Gold coatings prepared in this way were bright and adherent, and voltammograms recorded on pure gold and gold-plated platinum surfaces were identical.

Gold electrodes that have been exposed to the atmosphere adsorb oxygen on their surfaces, which gives rise to a reduction current on the first negative-going potential scan. Standing the electrode in nitrogenated buffer, or potential cycling in the same media desorbs this oxygen. During rotated-disc experiments, care was taken to maintain a nitrogen atmosphere above the solution surface; the disc rotation had the effect of aerating the solution which could cause large oxygen reduction currents to flow (280 |iA at -0.5 V vs. SHE).

After polishing with 0.3 qm alumina powder until a mirror finish was obtained, the electrode was introduced into the working solution. Cathodic polarization and potential cycling were investigated as possible methods of electrode activation. Holding the electrode at a highly negative potential (-1.7 V vs SHE) removed the adsorbed oxygen, but subsequent voltammograms recorded in sulphide solutions showed current densities lower than those which were obtained after the electrode had undergone Sulphide Electrochemistry 48 potential cycling, suggesting the presence of adsorbed sulphur. It was found that potential cycling at 10 V s -1 between -1.25 V and +1.75 V vs. SHE produced an active gold surface. If the anodic limit of the potential scans was reduced to -0.2 V vs. SHE, the electrode surface was not activated; adsorbed sulphur on the gold electrode is not oxidised to sulphate until a potential of 0.5 V vs. SHE is exceeded [41]. After potential cycling, the electrode was held at the cathodic limit, prior to commencement of a potential scan or pulse.

Current densities were calculated from the geometrical surface area unless otherwise stated. The real surface area of the gold disc electrode was determined according to the method of Dickertmann et al. [91], which relies on the integration of the charge passed when a layer of gold oxide is formed. The gold electrode was placed in a solution containing 1 kmol HCIO 4 m"3, and the potential scanned from 0.5 V to 1.7 V vs. SHE at 10 mV s"1; polycrystalline gold with a roughness factor of one forms a monolayer oxide coating with the passage of 0.40 mC cm-2. Using this method the roughness factor of the polished gold electrode was found to be 1.3.

3.2.4 Experimental: Ion chromatography The experimental apparatus was assembled as shown in Fig. 3.5 (overleaf). A glass reservoir held the eluent, 0.1 kmol Na 2C03 above m-3, which was constantly sparged with nitrogen. The eluent was pumped through a Kontron 414T pump, through a pressure damper and injection port and into a Dionex AG3 guard column. This guard column prevented strongly adsorbing ions from poisoning the main ion exchange column. The main column was made from the same material and achieved the separation of the sulphur anions. Two detectors were provided: a LKB 2238 Unicord SII fixed wavelength UV detector (at 254 nm) and a Dionex ECD electrochemical detector. The latter consisted of a silver working electrode (held at -0.1 V vs Ag/AgCl), a gold counter electrode and an Ag/AgCl reference electrode. Any sulphide or polysulphide passing to this detector produced an oxidation current and formed Ag 2S at the working electrode.

Complete exclusion of air from the working solutions was found to be essential to prevent oxidation of the polysulphide species. Aqueous polysulphide solutions were prepared by diluting stock solutions (see section 3.2.1) with deoxygenated purified water to form solutions in the concentration range 0.1-1 mol m-3. Samples were withdrawn into a glass syringe, and analysed immediately. Sulphide Electrochemistry 49

The concentrations of poly sulphide ions in the injected samples were calculated from the rate constantsAthe equilibration between species given by Giggenbach [53] (see section 2.1.2). From the initial concentrations of S(0) and S(-II), and knowing the pH, the equilibrium concentrations of the polysulphide species were calculated. The program TKSOLVER,run on a Digital 350 microcomputer, was used to solve numerically the resulting set of simultaneous equations. Corrections were made for the effect of ionic strength by calculating the activity coefficients according to the method of Albert and Serjeant [92]. Sulphide Electrochemistry 50

3.3 Sulphide Voltammetry: Results and Discussion A voltammogram of a gold plated platinum electrode sulphide solution recorded at pH 9.2 is shown in Fig. 3.6. The main oxidation peak at +0.09 V vs. SHE is due to the oxidation of HS" ions producing layers of elemental sulphur. This non-conducting layer inhibits further oxidation and passivates the electrode. On the reverse potential scan the corresponding reduction was not observed until a potential of -0.43 V vs. SHE was reached. This large peak separation is evidence that sulphur formation is a highly irreversible process, as would be expected for a phase formation reaction.

Fig. 3.6 Voltammogram of HS" on Gold Plated Disc Electrode. [HS‘] = 10 mol m-3, pH = 9.2 (Borate Buffer), nth. cycle, 20 mV s_1.

The integrated charge under the oxidation peak was found to be approximately 20 C m-2 (based on the real surface area). A monolayer coverage of sulphur, assuming 2 e" discharge, has been calculated to correspond to charge densities of 3.5 and 2.3 C m r 2 [41,93] (depending on the assumptions made about the sulphur packing). It is therefore apparent that several monolayers of sulphur were formed, and this conclusion is in agreement with those of previous workers [41,42,94]. Sulphide Electrochemistry 51

There was a charge imbalance over a complete cycle, more charge being passed on the positive-going scan. There are several possible explanations for this imbalance: 1. The sulphur layer was not completely reduced. This would imply that the sulphur layer would build up after repeated cycles. However, prolonged potential cycling did not completely passivate the electrode, nor could any visible sulphur deposits be seen. 2. A reaction that produced soluble sulphur oxidation products had occurred in parallel with that of sulphur formation. Possible alternative oxidation products include polysulphides, thiosulphate, sulphite and sulphate. 3. The sulphur layer was reduced to form polysulphide rather than hydrosulphide ions. Polysulphide ions could diffuse into the solution before they were further reduced.

Since the sulphur layer does not build up to completely passivate the electrode after repeated cycles, the reason for the charge imbalance must be either (or both) of the second and third possibilities.

Hamilton and Woods [41] suggested that sulphate production in the positive-going scan, and polysulphide production in both the positive and negative-going scans, were the reasons for the charge imbalance. To investigate the possibility that polysulphide intermediates were formed, ring-disc electrochemical studies were conducted. The gold ring was held at a reducing potential, in order to detect any polysulphide ions, and the disc was subjected to either a triangular potential waveform or a potential pulse.

3.4 Thiosulphate Voltammetry: Results and Discussion Thiosulphate is known to be a metastable oxidation product from sulphide oxidation. Theoretically, it can be oxidised to tetrathionate (at potentials above 0.18 V vs. SHE), or reduced to form hydrosulphide ions (below a potential of -0.3 V vs. SHE). The actual redox behaviour at a gold electrode was investigated to determine whether thiosulphate can, in reality, be reduced in the potential range that is required to ensure polysulphide reduction in a ring-disc experiment.

A gold disc electrode was cycled between the potential limits -0.75 V and 0.3 V vs. SHE in a solution containing 10 mol Na 2S2C>3 m-3 at pH 8.2; no reduction currents above those obtained with the buffer solution alone were observed. Sulphide Electrochemistry 52

If the potential range was increased, voltammograms such as those shown in Fig. 3.7 were observed:

Potential vs. SHE / V Fig. 3.7 Cyclic Voltammograms of Sodium Thiosulphate. [Na2S203] = 10 mol m-3. 1st Scans. 100 mV s-1. pH = 8.2.

In the negative-going scan, it can be seen that no reduction currents were observed until the cathodic limit was reached, when hydrogen was evolved at a potential of -0.7 V vs. SHE. In the positive-going scan, an oxidation peak at 0.6 V vs. SHE can be seen which was due to the formation of gold oxide. At higher potentials, 1.1 V and 1.25 V vs SHE, further oxidation peaks can be seen. However, the magnitudesof these current peaks are approximately an order of magnitude lower than the diffusion limited current calculated from the Levich equation (7.14), even assuming only a one electron oxidation: ^ 2^ 3^" 1/2 S ^ g ^ - + e~ (3.8) This reaction occurred only after an overpotential of almost 1 V was applied. In fact it has been suggested that thiosulphate oxidation proceeds via a chemical reaction with hydrogen peroxide, which is evolved at an anode at these potentials [95].

It can be concluded that thiosulphate is electrochemically inactive at a gold electrode in the potential range of interest to the present study. Sulphide Electrochemistry 53

3.5 Polysulphide Voltammetry: Results and Discussion Polysulphide solutions were prepared by diluting measured volumes of stock solutions (see section 3.2.1) in nitrogenated buffer. Potential scans were commenced from the cathodic limit, or the electrode rest potential (-0.17 V vs. SHE); similar results were obtained in both cases. Typical results are shown in Fig 3.8:

Fig. 3.8 Voltammograms of Polysulphide Solution at a Gold Disc. [Sx] = 1 mol n r3. xav = 2. pH = 8.2. Scan rate 50 mV s-1.

Aqueous polysulphide solutions always contain a proportion of free HS“ ions, and so voltammograms of polysulphide and hydrosulphide solutions are very similar. Commencing at -0.8 V vs. SHE, the first positive-going scan showed a oxidation pre­ wave at around -0.5 V vs SHE. The integrated charge density under this peak (based on the real surface area) was 0.5 C m-2, which corresponds to the discharge of a sub- monolayer (~ 0.2 monolayers) of sulphide ions. Similar peaks have been observed in the voltammetry of dilute hydrosulphide solution by Hamilton and Woods [41,42], and were attributed to the formation of a gold sulphide phase at the electrode surface. This oxidation peak disappeared after prolonged potential cycling provided the positive limit was kept below 0.5 V, under these conditions an adsorbed sulphur layer is likely to be permanently present. If the [HS“] were to be increased, this would have the effect of decreasing the potential at which the phase forms; this explains why this oxidation peak only appeared only as a shoulder on the hydrogen evolution current in concentrated HS‘ solutions.

The main oxidation peak, which is due to the formation of multilayers of elemental sulphur, was observed at 0.05 V vs SHE. When the electrode was rotated, this had Sulphide Electrochemistry 54 little effect on the peak current density, which was much lower than the expected diffusion limited value (ip = 1.04 A m"2, iijm = 6.5 A m-2); moreover the current decreased as the potential was increased above 0.1 V vs. SHE. All this is consistent with sulphur passivation of the electrode surface. However, the fact that there was some increase in the oxidation peak upon rotation indicated that soluble oxidation products were also produced.

In the negative-going scan, at a stationary electrode, the reduction currents at -0.5 V vs. SHE were seen to consist of two waves. These are due to the reduction of multilayer sulphur (at -0.4 V vs. SHE) and the gold sulphide layer (at ~ -0.5 V vs. SHE).

An Eft-pH diagram showing the sulphur polysulphide system is shown in Fig. 3.9.

Fig. 3.9 E^-pH Diagram of the Sulphide/Polysulphide System. [S Species] = 10 mol n r3.

The diagram was produced using the program PPE [89,90], running on an Apple He microcomputer, and the thermodynamic data was taken from Zhdanov's review [31]. The Eft-pH diagram was not adjusted to account for the new value of AGf°(S2-), since the free energies of formation of the polysulphides were calculated from equilibrium potential measurements on sulphide/polysulphide system, and rely on the lower value for pKa 2(H2S) [52]. Despite the uncertainties in the thermodynamic data, it is apparent that at potentials lower than -0.45 V vs SHE, all poly sulphide species are thermodynamically unstable and can be reduced to form S (—11). Sulphide Electrochemistry 55

In Fig. 3.8 it can be seen that the reduction of the polysulphide solutions at a rotated electrode resulted in an increased reduction current below -0.4 V. However, no clear, diffusion-limited plateau was seen in either the positive- or negative-going potential scans. The diffusion-limited current density calculated from the Levich equation (7.14), assuming a diffusion coefficient of 5.2 x 10 ' 10 m2 s-1 [42] and a rotation speed of 20 Hz, is around 4.2 A m -2 for a one electron transfer. The current density at -0.6 V vs. SHE was only 1.8 A m-2. If the lower potential limit was decreased, scans such as Fig. 3.10 were obtained:

Fig. 3.10 Voltammograms of Polysulphide Solution at a Gold Disc. [Sx2-] = 1 mol n r3. xav = 2. pH = 8.2. Scan rate 50 mV s-1.

There was a current plateau at a potential of -0.95 V vs. SHE; it is conceivable that this current might be due to the diffusion-limited reduction of poly sulphide ions: Sn2- + 2(n-l)e“ + nH+ nHS- (3.9) However, this current density is too high to be attributed solely to this reaction; the calculated diffusion-limited value is 8.4 A m -2 for the complete reduction of all polysulphide species, whereas the observed value was 65 A m-2 . Thus, the major proportion of the observed reduction current at -0.95 V vs. SHE must be due to hydrogen evolution (E0' H+/H 2 = -0.48 V vs. SHE at pH 8.2). Rotating the electrode increased the hydrogen evolution current, and adsorbed sulphur is known to have a dramatic effect on the hydrogen overpotential [96]. Sulphide Electrochemistry 56

Therefore, there is a problem in deciding which potential should be applied to detect polysulphide species. Oxidising potentials cannot be used, since elemental sulphur is formed and the electrode passivates. At a potential of -0.6 V vs. SHE, the polysulphide species may not be completely reduced. If the potential were to be lowered to around -0.95 V, a substantial hydrogen evolution current would flow; which may be modified by the local [HS~] (and [H+] ). Buckley et al. [42] showed a clearly defined current plateau at -0.6 V vs. SHE (at pH 9.2), and claimed that the magnitude of this reduction current was consistent with the complete reduction of all the polysulphide species. Nevertheless, they chose a substantially more negative potential in order to detect polysulphides at a gold ring: -0.92 V vs. SHE (unless they have erroneously quoted the potential vs. SCE rather than vs. SHE). At this potential a substantial hydrogen evolution current is likely to flow, and their assumption that this background current is constant (irrespective of surface sulphide concentration and pH changes) is questionable. In the present ring-disc study, a detection potential of -0.75 V vs. SHE was applied.

3.6 Ring-Disc Studies: Results and Discussion To determine whether polysulphide ions were produced from the oxidation of HS" ions, and from the reduction of elemental sulphur, ring-disc electrode studies were undertaken. An experimental problem was that the electrochemical activity of the ring (even when held at -0.75 V vs. SHE) decayed with time. The experiment was conducted after the initial decline in activity had stabilised, and periodically the electrode could be reactivated by potential cycling. Buckley et al.[42] noted a similar problem in their studies, and reactivated the electrode by pulsing to a highly positive potential. This deactivation suggests that sulphur was still adsorbing onto the electrode surface, even at these low potentials, implying that that the polysulphides may not be reduced under mass transport control at the ring. The potential of the ring was maintained at -0.75 V vs. SHE, while the disc potential was swept from -0.75 V to 0.3 V vs. SHE, returning to -0.75 V. The resulting ring and disc currents are shown in Fig. 3.11 (overleaf).

At the disc, similar results to those described previously were observed; a small oxidation peak at -0.53 V vs. SHE was seen in the positive-going scan, as the gold sulphide phase was formed, and the major oxidation peak was seen at 0.065 V vs. SHE. The integrated charge under this peak was 358 qC, which corresponds to 2-3 monolayers of sulphur. Sulphide Electrochemistry 57

Fig. 3.11 Ring-Disc Voltammetry of Sulphide Solution at Au RRDE. [HS-] = 10 mol m"3. co = 9 Hz. Scan rate = 100 mV s-1. Ring potential = -0.75 V vs. SHE.

At the ring, there was little response in the positive-going (disc potential) scan. A slightly increased reduction current was seen, which reached to a maximum value of —2 jiA. Assuming that this was due to polysulphide reduction, and that the polysulphide ions were produced at the disc, the corresponding disc oxidation current would be 11.8 |iA. This represents only about 5 % of the peak oxidation current at the disc, and supports the theory that polysulphides are rapidly oxidised at a gold electrode to form elemental sulphur [42]. Thus, only a small proportion of the poly sulphides that are produced can diffuse into solution and reach the ring.

On the negative-going sweep, a reduction current began to flow at the disc when the potential reached -0.3 V vs. SHE and this increased in magnitude as the potential was decreased further. The integrated charge that was passed in the negative-going scan, after subtraction of the hydrogen evolution charge, was 200 jiC. This includes a Sulphide Electrochemistry 58 component which is due to the reduction of the gold sulphide surface phase; Buckley et al. [42] quoted a value of 0.9 C m "2 for the reduction of this layer , which corresponds to 45 pC on the experimental surface area. Therefore the charge that was passed in reducing the multilayer sulphur, Q^, was 155 JJ.C (which was 43 % of the charge that was passed to form the sulphur).

At the ring, during the negative-going (disc potential) scan, a reduction current was seen which reached a maximum when the disc potential was -0.535 V vs. SHE. This current arose from the reduction of polysulphide ions which were swept out from the disc; if the electrode was stationary, no reduction currents were observed. There are two possible ways in which these polysulphide ions can be produced: from the electrochemical reduction of adsorbed sulphur (3.10), or from the chemical dissolution of sulphur in HS" solution (3.11). n S + 2 e" —» Sn2' (3.10) nS + nHS" + nOH" —» Sn2" + nH20 (3.11) At a disc potential of -0.15 V vs. SHE, no current flowed at the disc, yet a reduction current of about 2 pA below background was seen. This must have been due to the production of polysulphide ions by chemical dissolution of the sulphur layers (3.11). An estimate of the current due to polysulphide production due to the electrochemical reduction of sulphur can be gained by subtracting this 2 qA from the ring reduction currents observed at a lower disc potential. The integrated charge due to the reduction of electrochemically produced poly sulphide, Qr, was found to be 21.2 |iC.

If it is assumed that: 1. At the disc, all the sulphur was reduced to form polysulphide ions according to equation (3.10) 2. At the ring, all the polysulphide reaching this electrode was fully reduced to HS": Sn2~ + 2(n-l)e" + nH+ -> n HS" (3.12) Then the ratio of the two charges will be given by: N Qd/ Qr = 1 / (n- 1) where N = the ring collection efficiency (3.13) From equation (3.13) the average poly sulphide chain length (n) can be calculated. Substituting in the values; = 155 |iC, Qr = 21.2 jiC and N = 0.17 gives n = 1.8.

In a similar analysis, Buckley et al. [42] calculated that the polysulphide had n = 3.3 (at pH 9.2 and [HS‘] = 0.2 mol m~3). It can be concluded that in both cases the reduction product is likely to contain a mixture of polysulphide species. This is not unexpected, bearing in mind the predominance of different polysulphide species as the potential range was scanned (see Fig. 3.9). Sulphide Electrochemistry 59

The reduction of the sulphur layers was found to correspond to 155 p.C, and if this resulted in the formation of a poly sulphide of average stoichiometry 82-, the sulphur must have been deposited with the passage of 1.8 x 155 |iC, i.e. 279 fiC. As the sulphur was deposited, it was also chemically dissolved to form more polysulphide ions. Over the time span of the deposition of the sulphur (about 8 s) a reduction current at the ring of 2 (lA was observed due to the reduction of these polysulphide ions. Thus, at the ring a further charge of 16 |iC was passed. Assuming a collection efficiency of 0.17, the amount of sulphur dissolved would have required the passage of 16 / 0.17 = 94 {iC for its production at the disc. Thus, in total, the 279 + 94 = 373 JJ.C would be expected to have been passed on the anodic scan, which is is approximate agreement with the observed anodic charge of 358 (iC. Therefore, the charge imbalance at the disc can be attributed to the production of polysulphides in both the positive- and negative-going scans; there must have been little or no direct oxidation to produce sulphoxy species.

One criticism of the above approach, is that it is only legitimate to use ring and disc charges rather than currents when the collection efficiency and the composition of the intermediate species remain constant as the disc potential that is scanned. A constant collection efficiency requires that the reduction of polysulphide ions at the ring always operates under mass transport control, irrespective of the polysulphide flux over the ring. The nature of the polysulphide ions that are produced from sulphur reduction are likely to vary with potential, and so the above calculations will only lead to an average value for the chain length of the polysulphide. Voltage pulse studies can provide a more accurate estimate of the polysulphide that is produced under a particular reduction potential. Sulphide Electrochemistry 60

In a highly alkaline, concentrated solution of HS', the chemical dissolution of sulphur is favoured (3.11). A ring-disc electrode study in a solution containing 1 kmol Na2S.9H20 m'3 and 1 kmol NaOH m -3 showed that the polysulphide species could be detected in the positive-going scan, as shown in Fig. 3.12:

Fig. 3.12 Ring-Disc Voltammetry of Sulphide Solution at Au RRDE. [NaOH] = 1 kmol n r3, [HS“] = 1 kmol m“3, co = 4 Hz, nth* scan. Scan rate = 20 mV s"1, ring potential = -0.90 V vs. SHE.

Although a polysulphide reduction current was still detected on the negative-going (disc potential) scan, the largest current was seen during the positive-going scan. Once sulphur layers have been formed, they can be dissolved by the high concentrations of HS" flowing over the electrode surface, to produce polysulphide ions which can be reduced at the ring.

Potential pulse studies also confirmed that polysulphides were produced during the formation and reduction of elemental sulphur. The experiments were conducted in deoxygenated solutions containing 10 mol HS" m -3 at pH 9.3. The ring potential was maintained at -0.7 V vs. SHE throughout, and the disc potential was stepped from this to a more positive value for 4 s, and then stepped back to -0.7 V vs. SHE. The Sulphide Electrochemistry 61

experiment was repeated, with and without electrode rotation, for different values of the disc potential step. The resulting current response is shown in Fig. 3.13:

0 0 Vvs. SHE Us.----- Disc Potential -0 7 V

Fig. 3.13 Ring-Disc Potential Pulse Study. Au RRDE. [HS“] = 10 mol n r 3, co = 9 Hz. pH 9.3.

The potential step to 0.0 V vs. SHE was sufficient to form multilayers of elemental sulphur, and at the ring a small reduction current was seen due to the detection of polysulphides. When the potential was pulsed back to -0.7 V vs. SHE, the sulphur at the disc was reduced, and an increased reduction current was observed at the ring. If the electrode was stationary, the ring current decreased to zero (apart from the capacitive current spikes which were produced when the potential was pulsed). The ratio of ring to disc currents were lower than those expected for the production of a poly sulphide on average chain length 1.8. This suggested that either that less polysulphide was produced (a calculation of the average chain length gave n = 1.1) or that the ring electrode had become deactivated. The ratio of ir to i^ rose rapidly in the first 250 ms following the potential pulse, thereafter remaining approximately constant at about 0.025. The reduction of the sulphur layers, produced when the potential was pulsed to -0.7 V vs. SHE, resulted in the production of a smaller proportion of polysulphide ions than was observed in the potential scan study.

If the disc potential was pulsed to potentials below that at which multilayers of sulphur can form, i.e. no higher than -0.1 V vs SHE, no current response was seen at the ring. When the disc potential was stepped to 0.1 and 0.2 V vs. SHE, higher ring currents were seen while the disc was held at these potentials. This indicated that more polysulphides were formed, either through direct oxidation of HS" ions or by dissolution of the sulphur layer. Sulphide Electrochemistry 62

3.7 Calculated Polysulphide Concentrations vs. Potential A knowledge of the equilibrium concentrations of polysulphide ions as a function of potential can be used in two ways. It can enable the chain length of a polysulphide produced under a particular electrode potential to be predicted, and it can be used to calculate the composition of a polysulphide solution from the solution potential.

The concentrations were calculated by considering the following equilibria: S22" + 2e- 2 S2' Ei (3.14) 2 S32- + 2 e- *-» 3S 22- e2 (3.15) 3 S42" + 2 e ‘ <-> 4 S32" e3 (3.16) 4 S52- + 2 e- 0 5 S42" e4 (3.17) S2- + H+ HS- p K - ^ S ) (3.18) The standard electrode potentials (Ej0- E 40) and pK^CH^S) were calculated from the free energy of formation data in Zdhanov’s review [31]. The redox potentials were all set equal. Since these potentials are dependent on the species concentration via the Nemst equation, a set of simultaneous logarithmic equations are produced, which were solved iteratively using a numerical method and the program TKSOLVER. The process was repeated for a range of potential values, and in this way a profile of the polysulphide concentrations was built up. The results are shown in Fig. 3.14 and Fig. 3.15:

Fig. 3.14 Polysulphide Distribution vs. Potential (pH = 14) Sulphide Electrochemistry 63

Fig. 3.15 Polysulphide Distribution vs. Potential (pH = 9)

These figures indicate that the potential ranges of predominance of the individual polysulphide species are small. A range of only 100 mV separates the areas of stability of S 52- and S 22-. This means that using the solution potential to calculate the polysulphide concentration is insensitive; a deviation of only a few millivolts in the recorded potential would seriously alter the calculated composition. Furthermore, when gold indicator electrodes were used to measure the potential of polysulphide solutions, values outside the theoretically predicted likely potential range were obtained; e.g. -0.14 V vs. SHE for a solution containing 0.1 mol Na 2S4 m~3 at pH 9.3. Solutions made from Na 2S4 disproportionate, forming a range of polysulphide species in equilibrium. If these species were not equilibrating reversibly at the gold electrode surface, the observed solution potential would be outside the theoretical range. Inaccuracies in the published thermodynamic data may also account for this difference. Certainly there is a conflict between equilibrium constants that are calculated from the thermodynamic data, and those which have been determined experimentally [ 53].

However, the above figures do indicate that there will always be more than one polysulphide present in significant amounts at any particular solution potential (if the species can equilibrate under thermodynamic control). Sulphide Electrochemistry 64

3.8 The Detection of Polysulphides Using Ion Chromatography Because of their possible importance as an oxidation intermediates in the Stretford Process, it was decided to investigate the possibility of detecting poly sulphides using the technique of ion chromatography. Although polysulphides absorb in the UV-visible spectral range, spectrophotometry can not be used routinely since other components of a Stretford solution absorb in the same spectral region.

3.8.1 Ion Chromatography: Results and Discussion Using the calculations outlined in section 3.1.4, it was determined that only three solution species were important in polysulphide mixtures in the pH range 10-14; S 42", S52" and HS". An injection of polysulphide solution resulted in the detection of three peaks, as can be seen from Fig. 3.16:

Electrochemical Detector , Response 20-

1-5-

1 0-3 mM HS"

2 10 mM Sj?" UV Detector Response 3 10 mM H$ Saturated with elemental Sulphur

£ 00005 cJ j O c_ IS)O 4

O l 3 O

Fig. 3.16 Ion Chromatography Results Sulphide Electrochemistry 65

The peak heights of the second and third peaks (those two peaks having the longest retention times) correlated well with the calculated concentrations of S 42' and S 52". However, the response of the first peak, which was assigned to the HS' ion, was larger than expected. This suggested that disproportionation had occurred as the sample traversed the column.

