UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Phase-transfer mediated electroreduction of oxygen to hydrogen peroxide in acid and alkaline electrolytes Gyenge, Elod Lajos 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-611035.pdf [ 13.63MB ]
Metadata
JSON: 831-1.0058669.json
JSON-LD: 831-1.0058669-ld.json
RDF/XML (Pretty): 831-1.0058669-rdf.xml
RDF/JSON: 831-1.0058669-rdf.json
Turtle: 831-1.0058669-turtle.txt
N-Triples: 831-1.0058669-rdf-ntriples.txt
Original Record: 831-1.0058669-source.json
Full Text
831-1.0058669-fulltext.txt
Citation
831-1.0058669.ris

Full Text

P H A S E - T R A N S F E R M E D I A T E D E L E C T R O R E D U C T I O N O F O X Y G E N T O H Y D R O G E N PEROXIDE IN ACID A N D A L K A L I N E E L E C T R O L Y T E S by Elod Lajos Gyenge M . A . S c , University of British Columbia, Vancouver, Canada, 1995 M.Eng. , University 'Babes - Bolyai ' , Cluj, Romania, 1990 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF D O C T O R OF P H I L O S O P H Y in T H E F A C U L T Y OF G R A D U A T E S T U D I E S D E P A R T M E N T OF C H E M I C A L A N D B I O L O G I C A L E N G I N E E R I N G We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y OF B R I T I S H C O L U M B I A January 2001 © Elod Lajos Gyenge, 2001 In p resen t ing this thesis in partial fu l f i lment of the r e q u i r e m e n t s for an a d v a n c e d d e g r e e at the Univers i ty of Brit ish C o l u m b i a , I agree that t h e Library shall m a k e it f reely available fo r re fe rence and study. I further agree that p e r m i s s i o n fo r ex tens ive c o p y i n g of this thesis fo r scholar ly p u r p o s e s may b e g ran ted by the h e a d o f m y d e p a r t m e n t o r by his o r her representat ives . It is u n d e r s t o o d that c o p y i n g o r p u b l i c a t i o n of this thesis for f inancial ga in shall no t b e a l l o w e d w i t h o u t m y w r i t t e n p e r m i s s i o n . D e p a r t m e n t of C&^HJCKL qud CL&CriC^C °^ W M G~ T h e Univers i ty of Brit ish C o l u m b i a V a n c o u v e r , C a n a d a DE -6 (2/88) 11 A B S T R A C T Two novel methods for the electroreductioh of O 2 to H 2 O 2 have been developed and investigated in a variety of acidic and alkaline solutions. The first technique, unreported before in the literature, uses surfactants and exploits the associated interfacial effects, in order to enhance the direct, 2e~ reduction of Ch.The second method described in the present study is based on emulsion mediation of the electroreduction of O2. The latter procedure employs a three-phase, L / L / G system, composed o f an organic media, an aqueous electrolyte and oxygen gas. The organic phase is composed of a redox mediator (i.e. 2-ethyl-9,10-anthraquinone, E tAQ) , supporting electrolyte (i.e., tetrabutylammonium perchlorate, T B A P ) and cationic surfactant (i.e., tricaprylmethylammonium chloride, Aliquat® 336) dissolved in a solvent (e.g., tributyl phosphate, TBP) . Hydrogen peroxide is produced in-situ, by the mediating cycle: E t A Q electroreduction - homogeneous oxidation of the anthrahydroquinone. A n in-depth study and synergistic analysis was performed for several phenomena of importance for emulsion mediation, such as physico-chemical and electrochemical properties of the organic media, emulsification characteristics, partition of H 2 O 2 , mechanism of E t A Q - O 2 electroreduction and multiphase flow dynamics in three-dimensional electrodes. Also, an economic analysis is provided for in-situ H 2 O 2 production by emulsion mediation in an acidic electrolyte. The employed experimental techniques involved: cyclic voltammetry at u A current levels, batch electrolysis using currents up to 1 A and electrosynthesis in fixed-bed 'flow-by' electrochemical cells operated at currents up to 60 A . For the electrosynthesis experiments the cathode materials were either graphite felt or reticulated vitreous carbons with 10 to 100 pores per inch. For the surfactant modified, 'direct' O2 reduction, it was found that a triple-Cs chain cationic surfactant, Aliquat 336® (i.e. [ C H 3 ( C H 2 ) 7 ] 3 C H 3 N + C 1 " ) increased the standard rate constant for O 2 electroreduction to H 2 O 2 on vitreous carbon, minimized the H 2 O 2 electroreduction and improved the peroxide concentration and current efficiency during electrosynthesis in various acid and alkaline electrolytes. These effects were attributed to an increase of the surface p H induced by the Aliquat 336® surface structures. The emulsion mediation was most effective for acidic peroxide electrosynthesis, where at current densities above 500 A m"2 the mediated system yielded higher peroxide concentrations and current efficiencies than the surfactant modified 'direct' O 2 electroreduction. i i i Table of Contents Abstract i i List of Figures vi i List of Tables '. x i i Acknowledgments xv Motto xvi 1. Introduction 1 1.1 Trends in Wood Pulp Bleaching / Brightening: The Importance of H 2 O 2 1 1.2 Commercial Manufacture of Hydrogen Peroxide 3 1.2.1 Heterogeneous catalysis 3 1.2.2 Electrochemical synthesis 4 1.3 Obj ectives of the Present Work 7 2. Conceptual Description of the Electrochemical Systems under Investigation 8 2.1 Surfactant Modified Electroreduction of 0 2 to H 2 0 2 in Aqueous Electrolytes 8 2.2 Emulsion Mediated Electroreduction of 0 2 to H 2 O 2 8 3. Background and Literature Review 11 3.1 Fundamentals of Oxygen Electroreduction to Hydrogen Peroxide 11 3.1.1 Oxygen electroreduction in aqueous electrolytes 11 3.1.2 The effect of surfactants on the electroreduction of O 2 16 3.1.3 Oxygen electroreduction in organic media 16 3.2 Electrochemical Phenomena involving Liquid / Liquid Emulsions 18 3.2.1 Emulsions and surfactants: General aspects 18 3.2.2 Surfactant adsorption at interfaces involving a solid phase 21 3.2.3 Charge and mass transfer in emulsions 26 3.3 Electroreduction of Anthraquinone Derivatives in L / L Emulsions 30 3.4 Elements of Porous Electrode Theory 34 iv 4. Experimental Methods, Apparatus and Materials 39 4.1 Fundamental Techniques 39 4.1.1 Cyclic voltammetry 39 4.1.2 Spectrophotometry '. 