3.9 Summary Oxidation of HS" ions at pH 9.3 has been shown to produce a sub-monolayer of adsorbed sulphur on a gold electrode at low potentials (-0.4 V vs. SHE), and multilayers of sulphur at higher potentials (0.05 V vs. SHE). Associated with the formation of elemental sulphur is the production of polysulphide anions, Sn2" (n = 2 to 5), which can also be produced by the dissolution of the initial sulphur layer. The production of such polysulphide species accounted for the difference in charge between the positive and negative-going scans.

Upon reduction of the sulphur layers, polysulphide ions were produced, which were detected at a ring electrode in a rotating ring-disc electrode study. By a comparison of the charges passed in the production of these ions from elemental sulphur, and their reduction to HS" ions, an estimate of the average polysulphide chain length could be gained. This was calculated to be 1.8, which suggests that a mixture of poly sulphides was produced. This is consistent with thermodynamic predictions which show that a number of polysulphides can exist in solution at comparable concentrations at any given potential.

Ion chromatography was investigated as a means of detecting polysulphide ions in solution, but disproportionation of the polysulphide species as they traverse the ion exchange column and the air-sensitive nature of the solutions, made the method unsuitable for routine use. Vanadium Review 66

4. Vanadium Vanadium is a lustrous, corrosion-resistant metal which is used in large quantities to make steel alloys tougher. There are only a few concentrated deposits of vanadium minerals, and most vanadium is generated as a co-product from the processing of iron, phosphorus or uranium ores. The vanadium content is dissolved from these ores under oxidising and acidic conditions, forming a solution containing the VC> 2+ ion; the pH is then raised and vanadium anions are formed (see section 4.1) which can be separated from the aqueous solution using solvent extraction. Recently ion-exchange columns have also been suggested for this purpose [97]. The vanadium is then stripped from the organic solvent or ion-exchange resin to form a concentrated aqueous solution, and precipitated as the oxide. Conventional pyrometallurgical reduction of the oxide is not facile, since vanadium reacts with oxygen, nitrogen and carbon at high temperatures; instead, most vanadium is produced as the iron alloy, ferrovanadium. Very pure vanadium can be produced by the reduction of the chloride VCI 4 with magnesium, the reduction of the pentoxide V 2O5 with aluminium, or using the Van Arkel method which relies on the decomposition of the iodide VI 3 at high temperatures.

Vanadium has atomic number 23 and mass number 51, and is in the first row of group VB in the transition series; it can exist in oxidation states between II and V in aqueous solution. Vanadium, like other metals in this region of the periodic table (Nb, Ta, Cr, Mo and W) shows a marked tendency to form polymeric ions in aqueous solution, which makes the chemistry of vanadium both interesting and complex.

Various authors have produced E^-pH diagrams showing the thermodynamically most stable species [61,90,97-100]. These diagrams differ from each other for a variety of reasons. Firstly, AGf° values for the polymeric vanadium species are not always available; Pourbaix [61], for instance, did not consider the formation of any decavanadate species, although their existence and AGf° values have now been well established. Secondly, lines are drawn between two solution species under different criteria, Pourbaix used the criteria of equal activities, whereas Post and Robins [99] drew the lines to show when the total dissolved vanadium was distributed equally between the two species. When polymeric species are involved, this can lead to a considerable shifting of the equilibrium lines. Finally, diagrams are drawn under different total vanadium concentrations; Zipperian and Raghavan produced a diagram using a very low vanadium concentration of 0.2 mol m -3 and as a result show only monomeric species, whereas Post [98] considered the equilibria at a total vanadium concentration of 1 kmol n r 3 and showed the polymeric species V 2074- and V10O286" had areas of predominance.

Post [99] also noted that some of the available thermodynamic data on vanadium had been misprinted in the past, which had passed unnoticed by other authors; he corrected these values where necessary: Fig. 4.1 (overleaf) shows a diagram using his corrected data [98]: Vanadium Review 67

pH Fig. 4.1 Ejj-pH Diagram for the Vanadium-Water System [98] 1 kmol V m-3 T = 298 K.

4.1 Vanadium (V) Vanadium (V) solutions can be prepared by dissolving sodium metavanadate (NaVC^), ammonium metavanadate (NH 4VO3), or vanadium (V) oxide (V 2O5). It is the only oxidation state that is stable in air throughout the entire pH range. Babel et al. recently reviewed the use of vanadium (V) as an oxidising agent in acidic media [101]; they showed that it could be used to oxidise many organic and inorganic compounds (e.g. alcohols to aldehydes and SC^2- to SO 42'). Under these conditions, the predominant species are VC>2+ ions. If such a solution is made alkaline, tetrahedral VO 43- ions are produced. As dilute (< 0.01 mol V(V) m“3) alkaline solutions are made more acidic, VO43- ions protonate to form HVO 42- and H 2VO4- (at pH 13 and 8 respectively). Hydrated vanadic acid (usually written as HVO 3) precipitates at about pH 5 [102].

The situation is more complex with higher V(V) concentrations; consider the acidification of a moderately concentrated vanadium (V) solution (10 mol m-3) at pH 14. Initially, only VO 43- ions are present and the Raman spectrum is simple. As the pH falls below 14, new bands appear at 810, 503 and 228 cm-1, which have been assigned to V-O-V stretches [103]. This is consistent with the presence of the dimeric V2O74- ions. Similar changes are seen in the 5iV NMR spectrum [104]; in strongly alkaline solution a single absorption at 536.2 ppm (relative to VOCI 3) is observed, due to VO 43- ions, whereas below pH 14 this peak shifts slightly to 533 ppm and a second peak is seen at 556.2, due to the V 2O74- species. If the pH is further lowered to below Vanadium Review 68

11 this line broadens and shifts to 562 ppm, which is consistent with the protonation of these ions to form HV 2O73-. The presence of dimeric species is confirmed by the precipitation of Na 4V207 at concentrations in excess of 1 kmol V(V) m "3 .

In the pH region 7-8 there were early disagreements as to whether the vanadium was present in trimeric or tetrameric species. However, Ingri and Brito [105] showed that below 20 mol V(V) m"3, the V 3093" ions predominate, and at higher concentrations V4O124" ions are formed. Habayeb and Hileman studied vanadium (V) speciation using 51V NMR [32] and confirmed the presence of V 40i24"- They also collated the chemical shift data for the known vanadium (V) species; this compilation is shown as Table 4.1: Species 51V NMR Shift / ppm vo43- -536.2 hvo42- -533 v 2 o 74- -556.2 HV20 73- -562 V30 93- -573 V3O105- -410, -500, -516 1

O -577 h2v 4o 134- -582 V50 155- -586 O I> 1 -582

V 1 0 ° 2 8 6' -419,-495,-510

Table 4.1 51V NMR Chemical Shifts of V(V) Species.

Around pH 6.5, the species V 50i5^“ and V^Ojg6- have been reported [33], and they are seen as possible precursors to the decavanadate structure. It must be noted that acid titrations are difficult to interpret in this pH region, since some of the equilibria, especially those involving polymeric anions, are achieved very slowly. This may account for some of the early disagreements regarding vanadium (V) speciation. However, there is now no doubt that between pH 2 and 6, and at vanadium concentrations greater than 1 mol m-3, the orange decavanadate ions are formed; Vio028^“> and ^ V jqC^s4'- The presence of these species has now been established by potentiometric [105], cryoscopic [106], and spectroscopic [107] methods. Vanadium Review 69

Solid salts containing decavanadate anions have been isolated, and the minerals pascoite (Ca3Vio028-7H20 ) and hummerite (K 2Mg 2Vio 028-16H20 ) have been shown to contain them [108]. The structure of the V^gC^s6- ion consists of ten linked VOg octahedra [109]:

Fig. 4.2 Structure of the Viq0 286' ion.

If decavanadate solutions are allowed to stand for several weeks, or if they are warmed, sparingly soluble orange salts precipitate. Although they are termed trivanadates, and have stoichiometries such as KVgOg, they do not contain V 30g" ions, instead having a structure consisting of layers of linked VOg octahedra separated by layers of cations [110]. Vanadium Review 70

The available information on vanadium (V) speciation has been summarised in the form of an activity-pH diagram by Post and Robins [99]:

Fig. 4.3 Vanadium (V) Speciation [99].

4.2 Vanadium (IV) Vanadium (IV) is stable in aqueous solution provided that oxygen is excluded. Post [98] and Rossotti and Rossotti [111] showed that the predominant species below pH 3 was the blue vanadyl ion, V 02+. Above this pH they suggested that the two complexes, VO.OH+ and (VO) 2(OH)22+, were formed. Rohrer et al noted that a series of solid sulphate complexes were obtained between V 02+ and concentrated sulphuric acid [112]. At about pH 4 a solid precipitates from aqueous solution which has the composition VO(OH )2 (which could be regarded as V 2O4. 2H2O). They noted that this was soluble in excess alkali, forming brown, air sensitive "vanadite" solutions. Vanadium Review 71

Pope has recently reviewed the vanadium (IV) speciation in such alkaline solutions [33]. Crystalline alkali metal salts can be precipitated from vanadium (IV) solutions, and although they have the empirical formula M2V3C>7.nH2C), they have been shown to contain the polyanion Vig 04212"[113 ]. This anion consists of an almost spherical shell of linked VO 5 square pyramids, surrounding a central cavity about 0.45 nm in diameter:

Fig. 4.4 Structure ofV i804212‘.

In the solid salts the central cavity is occupied by a potassium ion or a water molecule. The anion appears to be stable in vanadium (IV) solutions between pH 9 and 13, and at concentrations above 2 mol V(IV) m"3 [33]. As it has only recendy been identified, no value for AGf° has yet been proposed, and for this reason it does not appear on any of the published E^-pH or activity-pH diagrams.

Vanadium (IV) has a magnetic moment of 1.73 Bohr Magnetons, which means that, unless the spins are completely paired, the 51'V NMR spectra are poorly resolved. This has limited the information available concerning the vanadium(IV) speciation in solution, and much less is known about vanadium (IV) anions than vanadium (V) anions. Post and Robins [99], although omitting any mention of the V ^gC ^12- ion, do show four vanadium (IV) species: V02+ (low pH), HV 2O 5- (high pH), (VO)2(O H )22+ and V 4O92- (at high V(IV) concentrations). The situation is summarised in Fig. 4.5 (overleaf). Vanadium Review 72

1 2 3 4 5 6 7 8 9 10 11 12 13 14

pH

Fig. 4.5 Vanadium (IV) Speciation in Solution [99].

4.3 Vanadium (V)/(IV) Compounds Several mixed-valence soluble polyvanadates have been reported in the literature; they have been formulated as partially reduced decavanadate structures, e.g.HV 3IVV7V0286“. Ostrowetsky reported six ions in the pH range 4 to 6.5 with y iv :yv ratios ranging from 2:8 to 7:3, of which the green 3:7 and 7:3 ions were the most stable [114]. Solid mixed-valence alkali metal vanadium oxides are also known [115]. Many of these compounds are semiconductors, and some of them show a metallic lustre; they have been termed "vanadium oxide bronzes" e.g. K 2V3O8.

It is possible to prepare a range of 19-nucleate blue-violet anions having y lv:VV ratios from 5:14 to 7:12 [33] (e.g K g H V ^ V V ^ O ^ .ll^ O ). The basic structure of these anions consists of an ellipsoidal cluster of 18 VOn polyhedra. These formulae are almost twice that of those proposed by Ostrowetsky [114] and it is possible that his assumption that the ions are based Y\ q clusters is incorrect..

Hayek and Pallasser [116] obtained the mixed-valence crystalline decavanadate structures Na 6VIV8Vv20 24.8H20 and K 6Vlv8VV20 24 .5H20; they acidified thiovanadate solutions containing V:S ratios in the range 1:1 to 1:4 using acetic acid. Vanadium Review 73

The pH was lowered to about 8.5, and after 12 hours heating the brown-black crystals precipitated from solution. The sulphide solution had effected the partial reduction and itself been oxidised to elemental sulphur, which was removed by washing with carbon disulphide. The mixed-valence ammonium salt (NH 4)2V30g.l /2 H2O was prepared in a similar manner, thought it did not to contain a decavanadate ion, but instead consisted of linked VIVC >5 square pyramids and VV 207 di-tetrahedral units. It is interesting to note that even in the presence of excess reducing agent, the vanadium (V) is not completely reduced to vanadium (IV) compounds, but instead forms a mixed-valence precipitate.

Post [98] studied the atmospheric oxidation of vanadium (IV) solutions in acid solutions, and found that at pH 2.5 and in the presence of sodium ions, blue or brown mixed oxidation state solids were obtained. He reported one solid with a Vv: V ^ ratio of 1:4 which had similar properties to the mineral corvusite (~V 2C>4 8 .1/2 H2O). Oxides with intermediate V/TV stoichiometries such as VgO^ can also be produced by heating the appropriate masses of the oxides V 2O5 and V 2O3 at 600 °C for 10 hours [117]. VgO^ can also be prepared by reducing V 2O5 in a stream of hydrogen [118]. Vanadium Review 74

4.4 Vanadium (III) The oxide V 2O3 is not amphoteric, unlike the vanadium (IV) and (V) oxides. It is insoluble in alkaline solutions, but dissolves in acid forming green V(H 20)63+ ions. Above pH 1 these hydrolyse to form VOH2+ and V 2(OH)24+, and if the pH is further raised to 7, V 2O3 precipitates. The vanadium (III) speciation is summarised in Fig. 4.6:

Fig. 4.6 Vanadium (III) Speciation in Solution [99].

4.5 Vanadium (II) Vanadium (II) represents the lowest accessible oxidation state of vanadium. The electronic configuration is cfi, which confers upon the aqueous species a kinetic inertness, and the ligand substitution reactionsAthe purple V(H 20)62+ ion are slow. It is a powerful reducing agent, and is oxidised by water. Because of this instability little is known of its hydrolysis behaviour [98]. It is readily oxidised by atmospheric oxygen, forming the green V(H 20)g 3+ ion, and it has been used to remove trace amounts of oxygen from inert gases [ 100]. Vanadium Review 75

4.6 Vanadium Electrochemistry Most studies on the electrochemistry of vanadium have been carried out in highly acidic solution, where each of the oxidation states V(II), V(III), V(IV), and V(V) can be produced by controlled potential electrolysis [100]. A comprehensive review of the work done up to 1976 is provided by Israel and Meites [30]. In acidic solution polarography of V(V) solutions is difficult because V(V) is capable of oxidising a mercury surface. However, in alkaline solution mercury metal should still be stable at potentials high enough to oxidise V(IV) to V(V) [61]. One problem with the interpretation of polarographic results is that vanadium coatings may catalyse other reactions; vanadium (V) has been shown to be an electrocatalyst for carbon oxidation [119] , and vanadium alloys reduce the overpotential required for hydrogen production [120] .

4.6.1 The V(V)/V(IV) Couple In acidic solution, vanadium (V) causes oxidation of mercury and platinum electrodes, and consistent pre-treatment is required to obtain meaningful results. In the pH range 7 to 10, of most relevance to the present study, relatively few investigations have been attempted.

Below pH 2, the reduction of a vanadium (V) solution proceeds in two stages; the reduction from VC> 2+ to V 02+ is reversible, but a further decrease in electrode potential causes the production of V(H 20)g2+ [100]. Magri-Elouadseri and Vittori [121], studied the electrochemical behaviour of carbon paste electrodes incorporating vanadium (V) solids at pH 0. They found that the vanadium (V)/(IV) couple was reversible and gave a half wave potential close to the expected standard potential (0.944 vs. SHE), but that at more negative potentials subsequent reduction proceeded directly to produce vanadium (II).

Van den Berg and Huang [122] conducted polarography on vanadium (V) at concentrations of 0.02 mol m -3 and at pH 7. They found that the vanadium (V) underwent reductive adsorption at potentials of -0.678 V vs. SHE, forming vanadium (IV) on the mercury surface, which was further reduced to vanadium (II) at a potential of —1.0 V vs SHE. Between pH between 2 and 9, up to four reduction waves were observed by Filipovic et al. [123]. The first two they assigned to the adsorption and reduction of hydrogen polyvanadate ions, the third was attributed to the reduction of dissolved vanadium (V) to vanadium (IV), and the fourth to a further reduction to form vanadium (II). From pH 9 to 12.5 they observed only the third and fourth waves. They noted that the reduction to vanadium (IV) occurred only after an overpotential of about 0.8 V had been applied, (e.g. when a potential of —1.16 V vs SHE was reached at pH 9.3). At solutions with a pH higher than 12.5, Filipovic [123] and other workers [30] have found a single irreversible reduction wave which has been attributed to the reduction of V(V) to form V(II). Vanadium Review 76

Stromberg et al. [124] found that vanadium oxide films were obtained when reducing potentials were applied to platinum or carbon electrodes in alkaline vanadate solutions. They proposed that the initial films (on carbon electrodes) consisted of V 2O3, and that this was converted to V 2O2 at potentials lower than -1.15 V vs SHE. However, they had no direct evidence of the film compositions, and relied solely upon thermodynamic predictions.

4.6.2 V(IV) Reduction It appears that the reduction of V(IV) at the dropping mercury electrode in acid media is totally irreversible and proceeds directly to V(II) [100,123]. At a carbon paste electrode, the reduction of V(IV) to V(III) was found to be very slow [121], and the potential had to be lowered until a reduction process forming V(II) occurred.

Gala et al. studied the deposition of vanadium from vanadium (IV) solutions onto steel cathodes at pH 10 [120]. They noted that in the metal could not be deposited from solutions containing vanadium alone, but suggested that elemental vanadium could be co-deposited with nickel, forming an alloy. To achieve this, the potential had to be lowered to such a level that hydrogen was also evolved. However, their analysis techniques did not distinguish between vanadium in a nickel alloy, and entrained grains of vanadium oxides. Indeed, they did not consider the possibility of entrained phases.

4.6.3 The V(III)/V(II) Couple The reduction of vanadium (III) in acid solution is reported to be reversible on mercury and carbon paste electrodes [100,121]. Filipovic et al. [123] report a half wave potential of -0.29 V vs. SHE, which is close to the expected standard potential of -0.263 V vs. SHE. In alkaline solution vanadium (III) forms solid V 2O3.

4.7 Oxidation of Vanadium (IV) Solutions using Oxygen The oxygen/water half cell can apply sufficient potential to oxidise vanadium (IV) solutions: E °(02/H20) = 1.23 V, E°(V(V)/V(IV)) = 0.944 V (at pH 0). Since the potential of the vanadium(V)/(TV) couple decreases with pH at a greater rate than the O2/H2O couple, the thermodynamic driving force for vanadium (IV) oxidation using oxygen increases with pH. This explains why acidic vanadium (IV) solutions can be handled without taking any special precautions to exclude air, whilst alkaline solutions are air sensitive.

Post [98] studied the oxidation of vanadium (IV) by oxygen under acidic conditions. Working at a temperature of 90 °C, he noted that the oxidation proceeded with a decrease in pH, due to reactions such as: 4 V02+ + 2 H2O + O2 —^ 4 V02+ + 4 H+ (4.1) This change in pH may alter the predominant vanadium solution species, and so alter the reaction mechanism. Vanadium Review 77

Dean and Herringshaw [125] looked at the air oxidation of vanadium (IV) in alkaline solution. They showed that oxidation to V(V) was rapid and complete (0.8 mol V(TV) m-3 being completely oxidised by the dissolved oxygen in air saturated solutions in 10 s at pH 14). They noted that under conditions of excess oxygen, hydrogen peroxide was produced as the oxygen reduction product, and that this itself was capable of oxidising more V(IV).

4.8 Vanadium Sulphides Vanadium can form a number of solid sulphide phases with varying ratios of V:S, some of which have not been fully characterised [126]. Like the oxide phases, mixed- valence compounds are known, and most vanadium sulphides possess sulphur-sulphur bonds. Mixed metal Mo-V sulphides are also known [127]. Vanadium sulphides are electrically conducting, and show paramagnetic behaviour owing to the presence of unpaired electrons. Table 4.2 shows some of the known vanadium sulphides (thermodynamic data are from Mills [128]):

Compound Comments AG f° / k j mol -1

v 3s Can exist in two metallic forms. V5S4 Metallic structure VS Non-stoichiometric solid -192 v 7s8 Hexagonal Structure. V3S4 Layered structure V2S3 Prepared by direct reaction -518 v 5s8 Monoclinic structure. V2S5 Prepared by decomp, of (NH 4)3VS4 v s 2 Non stoichiometric vs4 Exists as mineral Patronite -413 vs5 Amorphous semiconductor

Table 4.2 Some Known Vanadium Sulphides.

4.8.1 V3S, V5S4, VS The compound with the highest V:S ratio is V3S, which can exist in two forms, both of which are metallic. V5S4 is also reported to have a metallic structure [126]. As the sulphur content is increased the metallic character is lost and the compounds become semiconducting. Stoichiometric VS is unstable at room temperature and dis- proportionates to form cation-deficient V7Sg and cation rich V 9Sg; V7Sg has a hexagonal NiAs-type structure. Vanadium Review 78

A non-stoichiometric range of compounds with formula from Vq^ sS to Vq^ S are also known. Within this range, the compounds V 2S3 (Vq^ S ) and V 3S4 (V0/75S) have been prepared. V 2S3 can be prepared by heating vanadium pentasulphide to 300 °C in an inert atmosphere [129]: V2S5 -> V2S3 + 2S (4.2) V3S4 is prepared by combination of the elements at 800-1000 °C; it has a monoclinic unit cell and is thought to consist of alternate layers of V2+ and V3+ ions. V 3S4 absorbs water at room temperature, and loses H 2S when it is heated, forming an oxide phase. If V3S4 is heated in air or oxygen, it oxidises to form V 2O3, V2O4 and V 2O5 successively, evolving SO 2 [130].

4.8.2 V2S5 Vanadium pentasulphide is the sulphur analogue of vanadium pentoxide. It can be prepared by heating ammonium tetrathiovanadate (see section 4.9) to 100 °C in an inert atmosphere, whereupon ammonia and hydrogen sulphide are evolved [129]: 2 (NH4)3VS4 -> V2S5 + 6 NH3 + 3 H2S (4.3) V2S5 is a black amorphous powder, insoluble in water, alcohol, ether or carbon disulphide. If it is heated above 290 °C in the absence of air it decomposes to form V2S3; if air is present, it oxidises readily at 100 °C forming vanadium pentoxide: 2 V2S5 + 15 0 2 -> 2 V20 5 + 10SO2 (4.4)

4.8.3 VS2 and VS 4 Stoichiometric vanadium disulphide is not known, although a compound of stoichiometry Vj 2^2 has been reported. If the proportion of sulphur is further increased, VS 4 is produced. VS 4 exists in nature as the mineral patronite, and can be prepared in the laboratory by heating the elements together at 400 °C for several weeks. The structure is monoclinic,the vanadium atoms sitting in between S22- pairs; in this respect the mineral is similar to pyrite (FeS2).

4.9 Vanadium-Sulphur complexes If hydrogen sulphide is passed into an alkaline solution containing V, Mo or W anions a range of colours are produced. These colours are due to the thioanions of the transition metals and depending on the metal, pH, and metal to sulphide ratio, virtually any colour can be produced. Muller [131] recently reviewed the transition metal thiometalates. He noted that, as hydrogen sulphide was passed through an aqueous oxometalate solution, changes in the UV-visible and IR raman spectra were consistent with the successive formation of MO 411", MC^S11-, MO2S211", MOS3n_, and MS 4n‘ (M = V, Mo, W or Re). The rate of formation of these thiometalates was governed by the polarising power of the central metal atom; the lower the polarising power of the metal, the greater the electron density on the oxygen atoms and hence the faster the rate of complex formation. Vanadium Review 79

In aqueous solution thiometalates are not very stable, especially at low pH. They can be hydrolysed to form oxometalates, they can form solid metal sulphides, or they can undergo intramolecular redox processes: MfS2’^ -> Mr-2(S22-) (e.g. M = Mo, r = 6) (4.5)

Thiometalates can be attacked by nucleophiles to give a reduced metal centre, and this process of sulphur abstraction is more apparent in vanadium than molybdenum complexes: Mr-S + Nu Mr_2 + NuS (e.g. Nu = CN‘) (4.6) This reaction may explain why when VO43- was reacted with H2S in the presence of CN“ a polyvanadate with a low vanadium valence was obtained [132].

Ranade et al.[133] prepared a series of thiovanadate complexes by passing H 2S through weakly buffered ammoniacal solutions containing 0.1 mol V(V) m -3 at 5 °C. They found that initially a complex was formed which absorbed at 360 nm in the UV- visible spectrum (and weakly at 305 and 460 nm). This spectrum was similar in structure to the isoelectronic complex M 0O2S22", and so they attributed it to V 02S23-. (Because of the high affinity of vanadium for sulphide, the monothiovanadate (VO 3S3') could not be produced in aqueous solution, although it could be formed in a methanolic solution). As further H 2S was passed through an aqueous solution the trithiovanadate and tetrathiovanadate complexes were formed: VO2S23- + H2S VOS33" + H20 (4.7) VOS33- + H2S VS43" + H20 (4.8) Yatsimirskii and Zakharova [134] studied the hydrolysis of VS 43-. They concluded that on dissolution in sodium hydroxide solutions at pH 13-14, solid ammonium tetrathiovanadate dissolved with hydrolysis to form V 02S23-, and that this rapidly hydrolysed further to form the vanadate ion, VO 43".

From concentrated solutions, solid salts containing the tetrathiovanadate ion can be prepared. Busine and Tridot prepared the ammonium salt [129] by passing hydrogen sulphide through 5.9 kmol m ~3 ammonium sulphide solution containing 37 mol V(V) m-3; after several days at O °C intensely-coloured violet crystals were produced. If a more concentrated vanadium (V) solutions was used, or if the temperature was raised, the vanadium (V) became reduced and vanadyl hydroxide (VO(OH)2) precipitated. This precipitate would redissolve in excess ammonium sulphide, suggesting that vanadium (IV) thiosalts can also be produced. Vanadium Review 80

Harrison and Howarth [37] folloy&d Busine and Tridot's method and prepared a range of thiovanadate complexes. -Whey determined the 51V NMR shifts (relative to VOCI3) of the free anions and their.Trotonated forms. Their results, together with a summary of the UV-visible spectr&/(from [133]), are summarised in Table 4.3.

51V NMR Species Colour UV-visible Absorbances Chemical Shift X / nm (e / m 2 mol*1) ppm vs. VOCI3 VO43- Colourless -541 V 03(0H)2- Colourless -539 V03S3- (orange) 305,442 (in methanol) -250 HV03S2' -121

V 0 2 s 2 3 - Yellow/red 305, 360 (8-400), 460 (weak) 184 HV02S22- <230 VOS33- Red 295, 325, 459 (e~600), 521 740 HVOS32- 748 VS43- Violet 267, 351, 394, 538 1395 HVS42- 1392

Table 4.3 Spectral Summary of Thiovanadates [37,133].

4.10 Summary Vanadium, in common with other transition metals in the same region of the periodic table, shows a tendency to form polymeric anions in alkaline solutions. The degree of condensation in these species is highly dependent on the total vanadium concentration. Efo-pH diagrams are only of limited value in determining the predominant vanadium species under particular solution conditions. This is partly due to the above dependence of the vanadium speciation upon concentration, and partly because thermodynamic values are still not available for key polymeric species. However, in Stretford Process solutions it is likely that the dimeric and tetrameric species HV^C^" and V4O124" ^ present.

Hydrogen sulphide can interact with; vanadium (V) in two ways; as a reducing agent and as a complexing agent. Complete reduction of Stretford Process solutions to the V(IV) oxidation state is likely to produce the brown polyanion, V ig C ^12". On prolonged exposure to reducing environments it is conceivable that further reduction will occur, forming a precipitate of.vanadium (III) oxide, V 2O3. Mildly reducing conditions, or re-oxidation of vanadium (IV) solutions, can produce mixed-valence (V)/(IV) compounds (e.g.VIvgVX2024^~)* The sodium salts of these ions may precipitate if the sodium ion concentration is high. Vanadium Review 81

Thio complexes are known to be produced when (ammoniacal) vanadium (V) solutions contact H 2S. As sulphur is substituted for oxygen in the vanadate ion (VO 43-) the complexes VO 2S23-, VOS33- and VS 43- are formed.

Since it is known that isoelectronic thiomolybdate complexes can undergo intramolecular redox processes, it is possible that vanadium (V) catalysis of sulphide oxidation proceeds via thio-complex formation followed by an intramolecular redox reaction. Thus, initially a complex such as Vv 02S23- might be formed, which is converted to Vm 02S23-; the disulphide ion so produced may then desorb from the complex. In this way vanadium (V) could oxidise sulphide solutions producing polysulphide solutions and reduced vanadium species.

The electrochemical reduction of vanadium (V) in alkaline solution is slow, and large overpotentials are required to obtain measurable currents; the formation of thio- complexes may offer reaction pathways with a lower activation energy, and so allow a higher rate of sulphide oxidation. Vanadium Electrochemistry 82

5. Vanadium Electrochemistry The reduction kinetics of vanadium (V) in solution at pH 9 was investigated at a variety of electrode surfaces. In the absence of specific chemical interactions, oxidising agents that show reversible behaviour at electrode surfaces are reduced rapidly by chemical means, whereas oxidising agents which show irreversible reduction at an electrode react only slowly with chemical reductants.

5.1 Vanadium Electrochemistry: Experimental Cyclic voltammetry and potential pulse studies were carried out using a Tacusel hanging mercury drop electrode (HMDE). The electrode consisted of a mercury reservoir connected to a glass capillary. By turning a micrometer drive, a drop of mercury was made to hang from the capillary tube. The bore was cleaned prior to use with 6 kmol HNO 3 m~3 and triply distilled water, then rendered hydrophobic by treating it with a solution of dimethyldichlorosilane (2 % in 1,1,1-trichloroethane, BDH). A diagram of the HMDE is shown in Fig. 5.1:

/ /

Micrometer thread

\

Electrical contact

f . Silicone rubber seal

Mercury resevoir

Glass capillary

iHMercury bead

Fig. 5.1 Hanging Mercury Drop Electrode. Vanadium Electrochemistry 83

The surface area of one drop was calculated by making 25 complete revolutions of the micrometer drive, and measuring the mass of mercury ejected. From this, the average mass of mercury ejected by one revolution was derived; assuming that the drop had a spherical shape and knowing the density of mercury, enabled the surface area to be calculated (1.974 x 10 ~6 m2). The capillary bore was 100 |im in diameter, and it was shown that the area of attachment corresponded to only 0.4 % of the total drop area.

Gold, platinum and vitreous carbon rotating discs (see section 3.2) were also used as working electrodes. Bright platinum counter electrodes were used in all cases, and the potentials were controlled relative to saturated (KC1) calomel reference electrodes (EIL). All potentials are reported versus the standard hydrogen electrode (SHE), assuming that the potential of the saturated calomel electrode was 0.242 V vs. SHE. The electrochemical studies were carried out in a three compartment cell (see Chapter 7, Fig. 7.1) using a Thompson Ministat MP81 potentiostat. The control potentials were provided by a Hi-Tek PPR1 waveform generator and the currents were passed through a standard resistor. The resulting voltages were then applied to the inputs of a JJ PL4 chart recorder.

Most studies were undertaken at room temperature (~20 °C), but a series of experiments were recorded at 40 °C using a jacketed electrochemical cell which contained heating water maintained at 41 °C by a Grants thermostatted water bath. Voltammograms were commenced from the positive potential limit (0.331 V vs. SHE for a mercury electrode) for vanadium (V) solutions, or from the rest potential for vanadium (IV) solutions (-0.193 V vs SHE).

5.1.1 Solution Preparation A carbonate buffer of pH 9.3 was prepared by dissolving the appropriate mass of analytical grade chemicals (BDH) in triply distilled water to produce a solution containing 0.059 kmol Na 2C03 0.223 kmol NaHCC^ nr3 and 0.1 kmol Na2S04 m~3. A borate buffer of pH 9.2 was similarly prepared, containing 12.5 mol Na2B4Oy.lO H20 m"3,0.9 mol NaOH m "3 and 0.1 kmol Na 2S04 m-3.

A stock solution of 0.1 kmol V(V) m ~3 was prepared by dissolving the appropriate mass of NaVC >3 (BDH) in the carbonate buffer. The white crystals dissolved slowly with conventional stirring, but the dissolution rate could be increased by placing the flask in an ultrasonic bath. V(V) solutions were also prepared by dissolving vanadium pentoxide (BDH) in dilute sodium hydroxide, according to reaction (5.1): V2O5 + 3 NaOH —> HV20 73- + H20 + 3 Na^* (5.1) Vanadium Electrochemistry 84

The colourless stock solutions could be kept for many months without degradation, and they were diluted with the appropriate buffer solution before use. All solutions were thoroughly deoxygenated by sparging with White Spot grade nitrogen (BOC) for at least an hour before any electrochemical investigations were commenced. Identical results were obtained from vanadium (V) solutions which had been prepared from sodium vanadate and vanadium pentoxide starting materials.

A stock solution containing 10 mol vanadium (IV) m -3 was prepared by dissolving 0.635 g of blue vanadyl sulphate, VOSO 4.6H2O (BDH), in 250 cm 3 of oxygen-free carbonate buffer. Since the predominant V(IV) species are thought to be V ig C ^12" ions (see section 4.2), 6.7 cm 3 of 1 kmol NaOH m -3 solution was added to allow for the hydroxide ion consumption during reaction (5.2): I8 VOSO4 + 48 OH' -> V180 4212- + I8 SO42- + 24 H20 (5.2) The resulting dark brown solution was diluted tenfold with an oxygen-free buffer solution before electrochemical studies were made.

A solution of the complex VS 43- was prepared according to the method of Harrison and Howarth [37]. 10 cm3 of aqueous ammonia "0.880" (BDH) was added to 90 cm 3 of distilled water and the resulting solution was saturated with hydrogen sulphide. 1 cm3 of stock vanadium (V) solution (prepared from V 2O5 as detailed above) was then added; this produced a deep purple solution containing 1 mol VS 43- m"3. If this solution was allowed to contact air it turned orange initially and after a longer time became colourless, as elemental sulphur was precipitated. A solution for electrochemical studies, initially containing 0.2 mol VS 43- m"3, was prepared by diluting the above solution five fold with an oxygen-free carbonate buffer.

5.2 Vanadium Voltammetry: Results and Discussion A voltammogram of a 1 mol V(V) m "3 is shown in Fig. 5.2. A sharp reduction peak was observed on the negative going scan at 0.3 V vs. SHE. If the potential was maintained at 0.3 V and a new drop of mercury expelled an oxidation current was seen to flow for a short time. These peaks were peculiar to the mercury electrode and the charge under them corresponded to the passage of 1.1 C m"2. The value was independent of the sweep rate, the stirring rate and the vanadium (V) concentration (providing it was above 10 "3 mol m-3). This is consistent with the process responsible being the reduction of a monolayer of mercury (I) vanadate. The formation of a mercury (I) salt with vanadium anions has been reported [135], and used a means of determining the vanadium concentration in solution. In the concentration range 10 "4 to 10"2 mol m"3, less than a monolayer of mercury (I) vanadate is formed at a HMDE when it is held at an oxidising potential for 60 s [135]. The vanadium concentration determines the fraction of the surface that is covered, and hence the charge that is passed reducing this layer in a subsequent cathodic stripping potential scan. In this way cathodic stripping voltammetry can be used to determine the vanadium concentration in Vanadium Electrochemistry 85 solution. Calculations show that the close packing of mercury (I) ions results in a charge density of 2.9 C m-2, so it is likely that the monolayer coverage is determined by the packing of the larger vanadium (V) ions (e.g. HV 2O73-).

5 ■

^ -K H2 -10 -0-8 -0-6 -04 -02 6 02 Potential /V vs. SHE

Fig. 5.2 Voltammogram of Vanadium (V) in Borate Buffer at pH 9.2 First Scan, commenced at 0.35 V vs. SHE. 50 mV s"1.

Below 0.2 V vs. SHE, no further reduction was observed until a potential of -1.0 V vs. SHE was reached. This potential is considerably lower than the reversible potential required to reduce V(V) to V(IV) (-0.1 V vs. SHE), as can be seen from the E^-pH diagram for the vanadium-water system (Fig. 5.3). This diagram was produced using the computer program POURB, which was re-written in FORTRAN 77 from a listing provided by Froning et al [136]. The thermodynamic data, in the form of AGf° values, were taken from a recent review by Israel and Meites [30]. These values are shown in the Appendix.

A potential of -1.0 V vs. SHE at pH 9.3 is sufficient to produce vanadium (II) oxide. The peak current density at -1.2 V was about -2.8 A m -2 (see Fig. 5.2). This compares a value of -1.4 A m"2, which can be calculated for the peak current during a reversible one electron transfer (using equation 7.11 and assuming: x> = 0.05 V s"1, C0 = 1 mol m"3, D 0 = 5 x 10' 10 m2 s-1 and r = 3.96 x 10 -4 m). The fact that the observed reduction current was double the reversible one electron value suggests that the reaction may proceed to form vanadium (HI) oxide (Vj Oj) or V 3O5 (oxidation state 31/3) rather than vanadium (IV) ions. HV2O73" + 4 e" + 3 H2O —> V2O3 + 7 OH- (5.3) 3 HV20 73' + 10 e- + 8 H20 -> 2V 30 5 + 19 OH' (5.4) Vanadium Electrochemistry 86

Fig. 5.3 Eh-pH Diagram for the V-H20 System at 298 K. Activity of V species = 0.01.

However, the formation of solid phases often proceeds at lower current densities than those predicted by equation (7.11), and it may be that vanadium (II) oxide (VO) is formed: HV20 73- + 6 e" + 4 H20 -> 2 VO + 9 OH- (5.5) If the reduction products were soluble species, they would be dispersed away from the electrode surface as the solution was stirred and would not be available for re-oxidation on a subsequent positive-going scan. Therefore, stirring the solution would have the effect of suppressing the re-oxidation peak at 0.05 V vs. SHE. In fact, stirring did not suppress this peak, which implied that the reduction product was a solid film which was adsorbed on to the electrode surface.

The reduction of water to form hydrogen has an extremely high overpotential on mercury. However, the presence of the vanadium oxide phase facilitated hydrogen evolution, as can be seen from Fig. 5.2; this is consistent with the known catalytic activity of vanadium [ 120]. Vanadium Electrochemistry 87

At higher V(V) concentrations, similar results were obtained. A voltammogram recorded at a concentration of 10 mol m “3 is shown in Fig. 5.4 (this was taken at 40 °C, but the increased temperature did not substantially affect the voltammogram).

Fig. 5.4 Voltammogram of Vanadium (V) in Carbonate Buffer at HMDE. 1st. Scan, 50 mV s"1, pH 9.3, scan commenced 0.33 V vs. SHE. [V(V)] = 10 mol m-3, T = 40 °C.

The peak reduction current was shifted to a less negative potential than in Fig. 5.2, to -0.9 V vs. SHE. A small reduction current was also observed at -0.17 V vs. SHE, a potential which is accessible using H 2S as a reducing agent. The reversible potential for the V(V)/V(TV) couple according to equation (5.6) at this pH is -0.10 V vs. SHE. Therefore, it is possible that reduction of vanadium (V) to (IV) may be responsible for this peak, producing V jgC ^12- or V 4092" ions, as shown in equations (5.6) and (5.7). As yet, no thermodynamic data has been published for the V 18O4212" moiety, so it does not appear on the E^-pH diagram shown in Fig. 5.3. 2 HV2073' + 4e- + 3H20 V4O92- + 8 OH“ (5.6) 9 HV20 73" + 18 e- + 12H20 V180 4212' + 33 OH“ (5.7) From the stoichiometry of both of the above equations it can be seen that neither reduction is likely to proceed in a single step, since in both cases a large structural rearrangement is required. This explains why the reduction current was less than an order of magnitude lower than that predicted for a reversible one electron transfer.

At gold and platinum electrodes, a smaller potential range was available due to the low overpotential required for hydrogen evolution. Fig. 5.5 shows a voltammogram recorded on a gold flag electrode in a carbonate buffer solution (pH 9.3) containing Vanadium Electrochemistry 88

10 mol V(V) m-3. The peaks at 0.4 V vs. SHE on the positive going scan, and 0.3 V on the negative going scan were due to gold oxide formation and reduction, respectively. Using vanadium concentrations of 10 mol m ~3 and above, a small reduction current (approximately 1/10 of the magnitude of a reversible one electron reduction), was observed at a potential of -0.55 V vs. SHE. This current may have been due to the reduction of vanadium (V) to (IV) in solution. The re-oxidation peak at 0.02 V vs. SHE was observed only when a potential of -0.7 V vs. SHE was exceeded on the negative going scan, which suggested than a film of reduced vanadium oxide was again formed at these lower potentials.

Fig. 5.5 Cyclic Voltammogram of Vanadium (V) on a Gold Electrode. 1st. Scan, 50 mV s-1, pH 9.3, scan commenced 0.245 V vs. SHE. [V(V)j = 10 mol nr3, T = 19 °C. Vanadium Electrochemistry 89

Direct evidence that reduction of vanadium (V) results in the production of a solid film was provided by voltammetry using a vitreous carbon disc electrode. A typical cyclic voltammogram is shown in Fig. 5.6:

Fig. 5.6 Cyclic Voltammogram of V(V) on a Vitreous Carbon Electrode. 1st. Scan, 50 mV s"1, pH 9.3, [V(V)] = 10 mol n r3, T = 20 °C.

This showed a pattern of vanadium (V) reduction and re-oxidation similar to that observed in Fig. 5.4 and Fig. 5.5. Large reduction currents were observed only at highly negative potentials (-1.0 V vs. SHE in Fig 5.6) and there was an extremely large potential separation between the reduction and re-oxidation peaks. Repeated potential scans resulted in decreasing current densities. An inspection of the electrode after such scans revealed that it had become coated with an iridescent layer. Further experiments showed that the thickness of this layer could be increased by holding the electrode at a potential of -1.0 V vs. SHE. The vanadium oxide layers appeared blue, green or purple depending on their thicknesses; this behaviour is characteristic of the optical interference patterns produced by thin layers, and can only occur when the layer thickness is at least quarter of the wavelength of the incident light (i.e. > 0.1 |im). Vanadium Electrochemistry 90

Voltammetry of a solution containing 1 mol V(TV) m ~3 on a gold disc electrode revealed that no reduction currents could be detected above the background currents that were seen in the carbonate buffer, in the potential range -0.6 V to +0.3 V vs. SHE. Since the reversible potential for V(V)/(IV) at pH 9.3 is about -0.10 V vs. SHE, based on equation (5.6), an anodic limit of 0.3 V represents an oxidising overpotential of 400 mV.

The above evidence demonstrates that the reduction of V(V) to V(IV) and the oxidation of V(IV), are irreversible processes at a variety of electrode surfaces. If the vanadium (V) phase is HV 2073" and the vanadium (IV) phase is V 18O4212" (see section 4.1 and 4.2) then it is not surprising that V(V) reduction is slow, since the formation of V 1g 0 4212“ requires a considerable structural rearrangement. The overpotential for this reduction is so high that the potential has to be lowered to values where other reduction reactions can occur, producing oxide phases such as V 3O5, V2O3 and VO.

This suggests that the reduction of V(V) by hydrogen sulphide in the Stretford Process proceeds via a specific chemical interaction between the two species. The thiovanadate complexes (see section 4.9) are well known and are likely reaction intermediates. The complex VS 43- was prepared, and the cyclic voltammogram of this species was recorded (Fig. 5.7).

Fig. 5.7 Cyclic Voltammogram of VS43-, HS" on a Gold Disc. 100 mV s'1, pH 9.8, [VS43"] = 0.2 mol n r3, [HS“] = 0.36 kmol n r3. Vanadium Electrochemistry 91

As can be seen from Fig. 5.7, oxidation and reduction currents were observed with a large peak separation. This voltammogram is very similar to those observed on gold electrodes in hydrosulphide solution (see section 3.3). The saturation of a solution of ammonia with hydrogen sulphide results in the production of ammonium hydrosulphide: NH3 + H2S NH4+ + HS" (5.8) The 10 % aqueous ammonia solution, as used in the preparation of the VS 43" complex (section 5.1), contained 1.8 kmol NH 3 m'3. When this was saturated with hydrogen sulphide, a solution containing 1.8 kmol HS' m '3 was formed. This is about 1000 times greater than the VS43" concentration, and explains why the voltammogram is essentially that of a hydrosulphide solution. If the hydrosulphide concentration was lowered, the complex decomposed. Thus, the redox behaviour of VS 43' was masked by the large background currents due to the presence of HS' ions.

5.3 Summary The reduction of vanadium (V) was found to be irreversible on a variety of electrode surfaces, and led to the formation of solid oxide films (V 3O5, V20 3 and VO) rather than to V(IV) solution species. Irreversible behaviour is commonly observed when a large structural rearrangement is necessitated as the reactant is reduced. In the present case, HV2073" is the probable V(V) species and V 1g 04212'is the likely V(IV) species; it is clear that such a rearrangement will be required.

The fact that vanadium (V) is an effective oxidising agent for hydrogen sulphide in the Stretford Process suggests that there is some specific chemical interaction between them that facilitates V(V) reduction, such as the formation of thiovanadate complexes. An attempt was made to investigate the redox chemistry of the thiovanadate ion VS 43', using cyclic voltammetry, but large background currents due to the oxidation of HS' ions obscured any currents that might have been due to the reduction of VS 43' ions. Anthraquinone Review 92

6. Review of Anthraquinone Redox Chemistry Anthraquinones contain two carbonyl groups on an anthracene backbone. Fig. 6.1 shows the structure and nomenclature of 9,10-anthraquinone.

2 3

Fig. 6.1 9,10-Anthraquinone. In the following discussion the 9,10- prefix should be assumed.

6.1 Anthraquinone Reduction Each of the two carbonyl groups in the anthraquinone can be reduced to a hydroxy group. This reduction can be regarded as electron transfer followed by protonation. If only one carbonyl group is reduced the product is a semiquinol, if both are reduced a quinol is produced. Reduction to the quinol is shown in equation (6.1).

0 ^ 0 + 2 H+ + 2 e (6.1) o CM

This equation can be written in an abbreviated form: AQ + 2 H+ + 2 e- —^ AQH2 (6.2) AQ and AQH 2 represent the anthraquinone and anthraquinol respectively. This reduction could proceed through any one of seven intermediate species. Fig. 6.2 shows the possible reaction pathways: AQH22+ <-> AQH + <-> AQ H+ H+ T le - 1U e- Tvl e-

a q h 2.+ <-> AQH- AQ: H+ H+

t i e - U e - T i e- H+ H+ a q h 2 <-> AQH- <-> AQ2-

Fig. 6.2 Intermediates in the Reduction of Anthraquinones [137]. Anthraquinone Review 93

In acidic aqueous solution, Bailey and Ritchie [138] studied the reduction of a variety quinones, and found that this proceeded to invariably produce the corresponding quinol. Quershi [139] studied the electrochemical reduction of 18 hydroxy- anthraquinones derivatives and found a two electron reduction in all cases.

This reduction occurs reversibly at the dropping mercury electrode (DME), and the polarographic studies up to 1974 were reviewed by Chambers [137]. Since the reaction occurs reversibly, and the diffusion coefficents of the quinone and quinol forms are similar, the half wave potentials (E ^ ) are good approximations to the formal standard potentials (Eo').

Heyrovski and Kuta [140] noted that the electrochemical reduction of an anthraquinone is dependent on the stability of the corresponding semiquinones; this stability can be measured by the value of the semiquinone formation constant Ksq, defined by: AQ + AQ2' 2 AQ*“ (6.3) [AQ-12 K [AQHAQ2-] Polarographic and voltammetric results are dependent on the kinetics of the above reaction, as well as the value of its equilibrium constant. Only if the equilibrium is established rapidly,compared to the time taken to complete a potential scan, will the value of Ksq affect the polarographic reduction wave. Heyrovski and Kuta [140] found that if Ksq« 1 a single polarographic wave corresponding to a two electron reduction was observed; if K sq» 16 then two separate waves were observed, corresponding to two consecutive one electron reductions. However, when Ksq had a value between 1 and 16, a single wave was observed with a slope corresponding to anything between 2/3 and 2 electrons per molecule.

This explains why, in many instances, single reduction waves can be observed, indicating electron transfer numbers close to one while coulometry always results in a value close to 2. Exactly this kind of behaviour was observed in the present study when disodium 2,7-anthraquinone disulphonate (Na2AQ27DS) was reduced. The reduction of anthraquinones may or may not be accompanied by the uptake of protons, depending on the first and second acidity constants of the corresponding anthraquinols. From the pH dependence of the reduction potential the number of protons consumed can be determined.

Savenko and co-workers [141,142] noted that the polarography of anthraquinone- 1,5-disulphonic acid was affected by adsorption of the oxidised and reduced forms on a mercury electrode, and this resulted in the inhibition of the reduction reaction. Anthraquinone Review 94

6.1.1 Substituent Effects Zuman [143] reviewed the effect of substituent groups on the Ej/2 potentials of quinone/quinol couples and found that substituent groups increased the reduction potentials according to equation (6.4): AE1/2 = C5 (6.4) where: S = Log fKa (Subst. Benzoic acid)1 Ka (Benzoic acid) AEj/2 = the difference in the half wave potentials between the substituted and unsubstituted quinones. C = proportionality constant (for each group of quinones) The term S , the total polar substituent constant, is based on the ratio of the acidity constants of substituted and unsubstituted benzoic acids. Electron-withdrawing substituents on the benzene ring help to delocalise the negative charge on the benzoate anion, through polar and resonance effects. This same charge stabilisation occurs in substituted AQH*" and AQ2- anions. Thus, upon substitution, the acid-base equilibria of the quinol will be altered; this will decrease the concentration of free quinol, and so increase the reduction potential. Therefore, it would be expected that the nature and position of the substituent groups on an anthraquinone affect the molecule's ability to become reduced.

It is believed that the anthraquinone salts act as oxidation catalysts and regenerate the V(V) species in solution in the Stretford Process. Reduced quinols can be re-oxidised by bubbling air through a solution containing them. Randell and Phillips [144] investigated the effectiveness of various substituted anthraquinones in catalysing the oxidation of Stretford solutions, which had been reduced previously by HS- ions. Their results appear to indicate that at pH 9 many substituted anthraquinones show a catalytic ability as good as, or superior to, that achieved by AQ27DS.

Fig. 6.3 Anthraquinone 2,7-disuIphonate (AQ27DS). Anthraquinone Review 95

Randell and Phillip measured the time taken for the oxygen concentration (initially zero) to rise to 20 % and 80 % of saturation. Using AQ27DS these times were 9 and 16 minutes respectively. Simultaneously, a platinum electrode measured the solution potential, which rose from -0.188 V to +0.045 V (versus SHE). Two materials in particular appeared to enable the oxygen content to rise in about half the time taken using AQ27DS: a mixture of tetra-sodium disulphomethyl AQ2,6 and 2,7- disulphonamides; and tetrasodium disulphomethyl AQ 1,5-disulphonamide:

0

Na0oS-CHo-N-S0p J J Na c

Fig.6.4 Na4 NN'-disuIphomethylanthraquinone-2,6-disulphonamide.

No explanation of their increased catalytic activity was offered, except to state that the compounds showed a greater solubility than AQ27DS.

6.1.2 Photo-reduction The photochemistry of anthraquinones has been widely studied. Anthraquinones have been suggested as photocatalysts for solar energy storage and for the splitting of water [145]. In alcoholic solution, photo-reduction occurs and anthraquinols are the main products, but in water the photolysis becomes more complicated and reduced and hydroxylated products are formed (mainly a-hydroxy anthraquinone sulphonates).

Moore [146] studied the UV and Raman spectra of the species generated by the irradiation of 2,6 anthraquinone disulphonate using laser light at 351 nm. Triplet Tj 2,6 anthraquinone disulphonate (3n7t*) was initially produced; in the presence of reducing agents such as sodium nitrite the triplet state was reduced to form the radical anion: hv NaNC>2 AQ26DS -» 3AQ26DS -» AQ26DS-' (6.5)

The radical anion was extremely long lived in the absence of oxygen, and showed UV absorbances at 400 and 510 nm [146,147]; the protonated form only showed the absorbance at 400 nm. Anthraquinone Review 96

When oxygen was present, the radical was quickly quenched, a process which resulted in the production of superoxide ions: AQ26DS-" + 0 2 -> AQ26DS + 0 2 ~ (6.6) The rate constant for the above reaction was found to be 9 x 10 8 M' 1 s_1; calculations show that a solution containing 1 mol m_3 of AQ26DS-" would completely deoxygenate an oxygen-saturated aqueous solution in less than 1 s. The superoxide ions that are produced are powerful oxidising agents and will react with water to produce hydrogen peroxide [148]. 2 0 2 - + H20 -» 0 2 + H 02- + OH- (6-7) H 02- + H20 -> H20 2 + OH- (6-8) The AQ26DS •" radical anion can also be attacked by hydroxyl radicals to form hydroxyl substituted anthraquinones.