39 4.1.3 Ionic conductivity measurements 40 4.2 Electro synthesis Apparatus 41 4.2.1 Batch 'H ' -ce l l 41 4.2.2 Continuous 'flow-by' cells 42 4.3 Porous Electrodes and Membranes 52 4.3.1 Cathode materials 52 4.3.2 Cation exchange membranes 54 4.4 Chemicals and Analytical Methods 54 4.4.1 Surfactants employed 54 4.4.2 Organic solvent 55 4.4.3 Hydrogen peroxide analysis 55 5. Surfactant Modified Electroreduction of O2 to H 2 0 2 . Results and Discussion 56 5.1 Cyclic Voltammetry Studies 56 5.1.1 Kinetics of O 2 electroreduction on bare glassy carbon 56 5.1.2 Influence of surfactants on the O 2 cyclic voltammograms 59 5.1.3 Effect of A3 3 6 on the 0 2 electroreduction kinetic and mass transfer parameters 65 5.2 B atch Electroreduction of O 2 on Reticulated Vitreous Carbon 71 5.2.1 Constant current coulometry. Influence of surfactants 71 5.2.2 Influence of surfactants on the electroreduction of H 2 O 2 73 5.2.3 The effect of A336 on the figures of merit for batch 0 2 electroreduction 79 5.3 Electroreduction of 0 2 to H 2 0 2 in 'Flow-by' Cells 83 5.3.1 Graphite felt vs. Reticulated vitreous carbon cathodes 83 5.3.2 Factorial experimental design: Ac id electrolyte 87 5.3.3 Experiments on the 'long' (0.5 m) 'flow-by' cell: Acid electrolyte 93 5.3.4 Factorial experimental design: Alkaline electrolyte 96 I V 5.3.5 Experiments on the ' long' (0.5 m) 'flow-by' cell: Alkaline Electrolyte.. 103 6. Emulsion Mediated Electroreduction of O2. Results and Discussion 105 6.1 Exploratory Studies: Composition of the Organic Media 105 6.1.1 Aqueous / Organic partition of H 2 O 2 105 6.1.2 Effect of cationic surfactant and supporting electrolyte on the conductivity of the organic phase 109 6.1.3 Electrical conductivity of the emulsion: Determination of the continuous phase 112 6.1.4 Preliminary electrosynthesis experiments 118 6.2 On the Mechanism of the Emulsion Mediated 0 2 Electroreduction 119 6.2.1 Cyclic voltammetry of 2-ethyl anthraquinone in the absence of O 2 119 6.2.2 The Influence of proton sources on the electrochemistry of 0 2 in tributyl phosphate 129 6.2.3 Cyclic voltammetry of the 2-ethyl anthraquinone - O 2 system 132 6.2.4 Determination of the reaction zone for 2-ethyl anthraquinone reduction in emulsion: Spectrophotometric experiments 138 6.3 Emulsion Mediated Electrosynthesis of H 2 0 2 in a Batch Cell 142 6.3.1 Influence of the specific surface Area of R V C 142 6.3.2 Emulsion mediated electrosynthesis in the batch 'H ' - cell 145 6.4 Emulsion Mediated Electrosynthesis of H 2 0 2 in 'Flow-by' Cells 152 6.4.1 Aqueous phase: 1 M N a 2 S 0 4 ; pHi„ 3.1 152 6.4.2 Aqueous phase: 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 , pH,„9.3 169 7. Process Synthesis and Economics. An Outlook 180 7.1 Introduction 180 7.2 General Assumptions 180 7.3 Operational Conditions 182 7.4 Level 1 Analysis: Gross Economic Potential (GEP) 183 7.5 Level 2 Analysis: Net Economic Potential (NEP) and Return of Investment (ROI) 184 vi 8. Conclusions and Recommendations for Future Work 193 8.1 Conclusions 193 8.1.1 Surfactant modified electroreduction of O 2 to H 2 O 2 193 8.1.2 Emulsion mediated electroreduction of O 2 195 8.2 Recommendations for Future Work 199 8.2.1 Surfactant modified electroreduction of O 2 to H 2 O 2 199 8.2.2 Emulsion mediated electroreduction of O 2 200 References 201 Nomenclature 212 Appendices A: Calculation of the Oxygen Solubility in Various Electrolyte Solutions 216 B : Calculation of the Kinetic Parameters from Cyclic Voltammetry Data for Irreversible Charge Transfer under Mixed (Activation - Diffusion) Control 217 C: Theoretical Estimation of the Surface p H in the presence of Aliquat 336 Adsorption 218 D: Estimation of the Intra- Admicellar O 2 concentration for the Aliquat 336 Surface Fi lm 221 E : Structural Characteristics and Estimation of the Mass Transfer Limited Superficial Current Density of O 2 reduction to H 2 O 2 for Graphite Felt and Reticulated Vitreous Carbon 222 F: Elements of Factorial Experimental Design 226 G : Determination of the Aqueous / Organic Partition Coefficient for H 2 O 2 228 H : Spectrophotometric Investigation of 2-Ethyl Anthraquinone Electroreduction 230 I: Cyclic Voltammograms and Related Experimental Data for the Calculation of the 234 Standard Heterogeneous Rate Constant ks, for 2-Ethyl Anthraquinone Reduction J: Experimental Data for the Calculation of the Homogeneous Rate Constant kc 238 K : Estimation of the Organic-Liquid to Solid Mass Transfer Capacity for Co-current Upward, Three-Phase (L /L /G) Flow in Graphite Felt, R V C 100 ppi and R V C 3 0 ppi 241 L : Design and Cost Calculations for the Auxiliary Equipment employed in the Acid Emulsion Mediated Peroxide Electrosynthesis Process 243 List of Figures V l l Lis t of Figures 1.1 The principle of the catalytic anthraquinone process for hydrogen peroxide synthesis 3 1.2 The Dow-Huron 'Diaphragm Flow Control Trickle-Bed' electrochemical cell for the on-site electroreduction of 0 2 to H 2 O 2 in alkaline conditions (adapted from [01oman,1996]) 5 2.1 The general principle of L / L emulsion mediated electroreduction of 0 2 to H 2 O 2 (written for acidic conditions) 10 3.1 Models for O 2 adsorption and their influence on the reduction pathway 13 3 . 2 The bicontinuous emulsion structure (adapted from [Evans and Wennerstrom, 1994]) 18 3 . 3 Representative surfactant aggregates: a) Spherical micelle; b) Inverted micelle; c) Planar bilayer; d) Bicontinuous structure (adapted from [Evans and Wennerstrom, 1994] 20 3 . 