Kano and Matsuo [147] showed that the AQ26DS-" radical anions could be stabilised by adding surfactants that produced micelles; ions that were bound to these micelles were stable for several weeks, even in aerated solutions.

6.2 Anthraquinones in the Production of Hydrogen Peroxide The industrial preparation of hydrogen peroxide relies on the hydrogenation of anthraquinone derivatives in non aqueous media, and the subsequent re-oxidation with oxygen to produce hydrogen peroxide [149]: AQ + H2 -» AQH2 (6.9) catalyst

AQH2 + O2 —> AQ + H2O2 (6.10) An alkyl substituted anthraquinone (such as 2-ethyl anthraquinone) is used, dissolved in a solvent of methyl cyclohexyl acetate or alkyl benzene [150]. The reoxidised solution is contacted with distilled water, and the hydrogen peroxide partitions itself into the aqueous phase.

Keita and Nadjo [151] showed that analogous reactions could occur in aqueous solution; they reduced the water-soluble sodium salt of 2,6 anthraquinone disulphonate (Na2AQ26DS) electrochemically, and then re-oxidised it with oxygen to produce hydrogen peroxide with 100% efficiency. They suggested that this method could be used to produce relatively dilute hydrogen peroxide for immediate local use.

Therefore, it is feasible that, on re-oxidation, the reduced AQ27DS in the Stretford Process solutions generates hydrogen peroxide in-situ. Hydrogen peroxide in alkaline solution is a powerful oxidising agent which is capable of oxidising sulphide solutions and re-oxidising vanadium (IV) solutions. Anthraquinone Electrochemistry 97

7. Redox chemistry of anthraquinone 2,7-disulphonate The anthraquinone disulphonate (AQDS) used industrially in the Stretford Process is an isomer mix, but the most active isomers are believed to be the 2,7 and 1,5 disulphonates, of which AQ27DS is the most active. The redox chemistry was investigated using cyclic voltammetry, at stationary and rotating disc electrodes, and controlled potential coulometry was conducted. The reduction products and intermediates were analysed by UV-Visible spectrophotometry and e s r spectroscopy.

7.1 Purification of 2,7 anthraquinone disulphonate A 15 g sample of the crude di-sodium 2,7 anthraquinone disulphonate (L.B. Holliday and Co. Ltd., Huddersfield, England) was dissolved in 30 cm3 of distilled water, and this solution was placed on top of a column packed with alumina. The column was then eluted with distilled water, and a mobile pale-yellow band collected; an orange band remained adsorbed at the top of the column.

The AQ27DS was recrystallised from an 80:20 acetoneiwater mix using the following procedure: 200 cm3 of the above solution was added to 800 cm3 of boiling propanone (acetone), the solution was reboiled and then filtered hot. It was then cooled in an ice bath and the pale-yellow crystals were recovered by filtration. The solid was dried overnight in an oven at 110 °C; a yield of 7.9 g was obtained (52 %).

7.1.1 Analysis of the purified 2,7 anthraquinone disulphonate It has been shown [152] that liquid chromatography can separate the isomers of anthraquinone sulphonates. Using equipment at the British Gas London Research Station, the solid was analysed using High Pressure Liquid Chromatography (HPLC).

A sample of the pure 2,7 isomer was kindly provided by Dr. M. Bruce (Department of Chemistry, Manchester University), this had been prepared by a regio-selective synthesis route and was over 99% pure. Using this as a standard, it was shown that the material which had been purified as above contained 99.1 % AQ27DS.

The 13C NMR spectrum was also recorded, and showed six distinct carbon resonances; this is consistent with the structure of 2,7 anthraquinone disulphonate. Unpurified material showed resonances which could be attributed to the presence of other isomers.

Industrial AQDS contains a broad range of isomers. The material that is used currently in operating Stretford Plants (Elvada) contains only 22 % of the 2,7 isomer; with the 2,6, 1,5, 1,6, 1,7, and 1,8 isomers all present in significant quantities. It is believed that the 2,7 and 1,5 isomers are the most effective catalysts. Anthraquinone Electrochemistry 98

7.2 Experimental: Voltammetry Solutions of AQ27DS (in the range 1-5 mol m~3) were prepared by dissolving the appropriate mass of the sodium salt in conducting carbonate buffer (0.059 kmol Na2C 0 3 m-3, 0.223 Kmol NaHC03 n r 3, 0.10 Kmol Na2S 0 4 n r 3: pH 9.3). All working solutions were freshly prepared on the day of the experiment and were nitrogenated for two hours before use with white spot grade nitrogen (BOC pic).

Cyclic voltammograms were recorded using a conventional electrochemical cell design and a saturated calomel reference electrode (SCE electrode), as shown in Fig. 7.1 below. Either a hanging mercury drop electrode (HMDE), a gold flag or a platinum foil was used as the working electrode.

Counter Electrode

Fig. 7.1 Electrochemical Cell Design for Voltammetry Experiments.

A Hi-Tek PPR1 waveform generator provided control potentials for the Thomson MP81 potentiostat, and were also applied to the x-input of a JJ PL4 chart recorder; the currents were passed through a suitable standard resistor and the resulting voltage was applied directly to the y-input of the chart recorder. At sweep rates above 100 mV s_1 the voltammograms were recorded on a Nicolet Explorer I oscilloscope. Anthraquinone Electrochemistry 99

7.3 Experimental: Exhaustive Electrolysis In order to determine the number of electrons involved in the reduction of AQ27DS and to provide a supply of the reduced compound, exhaustive electrolysis was performed, using the apparatus shown in Fig. 7.2. This incorporated a cation exchange membrane (Nafion 425, DuPont) which prevented the diffusion of the reduced species to the anode where they would otherwise have been re-oxidised.

Reference electrode

Counter electrode Test solution Ion exchange membrane Mercury working electrode

Fig. 7.2 Exhaustive Electrolysis Apparatus. Anthraquinone Electrochemistry 100

The AQ27DS solution was reduced at a stirred mercury pool electrode, using a catholyte containing 3.54 x 10"4 moles dissolved in 35 cm3 of carbonate buffer (pH 9.3). The anolyte contained 1.27 x 10“2 moles of potassium hexacyano iron(II) solution (K 4Fe(CN)6) dissolved in the same buffer; using this solution meant that hexacyano iron(III) was formed at the platinised titanium mesh anode. Preliminary experiments had shown that if oxygen was allowed to be evolved at the anode, it could diffuse through the ion exchange membrane and chemically re-oxidise the reduced solution in the catholyte compartment. Both compartments were nitrogenated with Zero Grade nitrogen (BOC pic) prior to electrolysis, and a nitrogen atmosphere was maintained above the working solution throughout the reduction.

A Luggin probe placed close to the mercury surface was connected to a saturated calomel electrode (EIL); the liquid film around a closed ground glass joint provided an electrical connection whilst minimising diffusion from the reference electrode compartment into the working solution. The working electrode potential was controlled using a Solartron 1286 Electrochemical Interface, and the current was passed through an internal resistor, the resulting voltage was then fed to a Hi-Tek DIBS digital integrator.

7.3.1 Calculations: Exhaustive Electrolysis If it is assumed that the overpotential for reduction is sufficient to achieve complete reduction of the test solution, and there are no competing reactions, the integrated charge would be expected to rise asymptotically with time towards a value of znF C. (z = number of electrons transferred, n = number of moles of reactant present and F = Faradays constant).

Furthermore, if the electron transfer at the electrode is extremely rapid the current will be limited by the mass transport of reactant to the electrode : it 00 Ct (7.1) where Ct = concentration of reactant at time t The constant of proportionality will depend on the number of electrons transferred (z), the electrode area (A), the thickness of the Nemst diffusion layer (8) and the diffusion coefficient of the reactant (D0) according to the equation: it = zFD0ACt (7.2) 5 Anthraquinone Electrochemistry 101

If the system is closed, and of volume V, then the concentration of oxidised species at any time is given by: Ct = C0(l - x) where x is the fractional conversion and will be given by: x = Charge passed _ f i 3t Total Charge Required zFC0V .-. Ct = C0 - li2t- (7.3) zFV Substituting for Ct in equation 7.2 gives: it = zFD0AC0 - D0A f i 3t (7-4) 5 V5 Since the initial current, it=o, is given by:

it=o = zFDqACq (7.2) 6 Equation (7.4) becomes: it = it=o - D0A J i 3t (7.5) V5 Differentiating with respect to t:

dit = - DqA it dt V5

Rearranging, l d i t = - DqA dt it V5 and integrating gives: In it = lnit =0 - ^0At ^7'5^ 5V Thus a plot of In it vs. t would be expected to be a straight line with an intercept of In (it=o) and a gradient of -D 0A/5V.

7.3.2 Calibration of Exhaustive Electrolysis Apparatus The apparatus was calibrated using the reduction of potassium hexacyano iron (III), (potassium ferricyanide), which is known to undergo a reversible one electron reduction. A potential of -0.2 V vs. SHE (-0.442 V vs. SCE) was applied to the electrode, which was in contact with a catholyte solution containing 4.9 x 10 "4 moles of K 3Fe(CN)6. Care was taken not to allow the hexacyano iron (III) to contact the mercury cathode before potential control was established, since the solution is capable of oxidising the mercury surface: 2 Fe(CN)63- + 2 Hg + 2 OH’ -» Hg20 + H20 + 2 Fe(CN)64- (7.6) Anthraquinone Electrochemistry 102

Electrolysis was allowed to proceed for 2 x 104 seconds, after which time the reduction current had fallen from 28 mA to 22 pA. The charge passed after this time was 45.15 C; a value in reasonable agreement with the theoretical value of 46.98 C expected for the one electron reduction of the K 3Fe(CN)g.

A plot of log it vs. t is shown below:

Fig. 7.3 Plot of Log it vs t during the reduction of Fe(CN)<53_.

The slope of this graph gives a value of -DoA/2.303SV; the solution volume (V) and the electrode area (A) were measured directly, and a value for D0 of 1.02 x 10“9 m2s_1 was taken from the literature [153,154], and corrected for the viscosity (r|) of the working solution according to the Stokes-Einstein relationship (D0 = K^T / 67tqa). These values enabled the mean thickness of the Nemst diffusion layer (8) to be calculated to be 5.53 Jim (under the particular stirring conditions employed). Anthraquinone Electrochemistry 103

7.4 Voltammetry: Results and Discussion A typical voltammogram for AQ27DS in aqueous alkaline solution is shown in Fig. 7.4. The reduction was found to comply with many of the requirements of a reversible electrode reaction [155]: the peak separation was independent of the voltage sweep rate, the ratio of cathodic to anodic peak heights was one - independent of the voltage sweep rate - and the peak reduction current was found to be directly proportional to the square root of the sweep rate.

Fig. 7.4 Cyclic Voltammogram of AQ27DS. [AQ27DS] = 1 mol n r 3, Sweep Rate 5 mVs"1, pH 9.3.

The reduction was found to occur at a half wave potential of -0.25 V vs. SHE at mercury, gold and platinum electrodes. Since the diffusion coefficients of the oxidised (quinone) and reduced (quinol) forms are likely to be equal, the half wave potential provides an estimate of the formal standard potential at this pH. This potential is slightly greater than that required to oxidise HS" to elemental sulphur (E0' = -0.28 V when [HS-] = 10 mol m“3).

Only a single reduction peak was observed, even though potentials down to -1.4 V vs. SHE were accessible using a mercury electrode, before hydrogen was produced. This behaviour suggested that the reduction was proceeding directly to form the quinol in a two electron step. Anthraquinone Electrochemistry 104

The current at a hanging mercury drop electrode can be split into components due to planar and radial diffusion of the reactant to the electrode surface. It can be shown [155] that the peak current at a planar electrode is given by: ipl = 0.4463 zFAC0 ( zF f 2 o 1/2 D 0 1*2 (7.7) RT

where v = voltage sweep rate / Vs"1 and other symbols are as defined previously. At 20°C in aqueous solution this simplifies to: ipl = (2.71 x 105) z3/2ACod1/2 Dq1/2 (7.8) For a spherical electrode of radius r m, an extra term due to radial diffusion must be added: isp = ipl zFADqCq (})(Gt) (7.9) r (j)(at) is a dimensionless constant which is dependent on the applied overpotential. At the peak current obtained during a reversible reduction its value is 0.7516 [155]. Substituting this value into equation (7.9) gives: isp = ipl + (7-25 x IO^zADqCq (7.10) r Subtituting in for ipi*.

isp = (2.71 x IO ^z^ A C qD ^ D o1/2 + (7.25 x 104)zACoJ2o (7.11) r Equation (7.11) constitutes a quadratic equation in D01/2, which can be solved at a given peak current and sweep rate providing z, A and r are known. Following the methods described previously (section 5.1) the drop area and radius were calculated to be 1.97 x 10"6 m2 and 3.96 x 10"4 m respectively. From the peak current densities shown in Fig. 7.4, and assuming that z = 2, the value of D0 was found to be 3.74 x 10"10 m2 s"1.

The above calculations assume that the reduction occurs in a two electron process. The peak separation was found to be about 40 mV, in between that expected for one electron and two electron processes (59 and 29.5 mV respectively). This may be due to the two electron quinol product being in equilibrium with the semiquinone, as discussed in section 6.3. Anthraquinone Electrochemistry 105

Richardson and Taube [156] extended the theory first proposed by Polcyn and Shain [157] which relates the observed peak separation (AEp) to the comproportionation constant (Kg), where: Kc = [AQ27DS-~]2 (7.12 ) [AQ27DS] [AQ27DS2-] If the first electron is transferred at a standard potential of E^j, and the second at E ^ , then they showed that the conproportionation constant was equal to: (7.13) RT Their analysis assumes that both charge transfers are reversible, that the reaction rate is sufficient to maintain Nemstian concentrations at the electrode surface and that the reactant diffuses linearly towards the electrode surface. In practice the assumption of linear diffusion applies well to planar electrodes, and even the hanging mercury drop electrode used shows only about 10 % deviation due to it being spherical (see equation (7.11) above).

When (E°i - E02) is greater than 120 mV, two reduction peaks can be seen and the E° values estimated directly from the half wave potentials; however, in the present case the peaks are superimposed so that only a single reduction peak can be seen. Richardson and Taube [156] produced a table of values and a working curve to enable the value of (E^i - E°2) to be estimated from the peak separation between the negative and positive going scans (AEp). The observed value of was about 40 mV, which corresponds to (E°i - E ^ ) = 0. Substituting this value into equation (7.13) gives Kc ~ 1, although the error in AEp is such that K^. could lie in the range 0.2 to 4. Anthraquinone Electrochemistry 106

Cyclic voltammetry at a rotated gold disc electrode is shown in Fig. 7.5. Current crossovers can be seen, which may be due to adsorption of the quinol on the electrode surface.

Potential vs. SHE / V

Fig. 7.5 Cyclic voltammetry of AQ27DS at a rotated gold disc electrode. Scan rate = 20 mVs"1. pH = 9.23. C0 = 0.357 mol n r3.

The currents appear to be mass transport controlled, and would be expected to follow the Levich equation: ilim = 1.554 zFADo^co^n'^Co (7.14) where co = rotation rate / s_1 o = kinematic viscosity / m2 s-1 and other symbols are as defined previously. A plot of current vs. (rotation rate)1/2 would therefore be expected to be a straight line with a gradient of 1.554 zFAD02/3 o"1/6 C0. Experimental results are shown in Fig. 7.6 (overleaf). Anthraquinone Electrochemistry 107

Fig. 7.6 Plot of i vs. co1/2 for reduction of AQ27DS.

The gradient was found to be 2.07 x 10"5 A s1/2. Assuming two electron reduction enabled a value for the diffusion coefficient (D0) of 3.73 x 10“ *0 m2 s_1 to be calculated; this is in good agreement with the value reported previously. For comparison, Compton [158] found a value of 4.7 x 10“10 m2 s_1 for the similarly sized compound, 1,8-dihydroxyanthraquinone. Anthraquinone Electrochemistry 108

The pH dependence of the reduction potential was determined by conducting cyclic voltammetry at a hanging mercury drop electrode, after the pH had been adjusted by sparging the solution with carbon dioxide for a short time. The pH was monitored after each adjustment by withdrawing samples and measuring the pH with a Corning 150 pH meter. The pH could be lowered in this way from 9.3 to 7.1. A plot of the half wave potential vs. pH is shown in Fig. 7.7.

Fig. 7.7 Plot of Reduction Potential vs. pH for AQ27DS.

The slope of the graph was found to be -31.6 mV pH"1; it follows from the Nernst Equation that the reduction potential should decrease at a slope given by 59h/z, where h = the number of protons consumed in the reduction. Since z = 2, h must be equal to one; i.e. the reduction must proceed in a two electron, one proton process: AQ27DS + 2e- + H+ AQ27DSH" (7.15)

7.5 Exhaustive Electrolysis: Results and Discussion Electrolysis was performed at potentials of -0.382 V and -0.6 V vs. SHE; providing 132 and 350 mV overpotential respectively. Assuming Nemstian conditions apply, 132 mV is sufficient to ensure that the equilibrium concentration of the oxidised AQ27DS is reduced to less than 0.01 % of its initial value. At both these potentials identical results were obtained; the resulting plot of charge vs. time is shown in Fig. 7.8 (overleaf). Anthraquinone Electrochemistry 109

Fig. 7.8 Plot of Charge vs. Time During the Electrolysis of AQ27DS. Electrolysis potential = -0.6 V vs. SHE. pH = 9.3.

As expected, the charge rose to reach a maximum value which corresponded to complete, two electron reduction; experimental values ranged from 94 % to 107 % of the theoretical charge. When complete reduction had been achieved, the potential of the mercury pool electrode could be stepped to 0.0 V vs. SHE and the solution re-oxidised with the passage of 86 % of the cathodic charge.

The reduction current was found to decay logorithmically, as expected for a mass transport limited reaction (section 7.3.1). Using a value for the Nemst diffusion layer of 5.53 pm (section 7.3.2) enabled an estimate of the diffusion coefficient (D0) to be calculated.

The slope of the Log(i) vs. t plot (Fig. 7.9 overleaf) was found to be -1.91 x 10 ' 4 (correlation coefficient 0.998), this results in a calculated value of the diffusion coefficient of 8.9 x 10 ' 10 m2 s-1. This is only in moderate agreement with the values obtained from rotating disc experiments and hanging mercury drop electrode cyclic voltammetry (3.74 x 10" ^ ). However, the stirred mercury pool electrode does not provide a well defined hydrodynamic regime and diffusion coefficients calculated from such results are likely to be less accurate. Anthraquinone Electrochemistry 110

Fig. 7.9 Plot of Current vs. Time for Electrolysis of AQ27DS. Potential = -0.6 V vs. SHE. pH = 9.3. Anthraquinone Electrochemistry 111

7.6 UV-Visible Spectrophotometry: Experimental Observation of the electrode surface during cyclic voltammetry revealed that a deep red-brown colour was produced on the negative going scan; this colour was discharged on the return scan. This suggested that the production of the quinol could be followed spectrophotometrically,

An apparatus was assembled to continuously monitor the UV-Visible spectrum of a Na2AQ27DS solution throughout its reduction, by pumping the solution through a flow-through UV cell. The apparatus is shown in Fig. 7.10.

Fig. 7.10 Electrolysis with Linked UV-Visible Spectrophotometry.

The anolyte contained 0.5 kmol m-3 potassium hexacyano iron (II) (K 4Fe(CN)g) dissolved in carbonate buffer (pH 9.3) and the catholyte contained about 0.5 mol m-3 Na2AQ27DS dissolved in the same buffer. The two solutions were separated by a Nafion cation exchange membrane (DuPont) and were nitrogenated with oxygen-free (CP) grade nitrogen (BOC pic) before the reduction was commenced. Throughout the experiment a nitrogen atmosphere was maintained in the anolyte, catholyte and reference compartments.

A titanium working electrode of large surface was prepared by dissolving the oxide coating from a slotted titanium plate in hot 4 kmol m-3 sulphuric acid. The potential of Anthraquinone Electrochemistry 112 this electrode was then maintained at -0.78 V vs. SCE (-0.538 V vs. SHE). The reversible potential for hydrogen evolution at pH 9.3 was -0.638 V vs. SHE, and a preliminary voltammogram had shown that hydrogen evolution was not significant until a potential of -0.658 V were reached.

The potentials were controlled by a Solartron 1286 Electrochemical Interface, and the charge passed was monitored by a Hi-Tek DIBS digital integrator. Once the solution had been deoxygenated, 3.24 x 10"4 moles of Na2AQ27DS were added, to form a solution of concentration 1.392 mol m-3 . When the working potential was applied, an initial current of -3 mA flowed, which decreased as the AQ27DS was reduced, and eventually reached a "residual" value of -240 |iA. This was assumed to be due to oxygen diffusion into the apparatus, causing re-oxidation of the quinol formed.

Preliminary experiments had shown that diffusion of oxygen through plastic tubing and peristaltic pumps was a serious problem, and so PTFE-lined stainless steel tubing and an enclosed diaphram pump were used in the apparatus shown in Fig. 7.10. The reduced quinol was extremely oxygen-sensitive and the problem of oxygen diffusion was most apparent when the solution was essentially reduced. Data given by Esco (rubber) Ltd. showed that silicone rubber has a permeability of 3.16 x 10"4 moles of 0 2 m-2 s"1 (1 mm thickeness, AP accross membrane 1/5 atmosphere). Even the short lengths used to connect the inflexible metal tubing to the flow-through UV cell could, in theory, allow enough oxygen diffusion sufficient to sustain currents of -300 |iA. A correction was made for oxygen diffusion by subtracting the charge passed due to the experimentally observed "residual" current.

At regular charge intervals the UV-visible spectrum was recorded using a Hewlett Packard HP8451A diode array spectrophotometer, with the solution flowing through a Hellma 170.004Q quartz cell (path length 1 mm). The spectra were referenced against the carbonate buffer solution. 7 U-iil Setohtmty Rsls n Discussion and in shown Results are intervals charge C 15 Spectrophotometry: approximately at UV-Visible taken spectra UV-visible The .7 7 Fig. Fig. reduction. peaks new two and decreased AQ27DS, to due nm, 330 at peak that the seen be can It concentrations of AQ27DS, it was shown that the Beer-Lambert law (Equation 7.16) 7.16) (Equation law Beer-Lambert the that shown was it AQ27DS, of concentrations ocnrto, n ws on t b 403 m 460.3 be cm to found vs. was nm) and (330 absorbance concentration, of graph a from calculated was coefficient extinction The was followed, and so the absorbance at 330 nm could be used to monitor the AQ27DS AQ27DS the monitor to used be could nm 330 at absorbance theso and followed, was different containing solutions of nm 330 at absorbances the measuring By nm. 520 at shoulder broad a and nm 410 at peak sharp a proceeded: reduction the as appeared ABSORBANCE -1 8 opticalabsorbance = A = eC = A = opticalpathlength m / = / C extinctioncoefficient m / = mol 0 7.11. concentration mol / m = Spectra taken at 15 C Charge Intervals during AQ27DS Reduction. AQ27DS during ChargeIntervals C 15 at taken Spectra Concentration = 1.392 mol m-3. Path Length = 1mm. pH = 9.3 =pHm-3. 1mm.mol = Length Path 1.392 =Concentration -1 d 3). dm 0l (7.16) nhaunn ElectrochemistryAnthraquinone -3 2

mol Fig.7.11 -1

2 ml e463 n o-I units; non-SI in .(e=4,603 1 mol-

113 Anthraquinone Electrochemistry 114

A plot of the absorbance(330 nm) vs. charge (after correction for oxygen diffusion) during the reduction of AQ27DS is shown below:

Fig. 7.12 Absorbance(330 nm) vs Charge during AQ27DS Reduction Concentration = 1.392 mol n r3. Path Length = 1mm. pH = 9.3

The absorbance at 330 nm decreased linearly with charge passed as the AQ27DS was reduced (deviations were seen as the reduction neared completion, when the charge correction due to oxygen diffusion became very important). The charge taken to reduce the absorbance to half its initial value was 32.3 C, corresponding to 1.04 F mol-1 . This suggests that a two electron process is required to achieve complete reduction, a conclusion in agreement with earlier results.

7.7.1 Spectral Assignments The starting material, AQ27DS, shows absorbances at 330 nm (e = 460 m 2 mol-1) and 258 nm (s = 4450 m2 mol-1 ) in the UV spectral region. The reduction product at pH 9.3 is believed to be AQ27DSH", and so the absorbances at 410 nm (e = 980 m2 mol-1) and 277 nm (e = 10 4 m2 mol-1) are attributed to this species.

Work done by McQuillan [159] has shown that the AQ27DS*' radical anion can be produced by controlled potential electrolysis of AQ27DS at pH 13.2 in a solution containing 0.5 kmol m -3 tetraethyl ammonium hydroxide. In such a solution two reduction waves can be seen in a cyclic voltammogram, corresponding to two successive one electron reductions: AQ27DS + e--- > AQ27DS-” (7.17) AQ27DS-- + e----> AQ27DS2' (7.18) The tetraethyl ammonium hydroxide stabilises the radical anion, possibly by incorporating it within a micelle; Kano and Matsuo [147] found that micelles of Anthraquinone Electrochemistry 115 sodium laurate or sodium lauryl sulphate could stabilise the radical anion AQ26DS-". Alternatively, an ion-pairing interraction between the tetraethyl ammonium cation and the radical anion may be present.

By applying a potential sufficient to achieve only one electron reduction, McQuillan produced a solution containing AQ27DS-" as the major species, and therefore was able to measure its UY-Visible spectrum [159]. He found absorbances at 403 and 525 nm. It is likely that the shoulders at around 390 nm and 520 nm observed in Fig. 7.11 are due to AQ27DS-".

At more negative potentials, McQuillan produced the di-anion AQ27DS2" which absorbed at 454 and 540 nm. The pKa of AQ27DSH- is 10.8 [159], and so at pH 9.3 approximately 3 % o f the AQ27DSH- would be present as the di-anion. Therefore the shoulder observed at 450 nm in Fig. 7.11 is likely to be due to AQ27DS2-, present as a minor component. As the reduction proceeds, protons are consumed and, despite the buffering of the solution, the pH is likely to rise slightly. Calculations show that a change of only 0.3 of a pH unit would double the equilibrium concentration of the di­ anion. This explains why the shoulder at 450 nm became more pronounced as the reduction progressed.

The spectral assignments are summarised in Table 7.1:

Species ^max / nm 8 / m2 mol -1

AQ27DS 330 460 258 4450

AQ27DS-- 525 -660 403 -480

AQ27DS2" 540 -300 454 -1200

AQ27DSH- 410 -980 277 - 104

Table 7.1 UV-Visible Spectral Summary of AQ27DS Reduction Products Anthraquinone Electrochemistry 116

7.8 ESR Spectroscopy: Experimental The peak separation from voltammetry (section 7.4) and the UV-visible results (above) indicated that the radical anion, AQ27DS-", was produced during the reduction of AQ27DS, even though the major product was AQ27DSH". If the radical anion were present, then the solution would show a strong esr signal. The following experiment was designed to detect any esr signal produced during the reduction of the anthraquinone.

Solutions of 1 and 5 mol AQ27DS n r 3 were prepared by dissolving the appropriate mass of the sodium salt in conducting carbonate buffer (0.059 kmol Na 2C03 m-3, 0.223 kmol NaHC 03 m~3, 0.1 kmol Na 2S04 m“3: pH 9.3). The solutions were deoxygenated by passing purified nitrogen through the solution. The nitrogen was purified by passing it through a series of Dressel Bottles containing: reduced anthraquinone-2-sulphonate (AQ2S) to remove oxygen, water to remove spray, and finally drying agents to remove traces of moisture. The AQ2S was reduced by contacting an alkaline solution with a zinc/mercury amalgam.

The deoxygenated solution was loaded into the drive syringe of the electrochemical esr apparatus that is shown in Fig. 7.13.

- to waste r Platinum tube Tube cross-section counter electrode

esr davify LL___ Reference electrode

$&( Porous plug

Platinum half -cylinder working Solution electrode

Fig. 7.13 Electrochemical ESR Apparatus. Anthraquinone Electrochemistry 117

A stepper-motor driven syringe enabled the flow of solution through the apparatus to be carefully controlled. A potential of -0.358 V vs SHE (-0.6 V vs SCE) was applied to the platinum half-cylinder working electrode; the potential was controlled using a laboratory-built potentiostat incorporating a conventional operational amplifier circuit. The reduced solution flowed into the esr cavity of a Bruker ER200 TT esr spectrophotometer, through a 1 mm diameter quartz tube, and the esr spectrum was recorded using a field setting of 3395 Guass and microwave frequency of 9.54 GHz. The field was swept slowly over a 5 Guass range, using a 0.5 s integrating time constant. The integrated esr signal strength was proportional to the magnitude of the central resonance; the cavity response was calibrated by placing in it a MgO crystal containing a known number of spins (200 ppm Mn2+) and measuring the resulting esr signal.

7.9 Electrochemical ESR: Results and Discussion The flow-through tube electrode provides a well defined mass transport rate, and the mass transport limited current at a half-cylinder electrode should be given [158] by: iIim = 2.75 zF Xe2/3 D02/3 VfW C0 (7.19) where Xe = length of electrode / m (see Fig. 7.13) Vf = volumetric flow rate / m 3s_1 and other symbols are as previously defined. Thus the limiting current should be directly proportional to the third root of the flow rate; the slope of a graph of iiim vs. Vf 1/3 gives an estimate of the diffusion coefficient (D0), although it is acknowledged that rotating disc results (see section 7.4) are more accurate.

It was found that the above plot was linear, the value of the slope being 8.8 x 10-3 A (m3 s '1)"1/3; the value of D0 calculated from this was 1.0 x 10 -10 m2 s' 1 - somewhat lower than the value of 3.73 x 10_1° m2 s' 1 obtained using the rotating disc electrode.

Radical species produced at the tubular electrode are carried into the esr cavity, but the velocity profile across a section of tube is parabolic and not uniform, with the flow slowest near the walls (see Fig. 7.14).

Fig. 7.14 Flow Profile Across a Tube. Anthraquinone Electrochemistry 118

The radicals, produced at the tube edge, diffuse towards the centre of where the flow is fastest. Thus the number of radicals within the esr cavity is dependent on the flow velocity profile across the tube and the rate of radial diffusion towards the centre. This combination has been considered by Compton [158] who found that the resulting esr signal strength was given by equation (7.20). S = S0 (tt/4)2/3 j2/3 r2 inm V (7.20) zFD0l/3 Vf2/3 IK

where S = esr signal strength S0 = esr signal strength for one mole of spins within cavity. 1 = length of esr cavity / m r = radius of the tube / m IK and IK' are sensitivity factors, and are constant for a given geometry. Thus the normalised signal strength (S/ixim) should be directly proportional to V f2/3.

Such a plot is shown if Fig. 7.15:

Fig. 7.15 Normalised ESR signal (S/ium) vs. Vf~2/3. Anthraquinone Electrochemistry 119

As can be seen the resulting plot is not linear, showing a signal enhancement at low flow rates. This is consistent with a mechanism of radical production as follows: AQ27DS + 2e" + H+ -> AQ27DSH" (7.21) AQ27DSH- <-> AQ27DS2' + H+ (7.22) AQ27DS2" + AQ27DS <-> 2AQ27DS-- (7.23) The electrochemical reduction results in the formation of the protonated di-anion (7.21) which exists in equilibrium with the unprotonated form (7.22); this then reacts with the starting material to produce the radical anion (7.23).

If the flow rate is slow, the reduced material undergoes greater radial diffusion, towards a region of high AQ27DS concentration, which will shift the equilibrium (7.23) to the right, and result in an increased esr signal.

7.10 ESR Spectral Strucure Some difficulty was encountered in obtaining a fully resolved esr spectrum; a high concentration of AQ27DS-" increased the spin exchange owing to reaction (7.23), which had the effect of broadening the esr signals, and a low concentration meant that the spectrum was hard to resolve from the instrumental 'noise'.

Nevertheless, using a concentration of 5 mol AQ27DS m -3 enabled a partially resolved esr spectrum to be obtained. This was consistent with the radical structure shown below:

Fig. 7.16 Structure of AQ27DS*-

There are three groups of equivalent protons, labelled and H 3. This would be expected to produce 27 resonances, but spectral overlap and line broadening will reduce this number. Anthraquinone Electrochemistry 120

A spectrum was simulated using the program EPSIM77 assuming the following values: Spectral lineshape Lawrentzian Linewidth 0.19 Splitting Constants 0.21 Gauss (2 protons) 0.55 Gauss (2 protons) 1.025 Gauss (2 protons) The actual and simulated esr spectra are shown below:

Fig. 7.17 Actual and Simulated ESR Spectra of AQ27DS*"

7.11 Summary Anthraquinone 2,7 disulphonate (AQ27DS) was reduced in aqueous solution at pH 9.3 to give a deep red coloured air-sensitive solution. Cyclic voltammetry and exhaustive electrolysis indicated that the anthraquinone was reversibly reduced in a two electron, one proton process at a variety of electrode surfaces. From limiting current results at a rotating disc electrode the diffusion coefficient of AQ27DS was calculated to be 3.73 x 10 "10 m2 s"1. Anthraquinone Electrochemistry 121

UV-visible spectrophotometry confirmed that AQ27DSH" was the major reduced species, but also indicated that the di-anion (AQ27DS2-) and radical species AQ27DS-" were also present. A strong esr signal confirmed the presence of the radical and the spectral features were consistent with the chemical structure of AQ27DS-".

The esr signal strength variation with flow rate showed that the radical was not produced directly at the electrode in a one electron process, but was formed via a comproportionation reaction between the di-anion and the AQ27DS starting material. The peak separation from voltammetry enabled the comproportionation constant (Kc) to be estimated, and it was found to be in the range 0.2 to 4. Oxygen Reduction 122

8. Oxygen Reduction Oxygen and sulphur both lie in group VI of the periodic table and there are certain similarities in their chemistry: for example they both form compounds in the -II oxidation state and sulphur can often replace oxygen in its compounds (e.g. SO 42" and S20 32-).

However, oxygen is the first member of the group and many aspects of oxygen chemistry differ dramatically from those of sulphur The smaller size of oxygen means that the element is more electronegative than sulphur, and oxygen tends to form bonds with a high degree of ionic character. Thus, water is a highly polar molecule, which leads to its hydrogen-bonded liquid structure at room temperature, whilst hydrogen sulphide is non-polar and exists as a gas.

The fact that oxygen is restricted to eight electrons in its outer shell limits its maximum co-ordination number to four, and in practice it rarely exceeds two. Whereas sulphur, because of its d orbitals at available energy levels, is able to form compounds with higher co-ordination numbers. Commonly four co-ordinate compounds are formed (e.g. SO 42") and co-ordination number of up to six are possible (e.g. SFg). This expansion of the octet rule allows sulphur to form a range of compounds with a formal oxidation state greater than II (see section 2.1) whilst oxygen exhibits only the oxidation states -II, -I and zero.

The bond energy of the oxygen-oxygen double bond is more than three times that of the single bond (498 and 142 kj mol -1 respectively [160]), whilst for sulphur it is less than twice (425 and 226 kJ mol -1 [161,57]). This results in a tendency for sulphur, unlike oxygen, to form catenated molecules (e.g. the polysulphides and polythionates).

Oxygen can exist in two allotropes; ozone (O 3) and oxygen (O 2). Although oxygen has a bond order of two, it is paramagnetic and contains two unpaired electrons. This has been explained by molecular orbital bonding theory, which predicts that two tc* orbitals should be singly occupied. Because of this, oxygen has been termed a di-radical and many of the reactions of oxygen proceed via radical mechanisms (e.g. combustion).

By contrast ozone is diamagnetic. The molecule is bent with a central bond angle of 116 °. Ozone is prepared by the action of a silent electrical discharge in a stream of oxygen, or by the electrochemical oxidation of sulphuric acid at low temperatures. Ozone is an endothermic compound (AHf° = 142 kJ mol'1) and because of this thermodynamic instability it can decompose to form oxygen. However, this decomposition is slow in the absence of ultra violet light or transition metal catalysts, because of the high activation energy that is required. Like many sulphur compounds, ozone is an example of a metastable compound. Oxygen Reduction 123

Ozone is a more powerful oxidising agent than oxygen, as can be seen from a comparison of their standard reduction potentials: 0 3 + 2H+ + 2e- -> 0 2 + H20 E° = 2.07 V (8.1) 0 2 + 4H+ + 4e- -» 2H 20 E° = 1.23 V (8.2) In fact few oxidants are as effective in acid solution; ozone is capable of oxidising sulphide to form sulphate [83].

Oxygen itself should also be a good oxidant. Certainly at high temperatures it is extremely effective, being reduced to form water or carbon dioxide in the oxidation of fuels, for instance. However, in aqueous solution at room temperature, oxygen is often a much less effective oxidising agent than its standard electrode potential would indicate. There are two reasons for this; the low solubility of oxygen in aqueous solution and the kinetic inertness of the 02 molecule towards reduction.

Oxygen is only slightly soluble in water; a saturated solution in equilibrium with pure oxygen at one atmosphere pressure contains only 1.25 mol 0 2 m"3. Like other gases, the solubility of oxygen decreases as the temperature is increased. This is why increasing the temperature in an attempt to increase the rate of an oxidation can sometimes have the reverse effect. However, the diffusion coefficient (which can limit the transport of oxygen to a reacting surface) increases with temperature, so it follows that the product of the diffusion coefficient and the solubility must go through a maximum. This maximum is found at 60 °C in aqueous solution. Gold dissolution in cyanide solution (8.3) is limited by the transport of oxygen to the metal surface. 4Au + 0 2 + 8 CN- + 2H 20 -> 4Au(CN)2‘ + 4 OH- (8.3) The fastest rate of gold dissolution is found to occur at 60 °C.

The kinetic inertness of oxygen is due to the high strength of the oxygen-oxygen bond (498 kJ mol-1 ) and the need for four consecutive electron transfer steps. Oxygen reduction can sometimes proceed to form hydrogen peroxide (H 20 2) rather than water, since this does not require the cleavage of the oxygen-oxygen bond. Oxygen Reduction 124

8.1 The Oxygen / Water Couple Water represents the lowest oxidation state of oxygen and is produced on the complete reduction of molecular oxygen. Theoretically, this can be achieved by applying a potential lower than line (b) in Fig. 8.1:

Fig. 8.1 Eh-pH diagram of the 0 2/H20 System at 298 K [61].

However, complete reduction of gaseous oxygen (8.4) involves a four electron transfer, and is a highly irreversible process. 0 2 + 4H+ + 4 e “ -> 2H 20 E°= 1.23 V (8.4) The standard electrode potential for reaction (8.4) has proved difficult to achieve in practice; exchange current densities for the reaction on Pt and other noble metals are typically 10"10 - 10-11 A [62]. Side reactions, which would otherwise be considered slow, can compete with reaction (8.4) in determining the rest potential. Ordinary platinum electrodes in pure acid and in the presence of 1 atm. 0 2 usually achieve a potential of around 1.0 V vs. SHE. Oxygen Reduction 125

A general scheme for oxygen reduction for oxygen reduction has been reproduced in reviews by Tarasevich et al. [162] and Schiffrin [163]: I 1 Oj <-» (02)Sur ^ (C^ads ^ (^Q^ads “> H20 (8.5) u

(H2°2)sur H2°2 In this scheme 0 2, (0 2)sur, and (0 2)a(js correspond to molecular oxygen in the bulk solution, at the electrode surface, and in the adsorbed state, respectively. There exists two basic reaction pathways; the direct reduction to produce water, or a consecutive reaction pathway proceeding through hydrogen peroxide (which can also disproportionate chemically, to produce oxygen and water). At the reversible potential of equation (8.4), nearly all metal electrodes are covered in an oxide film, the nature of which is potential and time dependent. These films have a pronounced effect on the oxygen reduction kinetics, and make the observed reduction waves particularly complex to interpret. Vesovic et al. [164] recently pointed out that despite years of intensive study, the mechanism of oxygen reduction at many electrode surfaces has not yet been established.

Rotating ring-disc electrode studies have been made on the reduction of oxygen, and were recently reviewed by Tarasevich [162]. He pointed out that the method was incapable of yielding rate constants for all the reactions in scheme (8.5), but drew the following qualitative conclusions: i) On gold, mercury, pyrite (FeS2) and carbon electrodes the consecutive reaction pathway is operative, hydrogen peroxide is formed as an intermediate and can be desorbed slowly from the electrode surface. ii) On platinum, palladium and silver electrodes direct, four-electron reduction predominates. A simple explanation of this distinction between the two groups was suggested in the differing affinities of the materials for molecular oxygen and hydrogen peroxide; the first group possess a low affinity for these species, which is insufficient to break the oxygen-oxygen bond.

When oxygen reduction is carried out a rotated electrode, any hydrogen peroxide intermediate that is formed can be swept away from the electrode surface. This tends to occur at high rotation rates and low overpotentials. At pH 5, with a slow negative­ going potential sweep at a rotated pyrite (FeS2) electrode in an oxygen-saturated solution, Biegler et al. [165] observed two distinct reduction waves. They attributed the first wave to reaction ( 8.6) and the second wave to reaction (8.7). 0 2 + 2 H+ + 2 e" —> H20 2 (8.6) 02 + 4 H+ + 4 e- -> 2 H20 (8.7) Vesovic et al. [164] noted a similar change from a two to a four electron process at a gold electrode as the overpotential was increased. Oxygen Reduction 126

In fact, hydrogen peroxide can be produced by the electrochemical reduction of oxygen at porous carbon electrodes. This has been used as an industrial method of producing hydrogen peroxide, but has now been largely superceded by methods based on the hydrogenation and re-oxidation of anthraquinones (see section 6.2).

Behret et al [81] studied the oxygen-reduction activity of transition metal sulphides. They found that the sulphides of the metals cobalt, iron, and nickel showed the greatest activity. Since the sulphides of these metals are known to be catalytically active for sulphide oxidation (section 2.3.3) and oxygen reduction, it is likely that their known catalytic action on sulphide oxidation proceeds via a coupled electrochemical mechanism.

The fact that the four electron reduction of oxygen requires a high overpotential at most electrode surfaces has important technological implications. Only a limited number of metals are suitable for use in the construction of metal-air batteries, and efficient oxygen-consuming cathodes in fuel cell systems have remained reliant on expensive catalysts. As a consequence, the great potential of fuel cells for the efficient conversion of fuels to electricity remains largely untapped. Oxygen reduction has not yet replaced hydrogen evolution in electrolytic processes, such as those employed in the chlor-alkali industry, despite a possible saving of 0.8 V in the cell voltage if this practice could be adopted.

8.1.1 The Evolution of Oxygen The oxidation of water to produce oxygen is also irreversible: an overpotential of 0.6 to 0.8 V is required to produce oxygen from acid solution at a current density of 200 A m“ 2 on platinum or platinum / iridium anodes [166]. Even higher overpotentials are developed on the lead oxide anodes that are used industrially for electrowinning metals, and this voltage loss represents a considerable waste of electrical energy. On the other hand, the high overpotential is advantageous in aqueous batteries with positive electrodes which develop potentials greater than the reversible potential for oxygen evolution. The spontaneous generation of oxygen would rapidly discharge such batteries if the overpotential were low.

According to Tarasevich [162], the origin of the kinetic hindrance towards oxygen evolution lies in the nature of the electrode surface at the high potentials required. Relatively few metals are resistant to corrosion, and in those that are, this can usually be attributed to the formation of a passivating layer of metal oxide. Thick surface layers form a barrier towards electron transfers (although electrons are believed to be able to tunnel through the thin surface layer that forms on a platinum electrode in acidic solution). Despite years of research, the mechanism of oxygen evolution, even on platinum (the most intensively studied material), remains speculative. Oxygen Reduction 127

8.2 Hydrogen Peroxide Hydrogen peroxide is industrially produced by the reduction of an anthraquinone derivative with hydrogen, followed by re-oxidation with oxygen (see Section 6.2). It can also be prepared by the electrochemical oxidation of water, via a peroxodisulphuric acid intermediate. Sulphuric acid is oxidised at low temperatures and using a platinum anode; under these conditions it is oxidised to form peroxydisulphuric acid: 2H 2S 04 -> H2S20 8 + 2H+ + 2e- (8.8) This can be hydrolysed to form hydrogen peroxide (8.9), which can be removed by vacuum distillation. H2S20 8 + 2 H20 -> H20 2 + 2 H2S04 (8.9) Thus the overall reaction involves the oxidation of water: 2 H20 -> + 2H+ + 2e- (8.10) Hydrogen peroxide can also be prepared by the electrochemical reduction of oxygen (reaction 8.6) as was discussed above.

Hydrogen peroxide is also evolved when the solid peroxide salts react with water. These salts contain the 0 22' ion and are prepared by reacting group I or II metals with oxygen. Hydrogen peroxide is used as the starting material to form a range of organic peroxides, which are useful as oxidants or a source of free radicals (e.g. benzoyl peroxide).

Hydrogen peroxide has the structure shown in Fig. 8.2: 94°

Hx?97° 1.49 A < Q " “ H

Fig. 8.2 The Structure of Hydrogen Peroxide.

Pure hydrogen peroxide is a pale-blue viscous liquid, which possesses a structure containing a network of three-dimensional hydrogen bonds. However, it is not used as a solvent because of its oxidising nature and its ready decomposition.

Hydrogen peroxide is another example of a metastable compound; it is thermodynamically unstable, yet solutions can be stored for months without decomposition, because of the large activation energy required. If it is exposed to light (which can provide the large activation energy) or traces of transition metals (which provide a lower activation energy mechanism) then it decomposes rapidly. Oxygen Reduction 128

An Efo-pH diagram for the hydrogen peroxide / water system is shown in Fig. 8.3:

Fig. 8.3 Eh-pH Diagram for the H 20 2 / H20 System at 298 K [61].

Below lines (2) and (3) hydrogen peroxide can be reduced to form water: H20 2 + 2 H+ + 2 e" -> 2H20 (8.11) Above lines (4) and (5) it can be oxidised to form oxygen: H20 2 -> 0 2 + 2H+ + 2e- (8.12) Between these family of lines hydrogen peroxide is doubly unstable and can decompose according to reaction (8.13): 2H 20 2 -> 2H20 + 0 2 (8.13) Thus, if hydrogen peroxide contacts a metal surface having an electrode potential within this region, it will spontaneously decompose; this is an example of the electrochemical catalysis of a chemical reaction.

Since hydrogen peroxide can be reduced in the region below lines (2) and (3) in Fig. 8.3, it follows that it will act as an oxidant towards redox couples which have their solution potentials in this region. In this way hydrogen peroxide acts as a moderately powerful oxidising agent, both in acid and alkaline solution. Oxygen Reduction 129

Conversely, towards redox couples having their potentials above lines (4) and (5), hydrogen peroxide can act as a reductant; manganate (VII) solutions can be reduced to manganese (II), for example. Because of the slope of these lines with pH, it is possible for hydrogen peroxide to act as an oxidant at a low pH and a reductant at a higher pH.

The reduction of hydrogen peroxide is catalysed by the presence of transition metal ions in solution. Mo(VI), for example, forms a complex with hydrogen peroxide [167]: Mo0 42' + H20 2 Mo0 52' + H20 (8.14) This complex is more readily reduced than hydrogen peroxide: Mo052- + 2H+ + 2e- Mo042- + H20 (8.15) Many of the reactions of hydrogen peroxide can also proceed via free radical mechanisms.

8.3 Superoxides The action of oxygen on potassium, rubidium and cesium gives rise to yellow crystalline solids of the formula MO 2. They contain the superoxide ion (C>2~) which is an extremely powerful oxidising agent. In aqueous solution the superoxide ion will react with water to form hydrogen peroxide: 2 02" ■+■ 2 H20 —^ 0 2 + H20 2 (8.16) The superoxide ion is formed as the first intermediate during oxygen reduction, and its production in an adsorbed form was thought to be the rate-limiting step for oxygen reduction at a gold electrode in acidic solution [168].

8.4 Experimental A carbonate buffer of pH 9.3, containing 0.059 kmol Na 2CC>3 irf3, 0.223 kmol NaHC03 m-3 and 0.1 kmol Na 2S04 n r3, was prepared by dissolving the appropriate masses of analytical grade materials (BDH) in triply distilled water. This buffer solution was saturated with oxygen by sparging with pure oxygen (BOC) for one hour before the commencement of electrochemical measurements. According to the Kent oxygen meter handbook [169], such a saturated aqueous solution at 20 °C will contain 1.35 x 10-6 mol 0 2 m-3. Electrochemical measurements were made in a glass, three compartment electrochemical cell of conventional design (see Fig. 7.1).

A Hi-Tek PPR1 waveform generator provided the control potentials for the potentiostat, which was built in Imperial College using a conventional operational amplifier circuit design. A gold or platinum rotating disc electrode (see section 3.2) was used as the working electrode, a bright platinum flag as the counter electrode and a saturated calomel electrode (EIL) as the reference electrode. All potentials are quoted relative to the standard hydrogen electrode (SHE), assuming that the potential of the saturated calomel electrode was 0.242 V vs. SHE. The working electrodes were spun using a motor unit (Oxford Electrodes) which allowed the rotation speed to be continuously varied up to 50 Hz. The current flowing at the working electrode was Oxygen Reduction 130

passed through an internal resistor in the potentiostat and the resulting voltage was applied to the y-plates of a Nicolet 5091 storage oscilloscope. The potentiostat control voltage was applied to the x-plates which enabled a voltammogram to be recorded on the oscilloscope. Permanent copy was obtained on a Gould 60000 x-y plotter, by connecting this to the plotter output terminals of the oscilloscope.

The gold and platinum electrodes were pretreated by potential cycling from -0.8 V to +1.2 V vs SHE at 10 V s-1 (see section 3.2.3). Slow potential scan voltammograms were then recorded, starting from either the positive or negative potential limit, at a number of different rotation rates.

8.5 Oxygen Reduction: Results and Discussion Cyclic voltammograms showing oxygen reduction at gold and platinum electrodes are shown in Fig. 8.4:

Fig. 8.4 Cyclic Voltammograms Showing Oxygen Reduction Gold and Platinum Rotating Disc Electrodes, co = 9 Hz. pH 9.3. [ 0 2] = 1.35 mol n r 3. Scan rates, Pt = 10 mV s_1, Au = 1 mV s"l. Oxygen Reduction 131

The reversible potential for the O2 / H2O couple at pH 9.3 is 0.681 V vs. SHE. However, no reduction currents flowed at platinum or gold electrodes until the potential was reduced to below 0.4 V and 0.2 V vs. SHE respectively. This demonstrates that the direct reduction of oxygen to water is a highly irreversible process. The reversible potential for the O 2 / H2O2 couple (in equilibrium with 1 mol H 202 m-3) at this pH is 0.221 V vs. SHE. Thus, it can be seen that oxygen reduction at a gold electrode does not commence until this potential is reached.

At neither electrode surface was a clear, diffusion limited current plateau seen before hydrogen evolution commenced. Both electrode surfaces showed some degree of hysterisis, but this was most noticeable in the case of gold. This behaviour suggested that the gold surface was deactivated on the positive-going scan, at a potential of -0.35 V vs SHE. Hoare [168] noticed a similar deactivation after the first scan and suggested that it was due to a change in the electrode surface owing to the activity of a Au-0 layer, which changes with time. The same author [170] also noted that the presence of platinum sites on a gold surface can alter the electrocatalytic activity of a gold surface.

It is apparent from Fig. 8.4 that oxygen reduction requires a substantially lower overpotential on a platinum electrode, but the current still does not show a perfectly defined diffusion limited plateau (cf. to Chapter 7, Fig. 7.5). Nevertheless, the reduction current at a potential of -0.5 V vs. SHE did show an approximately linear dependence on the square root of the rotation rate, (co)1/2. The reduction currents that would be expected from the Levich equation (7.14) , assuming that the diffusion coefficent D 0(02) = 1.8 x 10"9 m2 s-1 [171], were also calculated. The experimental results (on platinum), together with the theoretical two and four electron reduction currents, are shown in Fig. 8.5 (overleaf). Oxygen Reduction 132

□ i / mA ♦ 2e n 4e

Fig. 8.5 Experimental and Calculated O2 Reduction Currents at a RDE. Pt RDE area = 3.85 x 10'5 m2, [ 0 2] = 1.35 mol m'3. T = 293 K.