4 Generic isotherm for ionic surfactant adsorption on oppositely charge surfaces (based on [Rusling 19941]) 22 3 . 5 Three-phase contact between a solid surface ( S ) and two fluid phases: 1 (e.g. organic) and 2 (e.g. aqueous). Schematic representation of the potentially beneficial effect on the surface wetting by phase 1 induced by surfactant adsorption 23 3 . 6 Schematic representation of the electrochemical reduction of an organic compound at the organic / electrode boundary in a two-phase L / L emulsion 27 3 .7 The 'square scheme' concept applied to the electroreduction of 2-ethyl anthraquinone (EtAQ) 33 3 .8 E - p H diagram for A Q reduction in a homogeneous mixture of H 2 O - 5% v D M F . [Revenga etal, 1994] 34 3 .9 Classification of porous electrodes based on the relative direction of electrolyte and current flows 35 3 . 1 0 Hypothetical potential and current distribution along the thickness of a porous electrode 38 4.1 Cyclic voltammetry set-up 40 List of Figures vi i i 4.2 Set-up and components of the 'H ' -ce l l for batch electrosynthesis 43 4.3 General experimental set-up for the continuous operation mode employing a 'flow- 45 by' fixed bed electrochemical cell 4.4 Front view and dimensions of the cathode compartment for the 'short' flow-by cell 47 4.5 Set-up and components o f the 'short' (0.14 m effective length) flow-by cell 48 4.6 Front view and dimensions of the cathode compartment for the 'long' flow-by cell 50 4.7 Set-up and components of the 'long' (0.5 m effective length) flow-by cell 51 5.1 Representative cyclic voltammograms for O 2 on bare G C 57 5.2 The effect of surfactant type and concentration on the first scan of O2 cyclic voltammetry. Electrolyte: 0.1 M N a 2 C 0 3 . Electrode: G C 60 5.3 The effect of surfactant type and concentration on the first scan of O2 cyclic voltammetry. Electrolyte: 0.1 M H 2 S 0 4 . Electrode: G C 61 5.4 Peak current for O2 reduction obtained on the first scan vs. Aliquat 336 cone. Electrolyte: 0.1 M N a 2 C 0 3 . Electrode: G C 63 5.5 Peak current for O2 reduction obtained on the first scan vs. Aliquat 336 cone. Electrolyte: 0.1 M H 2 S 0 4 . Electrode: G C 64 5.6 The effect of successive potential cycling on the O2 voltammogram in 0.1 M H2SO4 in the presence of 17 m M Aliquat 336. Electrode: G C 67 5.7 Peak current vs. square root of scan rate for the 1-st and n-th scans. Electrode: G C 68 5.8 Current efficiency for O 2 reduction to H 2 O 2 on a 30 ppi R V C vs. surfactant type and concentration. O 2 saturated electrolyte 72 5.9 Influence of surfactants on the electroreduction of H 2 O 2 on R V C 30 ppi. Constant current coulometry 75 5.10 Cathode potentials after 1 min. of H 2 O 2 electroreduction 78 List of Figures IX 5.11 The influence of Aliquat 336 concentration on the electrosynthesis of H2O2 in 0.1 M H2SO4 at 300 A m"2. Cathode: 30 ppi R V C 80 5.12 The influence o f Aliquat 336 concentration on the electrosynthesis o f H2O2 in 0.1 M N a 2 C 0 3 at 300 A m"2. Cathode: 30 ppi R V C 81 5.13 Electroreduction of 0 2 in 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 ( p H i n 9.6). Comparison between graphite felt (GF) and reticulated vitreous carbon 100 ppi ( R V C 1 0 0 ) cathodes 84 5.14 Electroreduction of O2 in 1 M Na2SC>4 (pH; n 3.1). Comparison between graphite felt (GF) and reticulated vitreous carbon 100 ppi (RVC_100) cathodes 85 5.15 Acid electroreduction of O2 in the 'long' flow-by cell. Influence of A336 concentration. 1 M N a 2 S 0 4 pH;„ 3.1 94 5.16 Alkaline electroreduction of O2 in the 'long' (0.5 m) flow-by cell. Influence of A336 concentration. 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 , p H i n 9.4 104 6.1 The aqueous / organic partition coefficient of H2O2 as a function of the total H2O2 concentration in the emulsion. (Aqueous / Organic) phase volume ratio: 3/1 108 6.2 The conductivity of tributyl phosphate (TBP) as a function of supporting electrolyte (TB A P ) and cationic surfactant concentration 110 6.3 Molar conductivities for T B A P and A336 in T B P vs. the square root of concentration 110 6.4 The conductivity of the emulsion and each of the two-phases for a typical catholyte composition 113 6.5 Comparison between two organic phases for emulsion mediated H2O2 electrosynthesis in 0.1 M H .2S0 4 118 6.6 Influence of supporting electrolyte on the E t A Q cyclic voltammetry in aprotic T B P 122 6.7 Influence of the acid and alkali aqueous phases on the electrochemical behaviour of E t A Q in the ab sence of 0 2 125 6.8 Influence of 'strong' and 'weak' proton sources on the electrochemical behaviour of O 2 in the organic phase 131 6.9 First scan cyclic voltammogram for the E t A Q - O2 system when the organic phase is in contact with 0.1 M H 2 S O 4 133 List of Figures X 6.10 First scan cyclic voltammogram of the E t A Q - O2 system when the organic phase is in contact with 0. 1 M N a 2 C 0 3 135 6.11 Schematic representation of the colour changes observed during the electroreduction of 0.1 M E t A Q in T B P 141 6.12 Batch electrosynthesis: the effect of R V C specific surface area. 300 A m"2 144 6.13 Comparison between the L / L / G , E t A Q mediated and the O 2 / A 3 3 6 systems. Aqueous phase 0.1 M H2SO4 146 6.14 Influence of A336 concentration on the emulsion mediated peroxide electrosynthesis at 500 A m"2 for 0.1 M H2SO4 as aqueous phase 147 6.15 Influence of E t A Q concentration and superficial current density on H 2 0 2 electrosynthesis in 2 M Na2SC>4, pHin,aqu 3.1 149 6.16 Comparison between the emulsion mediated and the O 2 / A 3 3 6 systems 151 6.17 H2O2 concentration in emulsion and current efficiency for one-pass, E t A Q mediated, electrosynthesis in flow-by cells 155 6.18 Comparison between E t A Q mediation and multiphase O2 reduction in a 'flow-by' cell equipped with a graphite felt cathode 156 6.19 The influence of flow regimes on