As can be seen from the above figure, the experimentally observed values fall in between those expected for two and four electron reductions. At very low rotation rates the current was close to that expected for a four electron process; as the rotation rate was increased the current tended towards the two electron limit. This was consistent with the suggestion that hydrogen peroxide is produced as a metastable intermediate. At a low rotation speed H 2O2 remains on or close to the platinum surface and can be further reduced to water, whereas at high speed more H 2O2 is dispersed into solution.

8.6 Summary Oxygen reduction was shown to be a slow reaction at gold and platinum electrodes. Platinum was a more effective electrocatalyst for oxygen reduction than gold; however, it did not show reduction currents large enough to be attributed to the complete four electron reduction of oxygen to water. The reduction currents that were observed were intermediate in magnitude between those expected for two and four electron processes. This behaviour suggested that hydrogen peroxide was formed as a metastable intermediate in a two electron reduction, and was then reduced further to form water.

This conclusion that hydrogen peroxide is an important intermediate is in agreement with the results of previous workers; it is due to the high strength of the oxygen- oxygen bond. A direct four electron reduction would involve the cleavage of this bond at an early stage during reduction, whilst a two electron reduction leaves the bond intact. Stretford Process Chemistry 133

9. The Redox Chemistry of the Stretford Process The Stretford Process achieves the oxidation of hydrogen sulphide to elemental sulphur. The gas containing the hydrogen sulphide is contacted with an alkaline solution containing vandadium (V) salts and anthraquinone disulphonates; the hydrogen sulphide dissolves and deprotonates in the alkaline solution and reacts with the two oxidising agents. The reduced solution is then passed to an oxidising vessel, where air is passed through the process solution. This serves to re-oxidise the solution and to recover the sulphur produced; sulphur is naturally hydrophobic and concentrates in the froth at the liquid surface, where it can be skimmed off and filtered. The oxidised solution is recycled to the gas absorber where it contacts more hydrogen sulphide. A more complete description of the Stretford Process is given in section (1.2).

In the process there are four linked redox couples; S(-II)/S(0), V(V)/V(IV), anthraquinone/anthraquinol and 0 2/H20. In the preceding chapters these redox couples have been investigated separately using electrochemical techniques. This section is concerned with the interaction between the redox couples, in order to determine the reaction mechanism that occurs in the Stretford Process.

A variety of techniques have been applied to study the chemical reactions involved: i) Stopped flow spectrophotometry has been used to follow the course of reactions that involve species which absorb in the UV-visible region of the spectrum. ii) The solution potential has been measured by the use of a suitable indicator electrode, in order to determine the extent of reduction that has occurred. iii) Small scale batch experiments have been conducted and the reaction products have been identified using 51V NMR spectroscopy, cyclic voltammetry and conventional chemical analyses.

9.1 Experimental A carbonate buffer solution of pH 9.3, containing 0.059 kmol Na 2CC>3 m-3, 0.223 kmol NaHC 03 m-3 and 0.10 kmol Na 2SC>4 was prepared by dissolving the appropriate masses of analytical grade materials (BDH) in triply distilled water. Similarly a borate buffer, having a pH of 9.2, was made up containing 12.5 mol Na2B4O7.10H2O m-3, 0.9 mol NaOH m ' 3 and 0.1 kmol Na 2SC>4 m-3. A stock solution containing 0.1 kmol HS" m"3, was prepared by dissolving an accurately weighed amount (about 12 g) of transparent, dried crystals of Analar sodium sulphide (BDH) in 500 cm 3 the appropriate deoxygenated buffer solution. The molarity of this stock solution was checked by conducting an iodate titration as detailed in section (3.2.1), and it was diluted with the appropriate volume of oxygen- free buffer before use. Stretford Process Chemistry 134

Polysulphide solutions were prepared either by dissolving the appropriate mass of Na2S4 in an oxygen-free buffer, or by dissolving elemental sulphur in a sodium sulphide solution. Stock solutions were made up containing 0.1 kmol S m "3 and were diluted for use with oxygen-free buffer solution.

Vanadium (V) solutions, containing 0.1 kmol V(V) m-3, were prepared by dissolving NaV03 or V 2O5 (BDH) in a carbonate buffer solution or a dilute sodium hydroxide solution respectively. The colourless stock solutions could be kept for many months without degradation, and they were diluted with the appropriate buffer solution before use.

Vanadium (IV) solutions containing 10 mol vanadium (IV) m "3 were prepared by dissolving 0.635 g of blue vanadyl sulphate, VOSO 4.6H2O (BDH), in 250 cm3 of oxygen-free carbonate buffer. 6.7 cm 3 of 1 kmol NaOH n r 3 solution were added to allow for the hydroxide ion consumption during reaction (9.1): I 8 VOSO4 + 48 OH- -» V180 4212- + I8 SO42- + 24 H20 (9.1) The resulting solutions were dark brown, but became green and eventually colourless if they were exposed to the atmosphere.

9.1.1 Stopped Flow Apparatus A diagram of the Stopped flow apparatus is shown in Fig. 9.1:

Fig. 9.1 Stopped Flow Apparatus. Stretford Process Chemistry 135

A Hi-Tech SFA-11 stopped flow attachement was modified for use with oxygen sensitive solutions by using glass syringes and PTFE-lined stainless steel tubing throughout. The attached quartz cell had four optical faces, so that it could be used with a 2 or 10 mm path length. The two reservoir syringes were filled with the reactants and the optical cell was placed in a Hewlett Packard 8451A diode array spectrophotometer. The two reactants were loaded into the drive syringes from the reservoir syringes. Then, when the syringe pistons were simultaneously depressed by the drive plate, they were mixed within the optical cell.

The diode array spectrophotometer was capable of recording a full UV-visible spectrum in 100 ms, and such spectra were recorded after fixed time intervals following the mixing of the reagents. The spectra were stored (as digital data) in the memory of the machine and could be transferred onto a magnetic disc for permanent storage. Using the Hewlett Packard program KINETICM, the spectra could be recalled and the absorbance values extracted at a fixed wavelength.

9.1.2 Experimental: Measurement of Solution Potential The solution potential was measured throughout the reaction between 2,7, anthraquinone disulphonate (AQ27DS) and sodium sulphide solution using a gold bead electrode. A diagram of the apparatus used is shown in Fig. 9.2:

seal electrode Solution flow

Fig. 9.2 Gold Indicator Electrode for Measuring the Solution Potential Stretford Process Chemistry 136

The potentials was measured relative to the saturated calomel electrode (EIL) and converted to the SHE scale assuming that the potential of the latter was 0.242 V vs. SHE. The indicator electrode assembly was connected to the stopped flow apparatus in place of the stop syringe, so that the chamber containing the indicator electrode was flushed with the reaction mixture at the same time as the optical cell was filled. Care was taken to completely expel all the air bubbles at this stage. The reaction mixture initially contained 0.16 mol AQ27DS m ~3 and 50 mol Na 2S n r 3 in deoxygenated carbonate buffer. The potential between the gold bead and the reference electrodes was measured at 600 s intervals, as the reaction between AQ27DS and HS“ proceeded.

9.1.3 Experimental: Preparation of Samples for 5iV NMR Four samples were prepared for 51 v Nuclear Magnetic Resonance (NMR) spectroscopy containing: 1 . 100 mol NaV 03 m~3 bi carbonate buffer. 2 . 10 mol NaV 03 m-3 in carbonate buffer. 3. 500 mol NaV 03 m"3 and 500 mol Na 2S m"3 in carbonate buffer. 4. 5 mol NaV 03 m"3 and 500 mol Na 2S m-3 in carbonate buffer.

All the samples were thoroughly deoxygenated before they were mixed, and were sealed into glass NMR sample tubes under a nitrogen atmosphere. Samples one and two were colourless, sample four became yellow as the reagents were mixed and sample three became a dark-brown colour and a black, hydrophobic solid precipitated from the solution.

Spectra were obtained using a Bruker 200 NMR spectrometer, using an exciting radiation frequency of 52.6 MHz. Liquid VOCI 3 was used as a reference material and all chemical shifs are quoted relative to it. Stretford Process Chemistry 137

9.2 Reaction Between AQ27DS and HS": Stopped Flow Results When equal volumes of solutions containing 0.32 mol AQ27DS m -3 and 100 mol Na2S m-3 were mixed in the stopped flow apparatus, the series of spectra shown in Fig. 9.3 were obtained:

Fig. 9.3 Spectra Taken at 600 s Intervals During Reaction between AQ27DS and HS\ [AQ27DS]0 = 0.16, [HS "]0 = 50 mol n r3. T = 17 °C. Cell Path Length = 1 cm.

There above spectra are very similar to those shown in Fig. 7.11, which were obtained during the electrochemical reduction of AQ27DS. This suggests that the reduction product (which has an absorbance peak at 410 nm) was the same in both cases. In chapter 7, evidence was presented that suggested that this reduction product was AQ27DSH". There was no visible deposition of elemental sulphur during the reaction, so it is likely that the HS" ions are oxidised to form polysulphide ions (e.g S42-): 3 AQ27DS + 4 HS- + OH" 3 AQ27DSH" + S42' + H20 (9.2) Stretford Process Chemistry 138

Polysulphides ions also absorb in the UV-visible region, and the spectrum of a polysulphide solution (prepared by dissolving Na 2S4 in a carbonate buffer) is shown in Fig. 9.4:

VAVEIENGTH (rut)

Fig. 9.4 UV-visible Spectrum of Sodium Polysulphide. pH 9.3.

However, complete reduction of all the AQ27DS, according to equation (9.2), would only produce a S 42' concentration of 0.053 mol m'3. Since emax at 380 nm is 112.5 m2 mol-1, the increase in absorbance at this wavelength due to the production of the polysulphide ions would amount to only 0.06 absorbance units. This is less than 10 % of the optical absorbance at 380 nm due to AQ27DSH". Thus, the presence of polysulphides would be expected to produce a shoulder at around 380 nm on the absorbance peak at 410 nm. An inspection of Fig. 9.3 shows that such a shoulder is present.

9.2.1. Reaction of AQ27DS and HS“: Rate Studies. Since there was a large excess (approximately 200 fold) of HS~ ions over the AQ27DS in the above experiment, the [HS'] was assumed to remain constant throughout the reaction. This enabled the rate order with respect to AQ27DS to be calculated; a zero- order reaction would cause a linear decrease in [AQ27DS] with time, a first-order reaction would produce a logarithmic decrease with time, and a second-order rate would produce a linear decrease of [AQ27DS]"1 with time. Stretford Process Chemistry 139

In fact, a logarithmic decrease in [AQ27DS] (as monitored by its absorbance at 330 nm) with time was observed, demonstrating that the reaction was first-order with respect to AQ27DS. The plot of In (Abs 330 nm) against time is shown in Fig. 9.5:

Oi

-1

Ln(Abs330 nm)

-2

'30 1000 2000 3000 WO O 5000 6000 7000 8000 Time / s

Fig. 9.5 Plot of ln(Abs 330 nm) vs. Time During Reduction of AQ27DS [AQ27DS]0 = 0.16, [HS‘]0 = 50 mol m-3. T = 17 °C. 1 = 1 cm.

It follows from themathematicsof first-order kinetics, that the concentration of reactant R (in this case AQ27DS), remaining after time t will be given by: In (R) = In (R0) - kt (9.3) where R0 = the initial concentration of reactant The concentration of AQ27DS is related to the absorbance at 330 nm (A) through the Beer-Lambert law (9.4): A = e R l (9.4) e = extinction coefficient / m 2 mol"1; 1 = path length / m By substituting (9.4) into (9.5) it follows that: In (A) = In (A0) - kt (9.6) Therefore, the first-order rate constant (k) is given by the slope of Fig. 9.5, which was found to be 2.53 x 10 "4 s"1.

This value means that at 17 °C, in the presence of 50 mol HS' m-3, half the anthraquinone would be reduced after 45 minutes (i.e. t 1/2 = 45 mins.). Assuming that the rate of reaction doubles for each 10 °C rise in temperature, means that at 40 °C (at which the Stretford Process operates) t1/2 will be reduced by a factor of four. Stretford Process Chemistry 140

Nevertheless, in order to achieve 75 % AQ27DS reduction, a residence time of 23 minutes would still be required. Early Stretford Plant liquors contained only anthraquinone disulphonates; these plants were characterised by long residence times (in the absorber and reactor vessels) and low H 2S throughputs.

9.2.2. Reaction of AQ27DS and HS-: Solution Potential Measurements Placing an inert metal indicator electrode in a solution containing an oxidising agent and a reducing agent which are reacting chemically, enables the extent of reaction to be monitored; as the reduction proceeds, the potential decreases. Indicator electrodes can be used in industrial processes; for instance, they can be used to follow the extent of oxidation during the oxidative leaching of uranium ores.

When there are two redox couples present in non-equilibrium conditions, the observed solution potential will lie between the reversible potentials that each couple would attain separately (given the concentrations of its oxidised and reduced forms). However, this value will lie closer to potential of the redox couple which shows the most reversible behaviour at the electrode surface. This situation is summarised in the "Evans Diagram" shown in Fig. 9.6:

Fig. 9.6 Evans Diagram Showing Anodic and Cathodic Polarisation Curves During the Establishment of a Mixed Potential. Stretford Process Chemistry 141

If separate polarisation curves were drawn for the electrochemical reduction of AQ27DS and oxidation of HS" they would appear as shown in Fig. 9.6, with the reduction reaction polarising cathodically and the oxidation reaction polarising anodically. When the two couples are allowed to react chemically, the reaction at the indicator electrode surface gives rise to a "short circuited" reaction current, which is dependent on the reaction rate. At this particular value of the current, the two potentials are both equal to the mixed potential. The oxidation of sulphide at a gold electrode has been shown to be highly irreversible (see Fig. 3.6) and so the anodic polarisation curve rises rapidly. Conversely, the reduction of AQ27DS at a gold surface was reversible (see section 7.4), and the cathodic polarisation curve falls gently. Therefore, the observed mixed potential of a gold bead electrode in a reacting mixture of AQ27DS and HS‘ ions will lie close to the equilibrium potential of the AQ27DS / AQ27DSH" couple; its value will depend on the relative concentrations of the quinone and quinol.

If it is assumed that only the AQ27DS and AQ27DSH- concentrations determine the potential of the gold indicator electrode, that this potential is attained rapidly compared to the rate of change of concentrations and that the reduction is a first order process, then the solution potential will be given by a modified form of the Nemst equation: g _ Eo _ RT { In ([AQ27DS]0 - exp(ln[AQ27DS]0 - kt)} (9.7) zF [AQ27DS]0 Since the value of k has been determined in section (9.2.1) and values of E° = -0.273 V vs. SHE and z = 2 can be estimated from section (7.4), the variation of the potential with time can be theoretically predicted. The calculated and experimentally observed potential measurements are shown in Fig. 9.7:

Fig. 9.7 Measured and Theoretical Solution Potentials vs Time. Solution Conditions as in Fig. 9.3. Stretford Process Chemistry 142

From Fig. 9.7 it can seen that there is a reasonable agreement between the theoretical and experimentally observed values. The potential fell as the reduction of the AQ27DS proceeded, and this decrease in potential as the reaction progressed was of the correct magnitude as that predicted when z = 2 in equation (9.7). If z were to equal 1, a drop in potential twice that observed would be predicted; therefore, these measurements provide evidence that AQ27DS is reduced in a two electron process.

However, the agreement between theory and experiment is not sufficiently close to allow the potential measurements to be used to predict the concentrations of the reduced and oxidised forms, nor to determine the first-order rate constant. The discrepencies are largest at the start of the reaction, when the assumption that the potential is attained rapidly compared to the rate of change of [AQ27DS] is most suspect.

9.3 Reaction between V(V) and HS_: Stopped Flow Results. The application of UV-visible spectrophotometry to the study of the reactions of vanadium (V) was limited by the high optical absorbance of V(V) solutions in the UV region; a 10 mm cuvette containing a solution of 10 mol NaVC >3 m~3 showed complete light absorbance below 370 nm (dilution showed that ^ max = 270 nm , £ = 320 m 2 mol-1). For this reason, it was not possible to study the reactions of concentrated vanadium (V) solutions, such as those used in the Stretford Process ([V(V)] = 32 mol m"3), using the Hi-Tech SFA-11 stopped flow apparatus. However, the optical path length (2 mm) was short enough to allow a study of the reaction between a solution containing 0.5 mol V(V) m -3 and excess HS' (44 mol m~3). Stretford Process Chemistry 143

Spectra were taken at two second intervals during the reaction between the two above solutions, and the resulting series is shown in Fig. 9.8:

Wavelength / nm

Fig. 9.8 Spectra Taken at 2 s Intervals During Reaction between V(V) and HS“. [V(V )]0 = 0.5, [HS-]0 = 44 mol n r3. T = 17 °C. Cell Path Length = 2 mm.

The reaction was extremely rapid, and an absorbance peak at 360 nm appeared within the first two seconds. An attempt to decrease the rate of reaction by reducing the [HS~] by a factor of ten succeeded only in decreasing the magnitude of the absorbance maxima. This behaviour suggested that an equilibrium was rapidly established between the HV2O73' ions and the HS" ions. An examination of the spectral properties of the known thiovanadate complexes (see Table 4.3) revealed that the complex ion V02S23- possessed an absorbance maxima at 360 nm. This suggested that an equilibrium involving this ion may have been established: HV20 73- + 4HS- 2 V 02S23" + 2H 20 + OH" (9.8) After the rapid formation of the absorbance maxima at 360 nm, there was a small increase in the optical absorbance in the spectral region 300-380 nm with time. Polysulphide species absorb in this spectral range, and they are likely to have been responsible. Indeed, separate experiments had shown that such an increase could also be seen when the HS" solution was allowed to react with an aerated buffer. Stretford Process Chemistry 144

Vanadium (IV) solutions did not show an absorbance maximum at 360 nm. Instead, these brown solutions showed a broad absorbance throughout the whole UV-visible spectral range. This complex and intense spectrum is consistent with vanadium (IV) existing as the complex polyanion V jgC ^12-. This spectral pattern was not observed in the above experiment, which shows that V ^g C ^12- was not formed under these conditions.

The 51V NMR spectra of 10 mol V(V) m -3 showed peaks at -547, -562 and -573 ppm vs. VOCI3; these were attributed to the species H2V2072-, HV2O73" and VgC^3- respectively. Upon reaction with the sulphide solution, all these peaks disappeared, which is consistent with the formation of thio complexes. However, no peak at 184 ppm (attributed by Howarth [37] to V 02S23- complex) could be detected, although it was not clear whether the detection limits of the machine would be exceeded at this relatively low concentration (5 mol V m"3). The spectrum was not scanned above 300 ppm, so it is not possible to rule out the presence of other thio complexes (see Table 4.3).

9.3.1 Vanadium (V) Reduction At higher vanadium (V) concentrations it was not possible to follow the reaction between V(V) species and HS" ions using UV-visible stopped flow spectrophotometry, because of the strong absorbance of the V(V). However, the following observations could be made: i) When a solution containing 10-100 mol V(V) m ' 3 was reacted with an equal volume of equimolar Na 2S, the mixture instantly turned a green-brown colour. If a large excess of sulphide was added, after several minutes a brown-black solid began to precipitate from the solution. ii) If this brown-black solid was separated and dissolved in hydrochloric acid, it dissolved to form a blue solution. iii) When hydrogen peroxide was added to the green-brown solutions, the solution instantly turned turbid, and a yellow solid (which was identified as sulphur) could be separated from a clear solution.

These observations imply that when more concentrated solutions of vanadium (V) were used, reduction of the vanadium (V) to vanadium (IV) was achieved. Vanadium (IV) solutions (prepared from VOSO 4) appeared brown, but when they were exposed to air they became a blue-green colour. After prolonged exposure to the atmosphere they turned colourless. The UV-visible spectra of vanadium (IV) solutions before and after exposure to the atmosphere are shown in Fig. 9.9 (overleaf). Stretford Process Chemistry 145

Fig. 9.9 Spectra of 10 mol V(IV) m“3, before and after aeration. pH = 9.3, T = 17 °C. Cell Path Length = 10 mm.

This behaviour suggests that vanadium (IV) solutions, which contain V 180 4212" anions, can be oxidised initially to form a mixed valence V(V)/V(IV) ion in solution. Mixed valence anions that are blue and green are now known, though still poorly characterised (see section 4.3). Prolonged aeration formed the V(V) ion, HV 2O73".

It is likely that the brown-coloured solutions that result when V(V) (at vanadium concentrations > 10 mol m"3) and HS" are mixed, contain Vig04212“ ions: 12HS“ + 9 HV20 73“ -» V180 4212- + 3 S42" + 21 OH" (9.9) This reduction appears to produce predominantly polysulphides, rather than elemental sulphur, since the precipitation of sulphur was not observed unless an large excess of vanadium (V) was used.

The brown solid that precipitates when vanadium (V) is in prolonged contact with sulphide solutions must contain vanadium in the (IV) oxidation state or lower; the blue solutions that were produced when the solid was dissolved in acid are characterisitic of V 02+ ions. The possibility that the solid was a vanadium sulphide is remote, since an examination of the E^-pH diagram for the vanadium-sulphur-water system (Fig. 9.10) shows that there is no area of thermodynamic stability for a sulphide phase at pH 9.3. Stretford Process Chemistry 146

Fig. 9.10 E^-pH Diagram for the V-S-H20 System at 298 K. Dissolved S species not shown. Activity of Dissolved species = 0.01.

Fig. 9.10 was produced using the computer program POURB, which was re-written in FORTRAN 77 from a listing provided by Froning et al [136]. The thermodynamic data, in the form of AGf° values, are shown in the Appendix and were taken from Israel and Meites [30] and Mills [128].

From Fig. 9.10 it can be seen that the first solid phase to be encountered as the solution potential is decreased at pH 9.3 is V 3O5; this would be expected to dissolve in acid to form the blue vanadyl cation (V 02+), and the green V3+ ions (which would be oxidised on exposure to the atmosphere forming more vanadyl ions): V3O5 + 8 H+ V02+ + 2V3+ + 4 H20 (9.10) It is also possible that the solid may have been a mixed-valence salt (see section 4.3). Thermodynamic data are not yet available for these species and so they cannot be included in Fig. 9.10. Stretford Process Chemistry 147

Hydrogen peroxide reacted rapidly with polysulphide solutions, producing elemental sulphur: 2H 202 + 2 S42" -> S8 + 4 OH' (9.11) By contrast, when hydrogen peroxide was added to solutions containing HS" at pH 9.3 no sulphur was produced; the oxidation proceeded instead to form sulphoxy compounds, such as thiosulphate: 4H 20 2 + 2HS- -> S2032- + 5H20 (9.12) However, when H20 2 was added to the green-brown solution, produced by mixing equal volumes of V(V) and HS" (both at concentrations of 10 mol m"3), elemental sulphur was again produced. This suggested that polysulphides were produced during the reaction between V(V) and HS- and that these were oxidised to sulphur by H 20 2. Hydrogen peroxide was also a sufficiently strong oxidant to convert vanadium (IV) to vanadium (V), forming a colourless solution: vl8°4212' + 9H 20 2 + 15 o h - 9HV2073- + 12HzO (9.13) Therefore, the addition of hydrogen peroxide to a reduced solution, containing vanadium (IV) and polysulphide ions, can produce elemental sulphur and vanadium (V) ions.

9.4 Interaction of AQ27DSH" Ions with Oxygen. Hydrogen peroxide is a metastable intermediate in the electrochemical reduction of oxygen at noble metal electrodes (see section 8.1), and can also be produced when reduced anthraquinones react with oxygen in aqueous solution (see section 6.2). Because of its possible role in the Stretford Process it was decided to analyse for the presence of hydrogen peroxide during the oxidation of AQ27DSH" solutions.

Hydrogen peroxide was detected by titrating with As(III) [172]. H 20 2 oxidises As(III) to As(V) and unreacted As(III) was then back-titrated in acid solution with iodate: A s02" + H20 2 —> A s0 3" + H20 (9.14) I03- + 2 H 3A s03 + 2H+ + Cl- -> IC1 + 2 H 3A s04 + H20 (9.15) Preliminary checks were made to ensure that As(III) could not be oxidised either through prolonged aeration or by direct reaction with anthraquinone 2,7 disulphonate (AQ27DS).

A solution of AQ27DS was reduced electrochemically using the exhaustive electrolysis described in section (7.3). The reduced solution was then re-oxidised externally by mixing the solution with pure oxygen in a gas syringe, and recording the volume of gas absorbed. The re-oxidised solution was then titrated with As(III) to detect the presence of hydrogen peroxide. Stretford Process Chemistry 148

It was found that the reduced solution absorbed oxygen in the molar ratio of O 2 to AQ27DSH", 1:2. No hydrogen peroxide was detected in the re-oxidised solution; this suggested that the oxygen was reduced to form hydroxide ions in a four electron process: 2AQDSH- + 0 2 -> 2AQ27DS + 2 OH" (9.16) However, if the AQDSH" was injected into an aerated solution of As(III), under conditions of excess oxygen at all times, then hydrogen peroxide was detected quantitatively according to equation (9.17): AQ27DSH- + 0 2 + H20 AQ27DS + H20 2 + OH“ (9.17) In a solution containing As(III), any hydrogen peroxide produced would react immediately according to reaction (9.14).

This behaviour suggested that hydrogen peroxide was produced as an intermediate during the reduction of oxygen. When a reduced species was available in solution, (such as As(III) or AQ27DSH"), the hydrogen peroxide reacted with it immediately (producing AQ27DS and As(V) respectively). In the Stretford Process, the reaction between AQ27DSH- and oxygen is likely to give rise to the in-situ production of hydrogen peroxide; which is capable of producing elemental sulphur from polysulphide solutions and oxidising V(IV) to V(V).

9.5 Stretford Solution Chemistry: Electrochemical Results When a Stretford Process solution (containing 33 mol V(V) m~3 and 8 mol AQ27DS m_3 in a carbonate buffer) was reacted with HS~ ions (10 mol m“3) the solution instantly turned a turbid dark brown colour. Over the succeeding twenty minutes this turbidity disappeared, leaving a dark-brown coloured solution. Both vanadium (IV) and AQ27DSH" ions absorb strongly in the visible region, and together would produce such a brown colouration. The turbidity may have been due to reduced vanadium oxide (e.g. V 203).