H2O2 concentration for the emulsion mediated electrosynthesis with acid aqueous phase 164 6.20 'Flow-by' cell with complete recycle of the catholyte: H2O2 concentration in emulsion and current efficiency for the E t A Q mediated electrosynthesis 167 6.21 'Flow-by' cell with complete recycle of the catholyte: Outlet p H and Cell voltage for the E t A Q mediated electrosynthesis 168 6.22 H2O2 concentration in emulsion and current efficiency for one-pass, E t A Q mediated, electrosynthesis in the flow-by cell 171 6.23 Comparison between 'direct' aqueous O2 reduction and emulsion mediation at atmospheric pressure 173 6.24 The influence of flow regimes on the alkali emulsion mediated peroxide electrosynthesis. Cathode: 100 ppi R V C 179 7.1 Conceptual flowsheet for Level 1 analysis 183 7.2 Level 2 flowsheet for the acid emulsion mediated electrosynthesis 186 List of Figures XI 7.3 The Net Economic Potential (NEP) and Return of Investment per year (ROI) for acid emulsion mediated peroxide electrosynthesis, as a function of superficial current density and individual cell voltage. Current efficiency: 85% 190 7.4 Installed Capital Cost (CIC) and Total Operating Costs (COP) for the electrochemical peroxide plant, as a function of superficial current density and individual cell voltage. Current efficiency: 85% 191 C - l The structure of the electric double layer in the presence of cationic surfactant bilayer (based on [Ingram and Ottewill, 1991]) 219 E - l Scanning electron microscopy images of graphite felt 223 E -2 Scanning electron microscopy images of reticulated vitreous carbon 100 ppi 224 H - l Absorption spectrum of the 'blank' organic phase, i.e. 0.1 M E t A Q in T B P . T: 295 K . N 2 purge 230 H - 2 Absorption spectra o f the organic layer for the acid (0.1 M H 2 S 0 4 ) emulsion 231 H-3 Absorption spectra of the organic layer for the alkali (0.1 M Na 2 COs) emulsion 231 H-4 Absorption spectrum of the organic phase after 10 minutes of mixing with 0.1 M N a 2 C 0 3 under N 2 purge. N o current applied 233 1-1 Cyclic voltammograms as a function of scan rate used for the calculation of ks for E t A Q reduction 235 1-2 Cyclic voltammograms as a function of scan rate used for the calculation of ks for E t A Q reduction. Aqueous phase: 0.1 M N a 2 C 0 3 236 J - l Cyclic voltammograms for E t A Q used in the calculation of kc with 0.1 M N a 2 C 0 3 as aqueous phase 240 L - l Inlet and outlet temperatures for the process and cooling water streams in the electrochemical cell 245 List of Tables X l l Lis t of Tables 3.1 The reactions and standard electrode potentials for O2 and H2O2 electroreduction in alkaline and acidic conditions [Sawyer, 1991; Hoare, 1985] 11 3.2 Selected exchange current densities for O2 reduction to H2O2 (25 °C) (based on [Kinoshita, 1992]) > 13 3.3 Surfactant number, preferred surfactant aggregate and corresponding emulsion type (based on [Evans and Wennerstrom, 1994; Rusling, 1994 II]) 20 4.1 Physico-chemical characteristics of the graphite felt 52 4.2 Physico-chemical properties of various ppi (pores per inch) reticulated vitreous carbon electrodes ( E R G Inc.) 53 4.3 Surfactant characteristics [Sigma, 1991] 54 4.4 Physico-chemical properties of tributyl phosphate (TBP) [Lange's, 1992; Merck Index, 1989] 55 5.1 Kinetic parameters for O2 electroreduction on bare G C at 295 K 58 5.2 Apparent kinetic and diffusion parameters for O2 electroreduction on G C in the presence of Aliquat 336 surface film at 295 K . Results based on 1-st scan cyclic voltammetry 69 5.3 Variables and their levels for the 2 4 +l factorial experiments in 1 M Na2S04 (pHi„ 3.1). Graphite Felt Cathode 87 5.4 Design matrix and 'responses' for the 2 4 +l factorial experiments in 1 M Na2S04 ( p H i n 3.1). Graphite Felt Cathode 89 5.5 Ma in and Interaction Effects for the O2 / A336 system. Catholyte: 1 M N a 2 S 0 4 (pH i n 3.1). Graphite Felt Cathode 90 5.6 Variables and their levels for the 2 4 +l factorial experiments in 0.5 M Na2C03 / 0.5 M N a H C 0 3 (pH i n 9.6). Graphite Felt Cathode 97 5.7 Design matrix and responses for the 2 4 +l factorial experiments in 0.5 M Na2C03 / 0 . 5 M N a H C O 3 ( p H i n 9 . 6 ) 98 5.8 Main and Interaction Effects for the O2 / A336 system. Catholyte: 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pH„9 .6 ) . Graphite Felt Cathode 99 List of Tables X l l l 5.9 Influence of A33 6 concentration on the measured pressure drop 101 6.1 Measured and Calculated Emulsion Conductivities 114 6.2 Standard heterogeneous rate constants ks, for E t A Q reduction in the presence of 'strong' and 'weak' proton sources 128 6.3 Pseudo first-order rate constant kc, for the homogeneous oxidation of the reduced E t A Q species 136 6.4 Estimated mass transfer coefficient, mass transfer capacity, electroactive bed thickness and superficial limiting current density for E t A Q reduction in acid emulsion using a 'flow-by' cell 154 6.5 Variables and their values used in the factorial design study of the emulsion mediated H 2 O 2 electrosynthesis using 1 M Na2S0 4 ( p H i n 3.1) as aqueous electrolyte 157 6.6 Factorial design: Emulsion H 2 O 2 concentration and current efficiency per one-pass, for the mediated electrosynthesis on graphite felt, using 1 M N a 2 S 0 4 (pH; n 3.1) as aqueous phase 159 6.7 Main and Interaction Effects for the Emulsion Mediated system with 1 M Na2S0 4 (pFfjn 3.1) as aqueous electrolyte. Graphite Felt Cathode 160 6.8 Variables and their values used in the factorial design study of the emulsion mediated H 2 0 2 electrosynthesis using 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 ( p H i n 9.3) as aqueous electrolyte. Cathode: 100 ppi R V C 174 6.9 Factorial design: Emulsion H 2 O 2 concentration and current efficiency per one-pass, for the mediated electrosynthesis on R V C J 0 0 , using 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 (pHin 9.3) as aqueous phase 175 6.10 Main and interaction effects for the emulsion mediated electrosynthesis using 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 (pHi„ 9.3) as aqueous phase. Cathode: R V C _ 1 0 0 176 7.1 Chemical costs and molar flow rates for Level 1 184 A - l Estimated O2 solubility in pertinent electrolyte solutions 216 H - l Absorption maxima for Figs. H-2 and H-3 23 2 1-1 Experimental data used in the calculation of ks 234 List of Tables xiv 1-2 Experimental data used in the calculation of ksj and ks,2, respectively 237 J - l Summary of experimental data used for the determination of kc using eqn. [6.17] .. 239 L - l Relevant Marshall-Swift Indexes 243 L - 2 Capital and utilities costs for the auxiliary equipment employed in the acid emulsion mediated peroxide electrosynthesis process 250 X V Acknowledgments It is my pleasure to thank Professor Colin W . Oloman, my research advisor, for creating the stimulating and friendly environment in which I conducted my doctoral work. His unabated interest and help with all aspects related to the present study (e.g., the research itself, conference presentations, publications), together with his thoughtful guidance and continuous encouragement to pursue new ideas on my own, were invaluable. Prof. Oloman was also influential as a teacher. It was through his courses, Electrochemical Engineering, Fuel Cells and Process Synthesis, that I learned the true meaning of synergy between fundamental research and engineering judgement and its importance for the large-scale application of any electrochemical system. I also thank the members of my committee, Prof. A . P . Watkinson, Prof. D . Dreisinger (Dept. of Metals and Material Sciences) and Dr. Z . Twardowski (Chemetics Inc.), for keeping in touch with my work and offering valuable suggestions especially in the crucial initial stages. The staff of both Chemical Engineering Department and Pulp and Paper Centre must be thanked for their help involving various and, sometimes dishevelled, aspects of graduate student life. I thank Antonio Lisboa and Joey Jung for their friendship, which meant a lot to me. The financial support from the Network of Centres of Excellence (Mechanical Wood Pulps) and a University Graduate Fellowship from U B C , are greatly appreciated. In my endeavours I have drawn upon the unconditional emotional support, encouragement and love of my mom, Marta. It is to her, and in the memory of my grandmother (Marta as well), this study is dedicated to, as a way of saying: 'Thank you for everything'. A l l these years I was fortunate to share my life (the good, bad and neutral) with Cristina. She created a home for us and gave new dimensions to my life, which made everything worthwhile. Thank you for being here next to me... xvi 'Support the unconventional, tolerate way-out and expensive ideas, and encourage the seemingly disheveled behavior of hacker life. Under these conditions - and only these conditions - will people maintain varied perspectives and demonstrate knowledge in different ways. Their many misunderstandings will ultimately be the root of new ideas.' Nicholas Negroponte, in 'Where Do New Ideas Come From?' The MIT Report, 3-4, March/April, (1996). 1 Introduction 1 C H A P T E R 1: Introduction 1.1 Trends in Wood Pulp Bleaching / Brightening: The Importance of H2O2 The pulp and paper industry is a major consumer of valuable natural resources (e.g. wood fibre), chemicals and energy; supplying ca 300 million tonnes per year of products valued at ca 250 billion U S $ per year [Oloman, 1996]. Therefore, the overall impact and 'well being' of the pulp and paper industry has paramount repercussions on the economy, environment and society in general. Starting with the early '90-s, due partly to increased environmental concerns and various economic factors (e.g. low profitability), the pulp and paper industry is going through a period of massive technological and structural (e.g. mergers) change. To increase competitiveness and profitability, there seems to be a consensus in the industry that from technology point of view, the best strategy is to produce 'value added' products with environmentally friendly processes [Ioanides, 1999]. For the bleaching / brightening operation this translates into higher pulp brightness (giving higher value*) and elemental or total chlorine free (ECF or T C F ) chemical pulp. Hydrogen peroxide, used alone or in sequence with other bleaching agents (e.g., C I O 2 , O 2 , etc.), could achieve the objectives of high brightness and environmental friendliness for both chemical and mechanical pulps. Following the lead of certain European countries (e.g. Sweden, Finland), this past decade has seen the emergence of new environmental regulations in North America, such as the E P A ' s 'Cluster Rules', which are aimed at abolishing CI2 as a bleaching agent. The chemicals that wil l benefit the most from the 'Cluster Rules', which calls for E C F chemical pulp bleaching in all the mills by 2004, are sodium chlorate (to produce on-site chlorine dioxide) and hydrogen peroxide. It is important to note that C I O 2 and H 2 O 2 can be used either separately or in combination (e.g. two-stage, C I O 2 - H 2 O 2 bleaching) to enhance the final brightness while preserving pulp strength [Whiteside, 1999]. * As a rule of thumb, a 5% increase in brightness increases the price of the final pulp / paper product by 50 $ per 20,000 m2 [Ioanides, 1999]. 