In the absence of oxygen, no precipitation of elemental sulphur was observed, even after standing for twelve hours, although the solution contained a stoichiometric excess of oxidising agents. However, when air or oxygen was bubbled through the solution, sulphur was formed immediately and the red-brown colouration was discharged simultaneously. Stretford Process Chemistry 149

Continual voltammetry of the above solution during these reactions was conducted using a gold flag electrode, with the potential limits set at +0.3 V and -1.2 V vs. SEE. Initially a reduction wave was seen at around -0.285 V on the negative-going scan, which was attributed to the reduction of AQ27DS. The re-oxidation peak was partially suppressed, and occurred at 0.05 V vs. SHE. Similar patterns were seen during the voltammetry of vanadium (V) solutions at gold electrodes (see Fig. 5.5), and were attributed to the re-oxidation of vanadium oxide films. This suggested that the formation of such films deactivated the gold electrode towards the re-oxidation of AQDSH-.

When the sodium sulphide solution was added, the series of voltammograms shown in Fig. 9.11 were obtained. Two new reduction peaks at -0.65 and -0.8 V vs. SHE appeared, and increased in intensity during the 50 minutes following the addition.

Fig. 9.11 Voltammetry of Stretford Solution During Reduction [AQ27DS] = 8 mol m-3, [V(V)] = 33 mol m-3, [HS-] = 10 mol n r3. Au Flag Electrode, pH = 9.3, Scan Rate = 200 mV s"1.

When the solution was oxygenated, these same peaks disappeared in about 20 minutes. A possible explanation for this behaviour is that the species responsible for the reduction peaks are polysulphide ions; cyclic voltammetry of polysulphide solutions at a gold electrode revealed two reduction peaks (see section 3.5). The potentials of the Stretford Process Chemistry 150

observed peaks in Fig. 9.11 differ from those seen in section (3.4), (-0.5 V and -0.95 V vs. SHE), but this may be due to the state of the gold electrode surface; which is likely to have been covered by films of elemental sulphur then vanadium oxide as the negative-going scan proceeded.

9.6 The Stretford Process: Possible Mechanism The results are consistent with the following mechanism occurring in the Stretford Process:

Hydrogen sulphide dissolves in the alkaline solution producing hydrosulphide ions: H2S + OH" HS- + H20 (9.18) In the absorber and reactor vessels, the hydrosulphide ions are oxidised to form poly sulphides. In practice, a mixture of polysulphides is likely to be produced, since there is always more than one polysulphide present in significant concentrations in an equilibrated solution (see Fig. 3.15). However, the following equations will be written assuming that the predominant polysulphide species is S42": 3AQ27DS + 4H S- + OH" 3 AQ27DSH" + S42' + H20 (9.19) 12 HS“ + 9HV20 73- -> V180 4212- + 3 S42" + 21 OH" (9.20) The reaction between AQ27DS and HS" ions has been shown to be slow, whereas the reaction with vanadium (Y) is more rapid. The thermodynamic driving force is greater for the V(V)/HS" reaction, since the reversible potential values at pH 9.3 are -0.28, -0.25 and -0.10 V vs. SHE for the HS"/S(0) AQ27DS/AQ27DSH" and V(V)/V(IV) couples respectively. However, it is likely that the oxidation of hydrosulphide solutions with vanadium (V) proceeds via the formation of a thiovanadate complex; it has been shown that dilute vanadium solutions (< 1 mol V(V) m“3) may form the complex ion V 02S23" when contacted with excess HS": HV20 73- + 4HS" <-» 2 V 02S23" + 2H 20 + OH" (9.21) Transition metal thiovanadates are known to undergo intramolecular redox reactions which can give rise to poly sulphides and reduced vanadium phases: vvo2s23- -> vmo2s23- (9.22) 2 V m 0 2S23- + H2 0 - > V 20 3 + 2 S 22- + 2 OH" (9.23) Colloidal V 2O3 can dissolve to form vanadium (IV): 9V 20 3 + 9HV20 73- -4 2 V 180 4212- + 3 OH- + 3 H20 (9.25) In this way, the formation of thiovanadate complexes can increase the rate of oxidation of hydrosulphide solutions^ Stretford Process Chemistry 151

In the oxidiser, the reduction of oxygen with AQ27DSH" gives rise to the in-situ production of hydrogen peroxide: A Q D SH - + 0 2 + H20 -> 2AQ27DS + H2 0 2 + OH' (9.26) This hydrogen peroxide can achieve the oxidation of the polysulphide ions to form elemental sulphur (9.27), and assist in the oxidation of the vanadium (IV) species (9.28) and colloidal V 2O3. 2 H2 0 2 + 2 S42“ S8 + 4 OH' (9.27)

V 18O4212' + 9 H 2 0 2 + 15 OH' -> 9HV2073- + 12 H20 (9.28)

V 20 3 + 2 H2 0 2 + 3 OH- -» HV2O 73- + 3 H20 (9.29) The brown vanadium (IV) species (V 1804212-) will react slowly with oxygen to form mixed-valence V(V)/V(TV) compounds and eventually form vanadium (V): 2 V i804212" + 9 02 + 30 OH" -> I8 HV 2O 73- + 6H20 (9.30) The rising air bubbles in the oxidiser also serve to recover the sulphur produced in equation (9.23) by froth flotation, because of the naturally hydrophobic nature of elemental sulphur. The re-oxidised solution is then re-cycled to the absorbing vessel where it can undergo another redox cycle. Conclusions 152

10. Conclusions The redox couples that are involved in the Stretford Process were studied using electrochemical techniques, the interactions between them were investigated and a process mechanism was proposed. The following results emerged:

10.1 The S(-II)/S(0) Redox Couple The literature concerning the oxidation of sulphide solutions was reviewed. It revealed that previous measurements of pK ^ for H 2S had been in error and that a new value of 19 ± 2 will have to be accepted. This means that the S2" ion will not be the predominant species in aqeous solution, even in highly alkaline media. At pH 8.5-9.5, at which the Stretford Process operates, HS" ions are present. The published Efo-pH diagrams revealed that elemental sulphur was not a thermodynamically stable oxidation product at the process pH; sulphur’s only region of stability was below pH 7. Operating the process below this pH would not only decrease the dissolution kinetics of hydrogen sulphide, but would also alter the vanadium (V) speciation, producing decavandate anions. These large anions are known to have slow reaction kinetics and upon reduction they can precicipitate the sodium salts of mixed valence (V/IV) vanadates.

Therefore, the Stretford Process is required to operate at a pH where the desired product is thermodynamically unstable. Nevertherless, elemental sulphur is still produced in high yield. However, this thermodynamic instability means that the formation of some higher oxidation state products, such as thiosulphate, is inevitable. Previous studies on the atmospheric oxidation of sulphide solutions showed that a wide range of reaction rates and oxidation products could be observed, depending on the particular temperature, pH and catalyst used.

The electrochemical oxidation of HS“ ions on gold electrodes at pH 9.3 was shown to produce a sub-monolayer of a gold sulphide phase at low potentials (-0.4 V vs. SHE) and multilayers of sulphur at higher potentials (0.05 V vs. SHE). Associated with the formation and reduction of this sulphur layer, was the production of polysulphide ions, Sn2“ (n = 2 to 5). The poly sulphide ions were detected at a ring electrode in a rotating ring-disc electrode study. By comparing the charges passed during their production and reduction, the average polysulphide chain length was calculated to be 1.8.

10.2 The V(V)/V(IV) Redox Couple The literature relating to the aqueous chemistry of vanadium in alkaline solutions was reviewed. This showed that vanadium, in common with other transition metals in the same region of the periodic table, displays a marked tendency to form polymeric anions. Early disagreements about V(V) speciation have been largely resolved, but uncertainty still exists as to speciation of the lower vanadium oxidation states. V(V) exists as the ions HV2O73' and V 40j2^" in the Stretford Process solutions; upon reduction these are likely to form the brown V(IV) polyanion V180 4212”. However, Conclusions 153

prolonged exposure to reducing environments can produce a precipitate of vanadium (III) oxide (V 2O3), and mild reduction may also produce mixed-valence (V)/(IV) compounds, such as VjqC ^ 6".

The electrochemical reduction of vanadium (V) was found to be irreversible on a variety of electrode surfaces. This reduction led to the production of vanadium oxide films (V3O5, V2O3 and VO) rather that to a solution species. Irreversible electrochemical behaviour and high overpotentials are commonly associated with processes that require a large structural rearrangement, as is the case with the reduction of HV 2073- to V 18O4212". The effectiveness of V(V) as an oxidising agent in the S tretford Process suggested that there was a specific chemical interaction occurring which facilitated V(V) reduction. A range of vanadium-sulphur complexes (the thiovanadates) are known, which are likely to formed under the chemical conditions prevailing in the absorbing vessel of the Stretford Process. It is possible that these thiovanadate complexes can undergo intramolecular redox reactions, producing polysulphide ions and a reduced oxidation state vanadium complex; in this way the formation of thiovanadate complexes may offer reaction pathways with lower activation energies than would otherwise be the case. This mechanism may explain the catalytic role of V(V) in the process.

10.3 The Anthraquinone/Anthraquinol Redox Couple The reduction of the anthraquinone 2,7-disulphonate (AQ27DS) can produce either the semiquinol in a single electron process, or the fully reduced quinol in a two electron process. An equilibrium can be established between these two species: AQ27DS + AQ27DS2" <-> 2 AQ27DS-' (10.1) The comproportionation constant for this reaction can determine the voltammetric behaviour of the compound. If the equilibrium constant is high, two consecutuve one electron reductions are observed, whilst if it is low, a single wave is seen corresponding to a two electron reduction.

The values of pKaj and p K ^ for the quinol (AQ27DSH2) are ~7 and 10.8 respectively. As would be expected from these values, the electrochemical reduction of AQ27DS at pH 9.3 was found to proceed in a two electron, one proton process: AQ27DS + 2e- + H+ AQ27DSH- (10.2) However, the presence of a strong esr signal from the reduced solution showed that AQ27DSH" existed in equilibrium with the radical species, AQ27DS*". From an analysis of the voltammetric peak separation, the comproportionation constant was estimated to be in the range 0.2 to 4. Conclusions 154

The radical anion reacts rapidly with oxygen to form the superoxide ion 0 2*", which can further react with water to form hydrogen peroxide: AQ27DS-- + 0 2 -> AQ27DS + 0 2-‘ (10.3) 2 0 2*" + 2 H20 -» 0 2 + H20 2 + 2 OH- (10.4) Reactions (10.3) occurs extremely rapidly, which has the effect of shifting equilibrium (10.1) to the right. In this way the oxygen can become reduced to form hydrogen peroxide, while the reduced AQ27DSH" is oxidised to form AQ27DS: AQ27DSH- + 0 2 + H20 -> AQ27DS + H20 2 + OH- (10.5) When a solution of AQDSH" was allowed to react with oxygen, hydrogen peroxide was detected quantitatively according to equation (10.5).

10.4 The 0 2/OH_ Redox Couple The direct electrochemical reduction of oxygen at platinum and gold electrodes gave rise to currents at a rotating disc electrode which were intermediate in value between those expected for mass transport limited two and four electron processes. At low rotation rates on platinum, the current approached the value predicted for a four electron process, and at higher rates it tended towards the two electron value. This suggested that hydrogen peroxide was formed as a metastable intermediate, via a two electron process (10.6). At low rotation rates H 20 2 was further reduced at the electrode surface to form hydroxide ions (10.7), whilst at higher speeds it was dispersed into solution. 0 2 + 2e- + H20 -> H 2 0 2 + 2 OH- (10.6)

H2 0 2 + 2e" -» 2 OH- (10.7) Previous workers have also found that hydrogen peroxide can be produced in alkaline solutions from the two electron reduction of oxygen, and this has been attributed to the high strength of the oxygen-oxygen bond. The direct four electron reduction of oxygen to form hydroxide ions involves the cleavage of this bond, whilst it remains intact during the two electron reduction to form hydrogen peroxide.

10.5 The Redox Chemistry of the Stretford Process Stopped flow spectrophotometric studies indicated that AQ27DS reacted with HS" ions to form the fully reduced quinol and polysulphide ions: 3AQ27DS + 4HS" + OH" 3 AQ27DSH- + S42“ + H20 (10.8) The reaction was found to be first order with respect to AQ27DS under conditions of excess HS" and the first order rate constant was determined to be 2.53 x 10 ~4 s"l (at a temperature of 17 °C and a concentration of 50 mol HS" m-3). Conclusions 155

The high optical absorbance of V(V) and V(IV) solutions prevented the reaction between concentrated vanadium (V) solutions (> 1 mol m"3) and HS" ions being investigated. However, the reaction with a dilute vanadium (V) solution (0.5 mol m"3) led to the formation of the thiovanadate complex VO 2S23': HV20 73- + 4HS- <-> 2 V 02S23' + 2H20 + OH' (10.9) The VO2S23' ion was identified by its optical absorbance at 360 nm, although the 51V NMR peak that had been attributed these species could not be detected.

Vanadium solutions at a concentration greater than 10 mol V(V) m "3 reacted rapidly with HS' ions, forming polysulphide ions rather than elemental sulphur: 12 HS- + 9HV20 73' -> V1804212- + 3S42' + 21 OH' (10.10) Solutions containing the Vi 804212- anion were re-oxidised slowly when they were exposed to air. This re-oxidation produced a blue-green mixed valence compound initially; vanadium (V) was regenerated only after prolonged oxygenation.

10.6 The Mechanism of Sulphide Oxidation in the Stretford Process The above evidence points to the following process mechanism:

Hydrogen sulphide dissolves in the alkaline solution producing hydrosulphide ions: H2S + OH' -> HS- + H20 (10.11) In the absorber and reactor vessels, these are oxidised to form polysulphides (such as S42") and the two oxidising agents in solution are reduced: 3AQ27DS + 4HS- + OH' -> 3 AQ27DSH' + S42“ + H20 (10.12) 12 HS- + 9 HV20 73' -> Vi80 4212- + 3 S 42- + 21 OH' (10.13) The reduction V(V) to V(IV) is likely to proceed via a mechanism of thiovanadate formation, followed by an intramolecular redox reaction, and the desorption of the resulting polysulphide.

In the oxidiser, oxygen is reduced by its reaction with the anthraquinol, which leads to the in-situ production of hydrogen peroxide: AQDSH' + 0 2 + H20 -> 2AQ27DS + H20 2 + OH' (10.14) This hydrogen peroxide can react with the polysulphide ions in solution, producing elemental sulphur: 2 H20 2 + 2 S42" —> Sg + 4 OH' (10.15) Hydrogen peroxide is also capable of re-oxidising the vanadium: Vig0 42^2” + 9 H20 2 + 15 OH“ —> 9 HV2073' + 12 H20 (10.16) although vanadium (IV) can be re-oxidised more slowly by its direct reaction with oxygen. 2 V180 4212" + 9 0 2 + 30 OH' -> 18HV2073' + 6 H20 (10.17) Conclusions 156

The rising air bubbles in the oxidiser also serve serve to recover the sulphur produced, which is naturally hydrophobic and concentrates in the froth. This froth is filtered and the sulphur produced can be further purified for sale. The re-oxidised solution is then returned to the absorbing vessel where it can commence another redox cycle.

10.7 Concluding Remarks There are several process implications arising from the above mechanism:- i) Polysulphide solutions tend to be oxidised to form thiosulphate when they contact oxygen directly. Therefore, maintaining an oxygen-free environment in the absorbing tower and reaction vessel (where the polysulphides are produced) would be expected to decrease the rate of thiosulphate production. Since sulphur is thermodynamically unstable at the process pH, some production of thiosulphate is inevitable. ii) If vanadium (V) is exposed to highly reducing conditions (i.e. when too much H 2S enters the absorber) a number of reduced-valence state vanadium salts can precipitate. The oxides V 3O5 and V 2O3 are thermodynamically stable at low solution potentials, and the production of the alkali metal salts of poorly characterised mixed-valence V(V)/(IV) anions (e.g N agV ioC ^.B ^O ) is also possible. The vanadium (V) concentration in the process solution is likely to be critical; dilute solutions do not oxidise the HS" ions (merely forming thiovanadate complexes) whereas concentrated solutions are more likely to precipitate the vanadium from solution. iii) To be an effective oxygen-reduction catalyst, the anthraquinone must produce the maximum yield of hydrogen peroxide in the oxidising vessel. This is governed by the stability of the radical semiquinol, which exists in equilibrium with the fully reduced quinol. If the equilibrium concentration of the semiquinol is low, it might not be sufficient to reduce the oxygen as fast as the gas enters solution. Conversely, if the semiquinol were highly stable, it would also be unreactive towards oxygen and so would be of no use as a re-oxidation catalyst. The stability of the semiquinol will be affected by the nature and position of any substituent groups; this explains the widely different catalytic activity of anthraquinone disulphonate isomers. iv) The rate-determining step in the Stretford Process will depend upon the operating conditions; in a poorly optimised plant, the transport of H 2S into solution in the absorber, or of oxygen into solution in the oxidiser, may limit the plant's throughput. However, in most operating plants it is the oxidation of HS" to form polysulphide in the absorbers and reaction vessels is likely to be the rate-limiting step. Early Stretford plants used solutions that contained only anthraquinone salts, and they suffered low throughput rates because of the slow reaction kinetics between anthraquinone and HS" ions (the first order rate constant between AQ27DS Conclusions 157 and excess HS" at 17 °C is only 2.53 x 10 "4 s_1). The addition of vanadium (V) salts to the process solution has greatly improved the oxidation kinetics of the process, which is probably due to the role of the thiovanadates (since the electrochemcial reduction of vanadium (V) at a number of electrode surfaces has been shown to be slow). The search for a more effective sulphide oxidation catalyst may be best focused on those transition metals which are also known to form thiometalates; for example molybdenum, tungsten, and rhenium. Appendix 158

Appendix: Thermodynamic Data Used in Eh-pH Diagrams The following thermodynamic data were used in the calculation of the E^-pH diagrams. Table A.1 was taken from the recent review by Israel and Meites [30], these values were in close agreement with the earlier compilation by Post [98] (from which some of the values were taken). Species AGf° / kj mol -1 V 0.0 y 2+ -218 VO -404.2 y3+ -251.3 V20 3 -1139 VO+ -451.8 VOH2+ -471.9 v 305 -1816 V407 -2473 v 204 -1318.6 V02+ -446.4 VOOH+ -657 (VOOH)22+ -1331 V4092- -2784 V60i3 -4109 v 205 -1419.4 v o 2+ -587 VOJ- -783.7 V043- -899.1 hvo 42- -974.9 H2v o 4- -1020.9 H3V04 -1040.3 v 20 74- -1720 h v 2o 73- -1792 h 3v 2o 7- -1864 V30 93- -2356 V40124- -3202 V10°286' -7675 HV1o0285' -7708 H2Vio0 284- -7729 v o 2.h 2o 2+ -746.4 v o .h 2o 23+ -523.4

Table A.l AGf° Values for Vanadium Compounds at 298 K. Appendix 159

Table A.2 shows the thermodynamic data that were used in the calculation of the Sulphur / water E^-pH diagrams. These were taken from the recent compilation by Zhdanov [31].

Species AGf° / k j mol S 0.0 S2- 86.31 S22" 79.5 S32’ 73.6 S42- 69.0 S52’ 65.7 SQ2(aq) -300.7 S0 32- -486.6 S042- -744.6 S20 32’ -518.8 S20 42- -600.4 S20 52- -791 S20 62" -966 S20 82- -1110.4 s 3 062 - -958 s 4 062 - -1022.2

s 5 o 62 ’ -956.0 HS' 12.1 H2S(aq) -27.9 HS03- -527.8 h s2o4- -614.6 HSO4- -756.0 H2S 03 -537.9 h 2so 4 -744.6 H2S204 -616.7 H2S20g -1110.4

Table A.2 AGf° Values for Sulphur Compounds at 298 K. Appendix 160

Table A.3 shows the AGf° values for the vanadium sulphides. AHf° and S° values were given for these compounds by Mills [128], and the AGf° values were calculated using S° values of 28.93 and 31.82 J k-1 mol-1 for elemental vanadium and sulphur respectively [173].

Species AGf° / kj mol -1 VS -191.3 V2S3 -516.2 VS4 -286.0

Table A.3 AGf° Values for Vanadium Sulphides at 298 K. References 161

1. Eliot R. C. (Ed) "Coal Desulphurisation Prior to Combustion" Chem. Technology Rev . 113(Pollution Technology), (1978), 45. 2. Tsai S. Chemical Desulfurizaton of West Kentucky Coal Using Air and Steam. Ind. Eng. Chem. Process Des. Dev. 25(1), (1986), 126-133. 3 . Garside P. G. KILnGAS Process for Producing Clean Gas from High Sulphur Coal: Major Technical Accomplishments. Energy Progress. 6(1), (1986), 5-12. 4 . Simbeck D. R. and Dickenson R. L. Integrated Gasification Combined Cycle for Acid Rain Control. Chem. Eng. Prog. 82(10), (1986), 28-34.

5 . Wilkie F. Sour Gas Corrosion. Process Engineering. 67(2),(1986), 65-71. 6. Russell F. G. Application of the DELSEP Membrane System. Chem. Eng. Prog. 80(10), (1984), 48-53. 7 . Stookey D. J., Patton C. J. and Malcolm G. L Membranes Separate Gases Selectively. Chem. Eng. Prog. 82(11), (1986), 36-41. 8. Say G. R., Heinzelmann F. J, and Savage D. W. A New, Hindered Amine for Simultaneous Removal of Carbon Dioxide and Hydrogen Sulphide from Gases. Chem. Eng. Prog. 80(10), (1984), 72-77. 9 . Hydrocarbon Processing. Special Report: Gas Process Handbook. Hydrocarbon Processing. 65(4),(1986), 75-81. 10. Kohl A. and Riesenfeld F. "Gas Purification". Gulf Publishing Co. (1974), 432-441. 1 1 . Bartoo R. K. Removing Acid Gas by the Benfield Process. Chem. Eng. Prog. 80(10), (1984), 35-40.

1 2 . Mohr V. H. and Ranke G. Acid and Sour Gas Treating Processes. Chem. Eng. Prog. 80(10), (1984), 27-35. 1 3 . Chou T-S. et A1 Selective Removal of H 2S from Biogas by Packed Silica Gel. Biotechnology Progress. 2(4), (1986), 203-210.

1 4 . Coskun I. and Tollefsun E. L. Oxidation of Low Concentrations of Hydrogen Sulphide over Activated Carbon. 5th. Canadian Symposium on Catalysis, Calgary Alberta 5, (1977). 1 5 . Bolmer P. W. H2S Removal from Hydrogen Gas by Electrolysis. U.S. Patent: 3,409,520(Mobil Oil Corporation), Nov. 5th. 1968.

1 6 . Dandapani B., Scharifker B.R. and Bockris J. O'M. Electrolysis of Hydrogen Sulphide in Aqueous Solution. Proceedings: Intersociety Energy Conversion Engineering Conference. 21 (1), (1986), 262-265. 1 7 . Lim H. S. and Winnick J. Electrochemical Removal and Concentration of Hydrogen Sulphide from Coal Gas. J. Electrochem. Soc. 131(3), (1984) 562-568. References 162

1 8 . Winnick J. Electrochemical Removal of Hydrogen Sulphide from Hot Coal Gas: Electrode Kinetics. Proc. International Gas Research Conference (1984). 1 9 . Kalina D. W. and Maas E. T. Indirect Hydrogen Sulphide Conversion. Int. J. Hydrogen Energy. 157, (1985), 163. 2 0 . Olson D. C. Removal of H 2S from Gaseous Streams. U.S. Patent: 4,540,561 (1985). 2 1 . Hardison L. C. Treating Hydrogen Sulfide: An Alternative to Claus. C hem . E n g . 92(2), (1985), 62-63. 22. O'Brien D. and Birkner F. B. Kinetics of Oxygenation of Reduced Sulphur Species in Aqueous Solution. Environ. Sci. Tech. 11(12), (1977), 1114-1120. 2 3 . Moyes A. J. and Wilkinson J. S. The Development of the Stretford Process. Conference Proceedings: The Control of Gaseous Sulphur Compound Emissions 10-12 April 1973. 2 4 . Vasan S. 'Holmes-Stretford Process Offers Economic Hydrogen Sulphide Removal' and 'Holmes-Stretford Process Proved in Use'. In "Sour Gas Processing and Sulphur Recovery". PPC Books Division (Tulsa, Ok. USA), (1979), 31-33, 36-38. 2 5 . Nicklin T. and Holland B. H. Removal of Hydrogen Sulphide from Coke Oven Gas by the Stetford Process. Proceedings European Symposium: Cleaning of Coke Oven Gas. (Saarbrucken) 21-22 March. (1963). 26. Murin P. J., Trofe T. W. and Dalrymple D. A. Stretford Process Status andR & D Needs: Draft Topical Report. GRI Contract 5083-253-0936 (1985).

2 7 . Malott M. G. Preliminary Investigations of Kinetic and Mechanistic Aspects of the Holmes-Stretford Process. McMaster University (USA) Final Year Project. (1981).

2 8 . Haley T. Investigation of the Interaction between Vanadium (V) and Organic Acids by Vanadium-51 NMR Spectroscopy. British Gas Document. ALP 166. L 235 (1983).

2 9 . Wilson B. M. and Newell R. D. Hydrogen Sulphide Removal by Stretford Process. Chem. Eng. Prog. 80 (10), (1984), 40-48.

30. Israel Y. and Meites L. "Standard Potentials in Aqueous Solution”. Eds. Bard A. and Parson R. Pergamon Press. (1985), 507-527.

31. Zhdanov S. I. "Standard Potentials in Aqueous Solution”. Eds. Bard A. and ParsonsR. Pergamon Press. (1985), 93-110.

3 2 . Habayeb M. A. and Hileman O. E. Jr. 51V FT-NMR Investigations of Metavanadate Ions in Aqueous Solutions. Can. J. Chem. 58, (1980), 2255-2261.

3 3 . Pope M. T. "Heteropoly and Isopoly Oxometalates". Inorganic Chemistry Concepts Vol. 8. Springer Verlag (Berlin: GDR). 8, (1983), 180 pp.

J References 163

34. Kuhn A.T. Kelsall G. H. and Chana M. S. A Review of the Air Oxidation of Aqueous Sulphide Solutions. J. Chem. Biotechnol. 33A, (1983), 406-414.

35. Andrzheevskii M. Yu. The Mechanism of Oxidation of Hydrogen Sulphide in the Stretford Process. Deposited Document with VINITI: 583-83, (1983), 13pp. (Russian). British Library: No. HE 03249.