1 Introduction 2 Due to the above developments, since the pulp and paper industry accounts for about 55% of H2O2 consumption world wide (e.g. 96% in Canada, 57% in U S A ) , the market for this chemical is predicted to grow at a rate of 5 to 6% per year over the next decade, exceeding 2 million tonnes per year in 2000 worth almost 2 billion U S $ [Whiteside, 1999; Webb and Mclntyre, 1996]. Traditionally, H2O2 was employed as brightening agent for mechanical pulps. The conventional peroxide brightening is done at 10 to 20% w pulp consistency, using 1 to 3%w H 2 0 2 charge per pulp at 40 - 60 °C for 1 to 3 hours [Oloman, 1996]. Employing N a O H / H2O2 weight ratio of ca 1, sets the p H of the bleaching liquor to the desired value of 10.5 - 11 (HO2" is the active bleaching agent). The liquor also contains peroxide stabilizing agents ( E D T A , MgS04, Na2Si03) to suppress the peroxide decomposition by traces of transition metal ions. Once the bleaching stage is over, the pulp is acidified to p H 5, in order to suppress the 'alkali darkening' effect. Typical problems associated with peroxide and other mechanical pulp brightening methods are: brightness reversion and yellowing. Recent research in peroxide brightening of mechanical pulps aimed at addressing the above-mention problems, reconsidered the need for alkaline conditions. It was found that a high charge of H2O2, i.e. 6 to 10% w per pulp, under almost neutral conditions (pH 7.5) gave excellent brightness stability and very low yellowing of the bleached mechanical pulp. Due to the novelty of this method and the associated intellectual property issues, very little additional information has been disclosed regarding this process [Wan, 1999]. Other research focused on the partial replacement of N a O H with Na 2 C03 (e.g. 50 to 80% substitution) in peroxide brightening of certain mechanical pulps. It was found that the Na2C03 substitution improved the brightness stability of the pulp by as much as 100%, e.g. only 2.2 points brightness loss in 9 month for the Na2C0 3 / N a O H liquor as opposed to a 4.5 points loss for the pulp brightened in N a O H solution without Na2C03 present [Rudie, 1993]. Hydrogen peroxide is a versatile bleaching agent for other types of pulp as well, such as chemical and chemi-mechanical pulps. Kemira Chemicals of Finland developed and implemented in a number of mills a novel molybdate activated, acidic peroxide bleaching process called 'GreenOx', for T C F bleaching of softwood kraft pulps [Paren and Tsujino, 1998]. Typical operating conditions are: 12% w pulp consistency, 1.5%w H2O2 charge per pulp, 0.045% w Na2Mo04 per pulp, p H 5, 90 °C, 3 hrs. A closely related processes uses activated H2O2 1 Introduction 3 simultaneously with CIO2 in the so-called ' D P c a t ' E C F bleaching [Koshitsuka, 1998]. The H 2 0 2 charge per pulp is only 0.2%,, with 0.05% w N a 2 M o 0 4 , 0.2% w C10 2 , 70 °C, p H 5, 3 hrs. Generally, bleaching of either wood or other type of fibers by various methods involving H 2 0 2 continues to be a very active area of research, which wil l likely produce new commercial processes by drawing upon state-of-the-art knowledge from several fields such as lignin and bleaching chemistry, homogeneous catalysis and chemical engineering [Anderson and Amini, 1996]. 1.2 Commercial Manufacture of Hydrogen Peroxide 1.2.1 Heterogeneous catalysis The most important industrial technology for H 2 0 2 synthesis is the catalytic anthraquinone process, which accounts for about 90% of H 2 0 2 produced worldwide. This process was first developed in Germany in the early '40s [Kirk-Othmer, 1985] (Figure 1.1). 2 Figure 1.1 The principle of the catalytic anthraquinone process for hydrogen peroxide synthesis. R: CH3CH2- or other functional group (usually alkyl). 1 Introduction 4 In the catalytic anthraquinone process the working solution containing an alkyl amthraquinone (e.g. 2-ethyl anthraquinone) dissolved in an organic solvent (e.g. 50: 50 mixture of benzene and a C 7 to C n secondary alcohol) undergoes first a catalytic hydrogenation using Raney N i or Pd based catalysts, to generate anthrahydroquinone (or anthraquinol). The product solution from the hydrogenation step, after purification to remove the traces of catalyst, is oxidized (usually by air) to reform the anthraquinone compound and simultaneously produce H 2 0 2 (Fig. 1.1). Next hydrogen peroxide is extracted from the organic solution into water, to obtain a 'crude' solution of ca 2 0 % w H2O2. The 'crude' peroxide solution is purified, concentrated and stabilized to produce the 35 to 7 0 % w merchant H2O2. Following the extraction stage, the organic working solution is subjected to extensive purification steps to remove the residual water and peroxide, before is recycled in the hydrogenator. A common technical challenge for the catalytic anthraquinone process is the possibility of excessive hydrogenation of the aromatic ring, leading to low selectivity for anthraquinol associated with 'contamination' of the working organic solution by undesired byproducts of hydrogenation [Kirk-Othmer, 1985]. Furthermore, there are problems with the catalyst, such as poisoning and pyrophoricity (e.g. Raney Ni) . Overall, the catalytic anthraquinone process is a multistep, capital intensive process, which is reflected in the fairly high price of merchant peroxide, (e.g., US$700 per ton on 100% basis is forecasted for the last quarter of 2000 [manufacturing.net, 2000]. 1.2.2 Electrochemical synthesis To address specifically the needs of the conventional peroxide bleaching processes (i.e. alkaline conditions, ca 3 % w H2O2, see Section 1.1), various on-site O2 electroreduction to H2O2 methods were developed, using carbon based porous electrodes, in trickle-bed or gas diffusion arrangements [Oloman, 1996]. Among them the Dow-Huron process using the so-called 'Diaphragm Flow Control Trickle-Bed Cel l ' was commercially the most successful so far [Nozaki, 1998; Dawe and Mathure, 1999], The cathode is composed of 1 - 3 mm graphite particles coated with a mixture of carbon black and Teflon, operated in co-current downward G / L flow at 0.1 M P a 0 2 pressure, ca 600 - 700 A m"2, 2.4 V per cell and 30 °C (Fig. 1.2). This process produces typically a product peroxide solution with a N a O H / H2O2 weight ratio of ca 1 Introduction 5 1.6 - 1.7, i.e. 7 - 8% w N a O H and 4 - 5% w H 2 0 2 , with a current efficiency of 85%. The entire electrochemical plant is completely automated and the life-time of the cell components (e.g. diaphragm, cathode) is estimated at ca 3 years [Dawe and Mathur, 1999]. The economic outlook is also very positive, since the cost of the on-site peroxide is ca 600 U S $ on a 100% basis, which is less than the price merchant H 2 0 2 , even without taking into account the transportation and storage costs associated with the latter. 0 2 0 2 I y N a O H 7% N a O H 7%; H 2 0 2 4% 0 2 + 2e" + H 2 0 -> H 0 2 ~ + O H " ; E u - -0.065 V S H E (pH 14) Figure 1.2 The Dow-Huron 'Diaphragm Flow Control Trickle-Bed' electrochemical cell for the on-site electroreduction of 0 2 to H 2 0 2 in alkaline conditions (adapted from [Oloman, 1996]). 1 Introduction 6 In spite of the above-mentioned economic advantages and technical elegance, the on-site O 2 electro-reduction technology has not so far significantly penetrated the peroxide bleaching liquor market*. The major concern of the pulp and paper industry is the high (i.e. 1.7) N a O H / H 2 O 2 weight ratio of the product peroxide solution, which is unsuitable as such for bleaching. Thus, the concentration of the product peroxide solution has to be adjusted (e.g. merchant peroxide has to be added), to obtain a N a O H / H 2 O 2 weight ratio of ca 1, which is the desired value in alkaline peroxide brightening (Section 1.1). Furthermore, due to the sensitivity of the 2e" O2 electroreduction kinetics to p H and electrolyte nature (NaOH or K O H are preferred at p H >13, see Section 3.1), an electrosynthesis process such as the Dow-Huron, cannot meet the peroxide requirements of some novel, neutral or acidic bleaching methods, such as GreenOx or D P c a t (Section 1.1). To increase the competitiveness of the alkaline O 2 reduction, two research strategies have been pursued. In one, the cathode reaction was coupled with simultaneous generation at the anode of useful chemicals such as NaC103 [Kalu and Oloman, 1990] and (NH 4) 2S208 [Dong et al, 1997] or electricity (i.e. peroxide fuel cell, [Webb and Mclntyre, 1996; Alcaide et al, 1998]). In spite of promising figures of merit on a laboratory scale (especially for the chlorate and persulfate systems), due to various technical and economic issues, none of the coupled electrosynthesis processes has been commercialized so far. In the other strategy, the research focused on lowering the N a O H / H 2 O 2 weight ratio [Foller and Bombard, 1995]. To accomplish this goal several ideas were proposed, such as using a three-compartment cell with an acid (e.g. 1 M H2SO4) middle compartment [Sudoh et al, 1991], or gas diffusion cathode with bipolar membrane for water splitting [Drackett, 1994]. None of these developments has yet reached the commercial stage. In addition to the catalytic anthraquinone and O 2 electroreduction technologies, there are a few other processes for industrial manufacture of peroxide but with very limited market share, such as the homogeneous oxidation of secondary alcohols (Shell Co.) or the electrochemical persulfate process [Kirk-Othmer, 1985]. These technologies are not directly relevant to the present work, therefore are not reviewed here. * There is only one Dow alkaline peroxide generation plant in operation on an industrial scale (i.e. 4 t / day), at the Muskogee pulp mill of the Fort Howard Co., put on-line in 1991. 1 Introduction 7 1.3 Objectives of the Present Work The general objective of the present work is to explore and develop novel electrochemical methods for the electroreduction of O2 to H2O2, which are employable in a wide range of p H (i.e. between 1 and 12) and for various electrolyte compositions (e.g. Na2C03, N a 2 S 0 4 or H 2 S 0 4 ) . Such methods would enable much greater flexibility in the end-use of the product peroxide solution in comparison with the existent O2 electro-reduction technologies. Both the fundamental electrochemistry and the electrochemical engineering aspects of the proposed O2 electro-reduction methods wil l be studied. 2 Conceptual Description of the Electrochemical Systems under Investigations 8 C H A P T E R 2: Conceptual Description of the Electrochemical Systems under Investigation The present work proposes and is concerned with two different phase-transfer mediated electrochemical systems for H2O2 synthesis, i.e. surfactant modified and L/L emulsion mediated electroreduction of 0 2 . 2 . 1 Surfactant Modified Electroreduction of O2 to H 2 O 2 in Aqueous Electrolytes The goal here is to put forward a different approach for the 2e" O2 reduction by exploiting certain interfacial effects, such as changes in surface p H and electric potential, induced by the adsorption of surfactants on the electrode surface. The influence of surfactant type (i.e. cationic, non-ionic and anionic) and concentration on various fundamental aspects of O2 electroreduction was investigated by cyclic voltammetry, constant current coulometry and batch electrolysis. Furthermore, experiments performed in two 'flow-by' cells of different size are presented, to elucidate the interplay in the presence of a surfactant, between the G / L mass transfer and electrode kinetics in porous electrodes. The novelty of this system lies in the fact that to the knowledge of this author there is no other account in the literature on the use of a surfactant to enhance the O2 reduction to H2O2 (see Chapter 3: 'Background and Literature Review'). 2 . 