3 6 . Boulegue J. Simultaneous Determination of S2", Sn2-, and S 2O32" as an Aid to Ore Exploration. J. Geochem. Exploration. 15, (1981), 21-36. 37. Harrison A. T. and Howarth O. W. V51 NMR Study of Sulphido and Oxosulphido Vanadate(V) Species. J. Chem. Soc. Dalton Trans. 7, (1986), 1405-1409. 3 8 . Meyer B., Ward K. and Peter L. Second Dissociation Constant of Hydrogen Sulphide. Inorg. Chem. 22, (1983), 2345-2346. 3 9 . Yalensi G. Contribution au Diagramme Potentiel pH du Soufre. Conference Report: Committe Internationa de Thermodynamique et de Cinetique Electrochimiques (CRCTTCE). 2, Tamburini, Milan, (1951), 51-67. 4 0 . Peters E. Direct Leaching of Sulphides: Chemistry and Applications Met. Trans. B. 7B, (1976), 505-517. 41. Hamilton I. C. and Woods R. An Investigation of the Deposition and Reactions of Sulphur on Gold Electrodes. J. Applied Eletrochem. 13, (1983), 783-794. 42. Buckley A. N., Hamilton I. C. and Woods R. An Investigation of the Sulphur(-II)/Sulphur (0) System on Gold Electrodes. J. Electroanal. Chem. 216, (1987), 213-227.

4 3. Meyer B. "Sulfur, Energy, and the Environment” Elsevier (New York). (1977).

4 4 . Myers R. J. The New Low Value for the Second Dissociaton Constant for Hydrogen Sulphide. J. Chem. Education. 63 (8), (1986), 687-690.

4 5 . Giggenbach W. Optical Spectra of Highly Alkaline Sulfide Solutions and the Second Dissociation Constant of Hydrogen Sulfide. Inorg. Chem. 10 (7), (1971), 1333-1338. 4 6 . Licht. S. and Manassen J. The Second Dissociation Constant of H 2S. J. Electrochem. Soc. 134(4), (1987), 918-921.

4 7 . Ellis A. J. and Milestone N. B. The Ionization Constants of H2S from 20 to 90 °C. Geochem. et Cosmochim. Acta. 31, (1967), 615-620.

48. Zhdanov S. Chap. 6: Sulphur. In "Encyclopedia of Electrochemistry of the Elements - Vol IV”. Ed. Bard A. Marcel Dekker Inc (New York) IV, (1975), 274-355. 49. Power G., Ritchie I. and Wylie M. The Electrochemistry of Silver in Sulphide and TetrasulphideSolutions. Analytica Chimica Acta. 26 (11), (1981), 1633-1638. References 164

5 0. Schwarzenbach G. and Fischer A. Die Aciditat der Sulfane und die Zusammensetzung Wasseriger Polysulfidlosungen. Helv. Chim. Acta. 43, (1960), 1365-91. 5 1 . Chen K. Y. and Gupta S. K. Formation of Polysulphides in Aqueous Solutions. Environmental Letters. 4 (3), (1973), 187-200. 5 2 . Maronny G. Constantes de Dissociation de l'Hydrogene Sulfure. Electrochem. Acta. 1, (1959), 58-69. 5 3 . Giggenbach W. Optical Spectra and Equilibrium Distribution of Polysulphide Ions in Aqueous Solution at 20 °C. Inorg. Chem. 11 (6), (1972), 1201. 5 4 . Giggenbach W. F. Equilibria Involving Polysulphide Ions in Aqueous Solution up to 240 C / Poly sulphide - Thiosulphate Disproportionation. Inorg. Chem. 13, (1974), 1724-1733. 55. Lesner P., Winnick J., McLarnon F. R. and Cairns E. J. Kinetics of Aqueous Polysulphide Solutions I and n. J. Electrochem. Soc. 133 (12), (1986), 2517-2522. 5 6 . Hamilton I. C. and Woods R. A Voltammetric Study of the Surface Oxidation of Sulphide Minerals. Proceedings of Symposium: Electrochemistry in Mineral and Metal Processing. 84(10), (1984), 259-285.

57. Cotton F. A. and Wilkinson G. "Advanced Inorganic Chemistry" 3rd. Edn. Interscience. (1972), 113. 5 8 . Habashi F. and Bauer E. Aqueous Oxidation of Elemental Sulphur. Ind. and Engineering Chem. Fundamentals. 5 (4), (1966), 469-471.

59. Cotton F. A. and Wilkinson G. "Advanced Inorganic Chemistry" 3rd. Edn. Interscience. (1972), 451. 6 0 . Samec Z. and Weber J. Study of the Oxidation of SO 2 dissolved in 0.5 M Sulphuric acid on a Gold Electrode. Electrochim. Acta. 20, (1975), 403-419.

6 1 . Pourbaix M. "Atlas of Electrochemical Equilibria in Aqueous Solution". Pergamon Press (Bristol, England) (1966). 6 2 . Allen P. L. and Hickling A. Electrochemistry of Sulphur. Trans. Farad. Soc. 53, (1957), 1626-1635.

6 3 . Albery W. J. and Hitchman M. L. "Ring-Disc Electrodes". Clarendon Press (London), (1971).

6 4 . Moscardo-Leveist M and Plichon V. Sulphur Chemistry in Equimolar Sodium Hydroxide-Water melt. Parts I and II. J. Electrochem. Soc. 131 (7), (1984), 1538-1551.

65. Remick R. J. and Camara E. H. Electrochemistry of the Sulphide!Poly sulphide Couple. Inst. Gas. Tech. (Chicago: USA) Project Report 65911 (1983). 36 p. 6 6 . Hodes G. and Joost M. Electrocatalytic Electrodes for the Polysulphide Redox System. J. Electrochem. Soc. 127, (1980), 544-549. References 165

67. Studzinska E. Discussion of the Kinetics of Oxidation of Hydrogen Sulphide in Solution. Gaz, Woda I Technika Sanitarna. 44 (7), (1973), 207-209. (Polish). 6 8 . Alferova and Titova. Study of the Reaction Rate and Mechanism of the Oxidation of Hydrogen Sulphide in Aqueous Solution by Atmospheric Oxygen. Zhur. Prikladnoi Khim. 42 (1), (1969), 192-196. 6 9 . Cox R. A. and Sandalls F. J. The Photo-oxidation of Hydrogen Sulphide and Propan-2-thione in Air. Atmos. Environ. 8, (1974), 1269-1281. 7 0 . Pelizetti E. Photoassisted Cleavage of Hydrogen Sulphide with Semiconductor Particles. In "Photo electro chemistry, Photocatalysis and Photoreactors". Ed: Schiavello M. D. Reidel Publishing Co. (Dordrecht: Holland). NATO ASI Series C, 146, (1985), 293-304.

7 1 . Boulegue J. and Michard G. Sulphur Speciation and Redox Processes in Reducing Environments. Chemical Modelling in Aqueous Systems. Am. Chem. Soc. Symposium. 93, (1978).

7 2 . Bowers J. W. et Al. Auto-oxidation of Aqueous Sulphide Solutions. Cherny. Ind. (1966), 65-66. 73. Lefers J. B., Koetsier W. T., and Van Swaaij W. P. M. The Oxidation of Sulphide in Aqueous Solutions. Chem. Eng. J. 15, (1978), 111-120. 7 4 . Snavely E. S. and Blount F. E. Rates of Reaction of Dissolved Oxygen with Scavengers in Sweet and Sour Brines. C o rro sio n - NACE. 25(10) , (1969), 397-404. 7 5 . Cline J. and Richards F. Oxygenation of Hydrogen Sulphide in Sea Water at Constant Salinity, Temp., and pH. Envir. Sci. Technol. 3, (1969) 838. 7 6 . Bhaskarwar A. and Kumar R. Oxidation of Sodium Sulphide in a Foam Bed Contactor. Chem. Eng. Sci. 39(9), (1984), 1393-1401. 7 7 . Chen and Morris. Kinetics of Oxidation of Aqueous Sulphide by Oxygen. Environ. Sci. Tech. 6(6), (1972), 529-537. 7 8 . Selmeczi J. G. Evaluation of Rate Determining Factors in the Oxidation of Sulphides in Water. Am. Chem. Soc. Division of Water, Air, and Waste Chemistry. 6(1), (1966), 14-23. 7 9 . Puri B. R., Singh D. D., and Verma S. K. Studies in Catalytic Reactions of Carbon: Part VIIL Oxidation, of Hydrogen Sulphide by Air in Presence of Active Carbons. Indian J. Chem. 18(A), (1979), 388-391. 8 0 . Oeste F. D. The Oxidaton of Hydrogen Sulphide on Activated Charcoal - An Electrochemical Process? C arb o n . 5, (1977), 225-228. 81. Behret H., Binder H., and Sandsted E. Electrocatalytic Oxygen Reduction with Thiospinels and Other Sulphides of Transition Metals Electrochim. Acta. 20,(1975), 111-117. References 166

8 2 . Freund P. L. and Spiro M. Catalysis by Colloidal Gold on the Reaction between Ferricyanide and Thiosulphate Ions. J. Chem. Soc., Faraday Trans. 1. 82(7), (1986), 2277-2282. 83. Zhaoling C., Yuchun W., Meilang L. and Shouzhi Y. Studies on the Oxidation of Low Valent Sulphur Compounds in an Aqueous Solution containing Vanadium with Ozone in a Bubble Column. In "Separation Processes in Hydrometallurgy". Ed. Davies G.A. Ellis Horwood Ltd. (Cirencester). Proceedings of Hydrometallurgy '87, UMIST, July 7-10 1987. 275-281. 84. Kneen W. R., Rogers M. J. W. and Simpson P. "Chemistry, Facts, Patterns and Principles" Revised Ed. Addison-Wesley Publishers (London), (1984), 450. 8 5 . Neumann L. S. Oxidative Absorption of H 2S and O 2 by Iron Chelate Solutions. AIChE Journal. 30(1), (1984), 64-69. 8 6 . Beavon T. K. Removal of Residual Oxygen to Prevent Thiosulphate Formation. U.S. Patent: 3,642,448 (Ralph M. Parsons Co.) 15 Feb. 1972. 87. Kapusta S. Viehbeck A., Wilhelm S., and Hackerman N. The Anodic Oxidation of Sulfide on Platinum Electrodes. J. Electroanal. Chem. 153, (1983), 175-183.

8 8 . Garrels R. M. and Christ C. L. "Solutions, Minerals, and Equilibria". Harper and Row (New York), (1965), 257-258. 8 9 . Angus J. C. and Angus C. T. Computation of Pourbaix Diagrams Using Virtual Species: Implementation on Personal Computers. /. Electrochem. Soc. 132(5), (1985), 1014-1019. 9 0 . Angus J. C. and Zappia. Potential-pH Diagrams for Complex Systems. J. Applied Electrochem. 17(1), (1987), 1-22. 91. Dickertmann D., Schultze J. W. and Vetter K. J. Electrochemical Formation and Reduction of Monomolecular Oxide Layers on Gold Single Crystals. J. Electroanal. Chem. 55, (1974), 429-443.

9 2 . Albert A. and Serjeant E. "The Determination of Ionisation Constants: A Laboratory Manual". Chapman and Hall (London) 3rd. Edn. (1984), 218.

93. Goodenough M. R. "ElectrohydrometallurgicalProcessing of Covellite (CuS) and Galena (PbS)". Ph.D. Thesis, Imperial College, University of London. December 1986.

94. Huang L. M., Bauldreay J. M., Wilhelm S. M., Kapusta S. D. and Hackerman N. The Anodic Oxidation of Sulfide on Gold and Platinum Electrodes. Proc. of Int. Symposium: Surfaces, Inhibition, and Passivation. 86-7, (1986), 493-506.

95. Zhdanov S. Chap. 6: Sulphur. In "Encyclopedia of Electrochemistry of the E lem en ts - Vol IV ”. Ed. Bard A. Marcel Dekker Inc (New York) IV, (1975), 328. References 167

9 6 . Marcus P, Montes S. and Oudar J. Electrochemical Behaviour of Silver and Gold Single Crystal Surfaces Covered with a Monolayer of AdsorbedSulphur. J. Electroanal. Chem. 119, (1981), 137-148. 9 7 . Zipperian D. C., and Raghavan S. The Recovery of V from Dilute Acid Sulphate Solutions by Resin Ion Exchange. Hydrometallurgy. 13, (1985), 265-281. 9 8 . Post K. The Kinetics of the Oxidation of Acidic Solutions of Aqueous Vanadium (IV) with Oxygen. Ph. D. Thesis. Univ. NSW Australia. Dissertation Abstracts International B. 37(2), (1976), 881. 9 9 . Post K. and Robins R. G. Thermodynamic Diagrams for the Vanadium- Water System at 298.15 K. Electrochim. Acta. 21, (1976), 401-405.

1 0 0 . House C. I. "The Electrogeneration ofC r(II) and V(II) Solutions and the Hydrometallurgical Reduction ofSn02, PbS and CuFeS 2". Ph. D. Thesis. University of London (Imperial College). January 1986. 101. Babel D. K., Banu R., Shanker R. and Bakore G. V. Oxidations with Vanadium(V). Journal of Scientific and Industrial Research. 43, (1984), 250-260. 102. Bailar J. C., Emeleus H., Nyholm R. and Trotman-Dickenson A. "Comprehensive Inorganic Chemistry". Pergamon Press (Oxford). (1975), 625-632. 1 0 3 . Griffith W. P. and Wickins T. D. Raman Studies on Species in Aqueous Solutions. Parti. The Vanadates. J. Chem. Soc.(A). (1966), 1087-1092.

1 0 4 . Howarth O. W. and Richards R. E. Nuclear Magnetic Resonance Study of Polyvanadate Equilibria by Use of 51V. /. C hem . Soc. 1, (1965), 864-870. 1 0 5 . Ingri N. and Brito F. Equilibrium Studies of Polyanions. VI. Polyvanadates in Alkaline Na(Cl) Medium. Acta Chem. Scand. 13, (1959), 1971-1995. 1 0 6 . Jahr K. F. and Schoepp L. Cryoscopic Measurements on Orange-Colored Sodium Polyvanadate Solutions. Z. Naturforsch. 146, (1959), 468-469. 1 0 7 . Schiller K. and Thilo E. Spectrophotometric Study of Vanadate Equilibria in Dilute Aqueous Solution. Z. Anorg. und Allgem. Chem. 310, (1961), 261-285. 1 0 8 . Evans H. T., Swallow A. G. and Barnes W. H. Structure of the Decavanadate Ion. J. Amer. Chem. Soc. 86(19), (1964), 4209-4210. 1 0 9 . Evans H.T. The Molecular Structure of the Isopoly Complex Ion, decavanadate. Inorg. Chem. 5(6), (1966), 967-977. 1 1 0 . Evans H. T. and Block S. The Crystal Structure of Potassium Cesium Trivanadates. Inorg. Chem. 10, (1966), 1807-1814.

1 1 1 . Rossotti F. and Rossotti H. Studies on The Hydrolysis of Metal Ions. Acta Chemica Scandinavia. 9, (1955), 1177-1192. References 168

1 1 2 . Rohrer C. et AI. A study of Heterogenous Equilibria in Aqueous Solutions of the Sulphates of V (IV) at 30 °C. J. Amer. Chem. Soc. 64, (1942), 2810-2816. 113. Johnson G. K. and Schlemper O. E. Existence and Structure of the Molecular Ion 18-Vanadate (IV). J. Am. Chem. Soc. 100, (1978), 3645-3646. 1 1 4 . Ostrowetsky S. Contribution a l'etude des isopolymolybdates et isopolyvanadates reduits: II and m. Bull Soc. Chim. Franc. (1964), 1012-1035. 115. Forslund B. On the Preparation and Properties of Mixed Valence Alkali Vanadium Oxides. Chemical Communicatons, Univ. Stockholm. 4, (1984), 55 pp. 1 1 6 . Hayek E. and Pallasser U. Alkalisalze eines neuen Vanadat (IV,V)-Ions. Monatschefte fur Chemie. 99(6), (1968), 2494-2497. 117. Volkov V. L. Chemisorption of O 2 on Oxides of the V2O5-V2O4 System. Russ. J. Phys. Chem. 59(4),(1985), 549. 118. Bosch H. Kip B. J., Van Ommen J. G. and Gellings P. J. Factors Influencing the Temperature Programmed Reduction Profiles of Vanadium Pentoxide. J. Chem. Soc. Faraday Trans. 1. 80, (1984), 2479-2488. 119. Gardner H., Lu W. and Rogers L. Chemical Evolution of CO 2 by Electrocatalytic Oxidation of Vanadium Impregnated Coke Anode Material. J. Applied Electrochem. 16, (1986), 357-364. 120. Gala J., Malachowski A. and Nawrat G. Electrolytic Ni/Mo/V Alloy Coatings as a Material with a Decreased H 2 Overvoltage. J. Applied electrochem. 14, (1984), 221-230. 121. Elouadseri M. and Vittori O. Reactivite Electrochimique de Quelques Oxydes de Vanadium en Milieu HC11 M. Electrochimica Acta. 31(10), (1986), 1335-1339. 122. Van den Berg, C. M. and Huang, Z. Q. Direct Electrochemical Determination of Dissolved Vanadium in Seawater. Anal. Chem. 56, (1984), 2383-2386.

123. Filipovic I., Hahl Z., Gasparac Z. and Klemencic V. Polarographic Characterisitcs of +5 Vanadium in Phosphate, Borate and Carbonate Buffers. J. Amer. Chem. Soc. 76, (1954), 2074.

124. Stromberg A. G., Dubova N. and Kaplin A. Mathematical Model for V(V) Electroreduction Kinetics under Conditions of Stripping Voltammetry. Soviet Electrochemistry. 21(1), (1985), 71-75.

1 2 5 . Dean G. A. and Herringshaw J. F. The Air Oxidation of Vanadium (IV) in Alkaline Solution. T a la n ta . 10, (1963), 793-799.

126. Rao C. N. R. and PIsharody K.P.R. "Progress in Solid State Chemistry - Transition M etal Sulphides" Pergamon Press. 10(4), (1974), 207-270. 127. Harris S and Chianelli R. Catalysis by Transition Metal Sulphides. J. Catalysis. 98(1), (1986), 17-32. References 169

1 2 8 . Mills K. C. "Thermodynamic D ata for Inorganic Sulphides Selenides and Tellurides". Butterworth (London), (1974), 669-670. 1 2 9 . Busine A. and Tridot G. Reduction et Sulfuration des Solutions de Vanadate de Sodium par le Sulfure d'ammonium. Bull. Soc. Chim. Fr. (1961) 1383-1387. 130. Basu S. K. and Taniguchi M. Thermal Analysis of the Oxidation of the Vanadium Sulphides V^Sg’ and V 3S4. /. Thermal Anal. 29(6), (1984), 1209-1223. 131. Muller A., Diemann E., Jostes R., and Bogge H. Transition Metal Thiometallates: Properties and Significancein Complex and Bioinorganic Chemistry. Angew. Chem. Int. Ed. (Engl.) 20, (1981), 934-954. 1 3 2 . Werle P. Diplomarbeit, Gottingen. (1970). 1 3 3 . Ranade A. C., Muller A. and Diemann E. Evidence for Existence of New Thioanions of Vanadium and Rhenium by their Electronic Spectra. Zeitschrift fur Anorganische und Allgemeine Chemie. 373, (1970), 258-265.

1 3 4 . Yatsimirskii K. B. and Zakhorava L. A. V S^- Decomposed via Two Intermediates. Zhur. Neorg. Khim. 10, (1965), 2065.

135. Vydra F., Stulik K. and Julakova E. "Electrochemical Stripping A nalysis". Ellis Horwood (Chichester: England), (1976), 252. 136. Froning M. H., Shanley M. E. and Verink E. D. An Improved Method for Calculation of E^-pH Diagrams of Metal-Ion-Water Systems by Computer. Corros. Sci. 61, (1976), 371-377.

137. Chambers J. Q. "The Chemistry of the Quininoid Compounds". Ed. Patai S. John Wiley and Sons (Interscience). Pt. 2 , (1974), 737-793. 1 3 8 . Bailey S. I. and Ritchie I. M. A Cyclic Voltammetric Study of the Aqueous Electrochemistry of Some Quinones. Electrochim. Acta. 30(1), (1985), 3-13. 139. Quershi G. Ali., Svehla G., and Leonard M. A. Electrochemical Studies of Strongly Chelating Anthraquinone Derivatives. The Analyst. 104(1241), (1979), 705-722.

140. Heyrovsky A. and Kuta B. "Principles of Polarography". Czechoslovak Academy of Prague, and Academic Press (New York), 1966, 181. 141. Savenko E. M., Leibman D. and Tedoradze G.Adsorption Waves During the Reduction of Anthraquinone 1,5-Disulphonic Acid. Soviet Electrochemistry (Eng. Trans.) 14, (1978), 404.

1 4 2 . Savenko E. and Tedoradze G. Adsorption Phenomena in Solutions of Some Anthraquinone Sulfonic Acids: Factors Responsible for Postwaves. Collection Checkoslovak Chem. Commun. 48, (1983), 550-558.

143. Zuman P. "Substituent Effects in Organic Polarography". Plenum Press (New York). (1967), Chap. VIII. References 170

144. Randell D. R. and Philips E. Removal of Hydrogen Sulphide from Gases and Liquid Hydrocarbons. UK Patent: GB 2 029386A, (1979). 145. Loeff I. Treinin A. and Linschitz H. The Photochemistry of 9,10- Anthraquinone-2-sulphonate in Solution. 2. Quenching vs. Radical formation at High Anion Concentration. J. Phys. Chem. 88(21), 1984, 4931-4937.

146. Moore J. N. "Time Resolved Spectra of W ater Soluble Anthraquinones". Ph.D. Thesis, University of London (Imperial College), (1986). 147. Kano K. and Matsuo T. Photochemistry of Micellar System I. Stabilizaton of the Radical Anions of Anthraquinone sulfonates. Bull. Chem. Soc. Japan. 74,(1974), 2836-2842. 148. Cotton F. A. and Wilkinson G. "Advanced Inorganic Chemistry" 3rd. Edn. Interscience. (1972), 417.

149. Considine D.M. (Ed.) "Chemical and Process Technology Encyclopaedia" . McGraw-Hill (New York). (1974).

150. Burns D. and Harle R. The Chromatographic Analysis of Some Anthraquinone Derivatives. Analytica Chimica Acta. 100, (1978), 563-569. 151. Keita B. and Nadjo L. Catalytic Synthesis of Hydrogen Peroxide: An Attractive Electrochemical and Photoelectrochemical route to O 2 Reduction. J. Electroanal. Chem. 145, (1983), 431-437. 152. Hunter A. Separation of Anthraquinone Sulphonic acids by Ion-Pair Liquid- Chromotography. Journal of Chromotography. (1985), 319-330. 153. Angell D. H. and Dickinson T. The Kinetics of the Ferrous Ferric and Ferro Ferricyanide Reactions at Platinum and Gold Electrodes. Part 1 Kinetics. J. Electroanal. Chem. 35, (1972), 55-71.

154. Weast R. C. (Ed). "CRC Handbook of Chemistry and Physics". 64th. Edn. CRC Press Inc. (Boca Raton, Florida, USA). 1983, F - 46.

155. Bard A. J. and Faulkner L. R. "Electrochemical Methods: Fundamentals and Applications" . John Wiley and Sons (New York), 1980, 218-220.

156. Richardson D. E. and Taube H. Determination of E ^-E 0} in Multistep Charge Transfer by Stationary-Electrode Pulse and Cyclic Voltammetry... Inorg. Chem. 20, (1981), 1278. 157. Polcyn D. S and Shain I. Multistep Charge Transfers in Stationary Electrode Polarography. Analytical Chemistry. 38(3), (1966), 370-375.

158. Compton R.G. "Electrochemical ESR". Ph.D. Thesis, Chemistry Dept., University of Oxford. 1980.

159. McQuillan A.J. Private Correspondence, 1987. Chemistry Dept. University of Otago, PO Box 56, Dunedin, New Zealand. 160. Kneen W. R., Rogers M. J. W. and Simpson P. "Chemistry, Facts, Patterns and Principles” Revised Ed. Addison-Wesley Publishers (London), (1984), 438. References 171

1 6 1 . Weast R. C. (Ed). "CRC Handbook of Chemistry and Physics". 64th. Edn. CRC Press Inc. (Boca Raton, Florida, USA). 1983, F-180. 1 6 2 . Tarasevich M. R., Sadkowski A. and Yeager E. Oxygen Electrochemistry; Chap. 6 in "Comprehensive Treatise of Electrochemistry Vol. 7". Eds. Conway B. E., Bockris J. O'M., Yeager E. and Khan S. Plenum Press (New York), (1983), 301-398.

1 6 3 . Schiffrin D. J. The Electrochemistry of Oxygen; Chap. 4 in "Specialist Periodic Reports - Electrochemistry Vol. 8 ”. Ed. Fletcher D. Royal Society of Chemistry (London), (1983), 126-170. 1 6 4 . Vesovic V., Anastasijevic N. and Adzic R. R.Rotating Disc Electrode: A Re-examination of some Kinetic Criteria with Special Reference to Oxygen Reduction. J. Electroanal. Chem. 218, (1987), 53-63. 1 6 5 . Biegler T., Rand D. and Woods R. Oxygen Reduction on Sulphide Minerals. Part I. Kinetics and Mechanism at Rotated Pyrite Electrode. J. Electroanal. Chem. 60, (1975), 151-162. 166. Miles M. H., Klaus E., Gunn B., Locker J. and Serafin W. The Oxygen Evolution Reaction on Platinum, Iridium, Ruthenium and Their Alloys at 80 °C in Acid Solution. Electrochim. Acta. 23(6), (1978), 521-526. 167. Chariot G., Badoz-Lambling J. and Tremillon B. "Electrochemical Reactions”. Elsevier (New York), (1962), 101. 1 6 8 . Hoare J. P. Oxygen Overpotential on Bright Gold - 1. Electrochim. Acta. 11, (1966), 311-320.

169. Kent Industrial Measurements. "Operating Instructions for Model 8012 Dissolved Oxygen Probe". Kent Industrial Analytical Instruments, Hanworth Lane, Chertsey, England KT16 9LF., p. 17. 1 7 0 . Hoare J. P. A Cyclic Voltammetric Study of the Gold-Oxygen System. J. Electrochem. Soc. 131(8), 1984, 1808-1815. 1 7 1 . Kelsall G. H. Private Correspondence (1987). Dept, of Mineral Resources Engineering, Imperial College, London.

1 7 2 . Vogel A. "Textbook of Quantitative Inorganic Analysis" (4th. Edn). Revised by Bassett J., Denney R., Jeffery G. H., and Mendham J. Longman (New York: USA), (1978), 377-379.

173. Weast R. C. (Ed). "CRC Handbook of Chemistry and Physics". 64th. Edn. CRC Press Inc. (Boca Raton, Florida, USA). 1983, D - 87.