2 Emulsion Mediated Electroreduction of O2 to H 2 O 2 This part deals with a multi-phase system composed of two liquid phases and a gas phase (i.e. O2). The liquid phases are: an organic media, containing among others an anthraquinone compound as electrochemical mediator, and an aqueous electrolyte of choice. The basic principle of this method is two carry-out an electrochemical version of the catalytic anthraquinone process, where the anthraquinone compound (e.g., 2-ethyl anthraquinone) is recycled in-situ with simultaneous generation of H2O2 (Fig. 2.1). Such a process would eliminate the need of a separate oxidation step (see Fig. 1.1) and also it would offer other significant advantages over 2 Conceptual Description of the Electrochemical Systems under Investigations 9 the catalytic method. For instance, typical problems associated with the catalyst would be suppressed such as, the costly separation and purification steps needed to remove the moisture and peroxide from the organic solvent, or issues related to catalyst poisoning, deactivation and regeneration (see Section 1.2). Furthermore, the selectivity is usually higher for electroreduction than for catalytic hydrogenation, therefore the hydrogenation of the aromatic ring and consequently, the loss of the anthraquinone compound could be avoided by the electrochemical route. From the peroxide end-use point of view, it is very important that theoretically, there is very little restriction on both the p H and composition of the aqueous part of the emulsion (except conductivity constraints), since the main route to H2O2 is the anthraquinone reduction from the organic phase and not the p H and electrolyte sensitive aqueous 0 2 reduction. There are a host of issues that need to be addressed in conjunction with the emulsion electrosynthesis such as: H2O2 partition between the two phases, electric conductivity of the organic phase (e.g. role of a supporting electrolyte), nature of the emulsion formed, wetting of the cathode surface by the organic phase (e.g. importance of a surfactant presence), kinetic aspects of anthraquinone mediator reduction under the specified conditions and multi-phase mass transfer in porous electrodes. Although there are a few relatively recent (i.e. 1992 onward) literature reports on emulsion mediation for H2O2 electrosynthesis (see Chapter 3 'Background and Literature Review'), the novelty of the present work lies in the different composition of the employed organic phase (using also a supporting electrolyte and a surfactant), in the nature of the various emulsions explored with various acidic and alkaline aqueous phases and in addressing in conjunction, a number of issues such as H2O2 partition, emulsification, electroreduction mechanism and multi-phase flow (see Chapter 6). Also a process cost projection for the emulsion mediation is provided (see Chapter 7). Figure 2.1 The general principle of L / L emulsion mediated electroreduction of O2 to H2O2 (written for acidic conditions). 3 Background and Literature Review 11 C H A P T E R 3: Background and Literature Review 3.1 Fundamentals of Oxygen Electroreduction to Hydrogen Peroxide 3.1.1 Oxygen electroreduction in aqueous electrolytes Both the 4e" and 2e" reduction of O2 are among the most important of electrochemical reactions, being relevant for many and diverse areas of research such as: bioelectrochemistry, corrosion, fuel cells and electrochemical synthesis. The 4e~, 'water' pathway*, is generally predominant on certain noble-metal electrodes (e.g. Pt, Pd, Ag) , metal oxides such as perovskites (e.g. SrTi04) and on certain transition metal macrocycle electrocatalysts (e.g. Fe - tetrasulfonated phtalocyanine). On the other hand, the 2e" 'peroxide' route is usually obtained on carbon and graphite electrodes, A u , Hg , N i and some transition-metal macrocycle electrocatalysts (e.g. Co -tetrasulfonated phtalocyanine) [Kinoshita, 1992]. The electrochemical reactions relevant to the electroreduction of O2 to H2O2 are summarized in Table 3.1. Table 3.1 The reactions and standard electrode potentials for O2 and H2O2 electroreduction in alkaline and acidic conditions [Sawyer, 1991; Hoare, 1985]. (02 at 0.1 MPa) Electrolyte Overall Electrode Reaction E° ( V S H E ) A L K A L I N E (pH=14) 1) 0 2 + H 2 0 + 2e~ H02" + OH" -0.065 2) H02" + H 2 0 + 2e"^30ir 0.87 A C I D (pH = 0) 3) 0 2 + 2 H + + 2 e " ^ H 2 0 2 0.70 4) H 2 0 2 + 2 H + + 2e~ 2H 2 0 1.76 * Might involve adsorbed peroxide species as intermediates, but does not lead to peroxide in the bulk solution. 3 Background and Literature Review 12 As shown by E° in Table 3.1, the H2O2 reduction is thermodynamically favoured over the reduction of O2 for both alkaline and acidic conditions (compare eqns. [1] - [2] and [3] - [4]). Moreover, the thermodynamic criterion favours electroreduction of O2 in acid vs. alkaline conditions (eqn. [3] vs. eqn. [1]). However, the eletroreduction rate on materials such as graphite and other carbons, is higher for O2 than for H2O2, particularly in alkaline solutions. The mechanism of O2 electroreduction is dependent on many interacting factors besides the electrolyte composition, such as the bare electrode surface properties and morphology, O2 adsorption mode, presence of surface active additives, etc. For the electrode materials most relevant for the present work, i.e. graphite and glassy carbon, the accepted mechanisms for the alkaline 2e" reduction of O2 are: [Bockris and Khan, 1993; Kinoshita, 1992] Graphite O2, (bulk) —*• O2, (ads) 02,(ads)+ e~ —• 02~',(ads); (rate determining step) 2 0 2 ~ \ (ads) + H 2 0 0 2 + H 0 2 " (bulk) + O H " [3.1] [3.2] [3.3] Glassy Carbon O2, (bulk) —*• O2, (ads) 0 2 , (ads) + e —• [02, (ads)] [3.4] [3.5] [02, (ads)] * —>• O2 ", (ads) [3.6] 02 _', (ads) + H2O —• H 0 2 ' , (ads) + O H " ; (rate determining step) HO2", (ads) + e —• HO2 , (ads) HO2 , (ads) —* HO2 , (bulk) [3.8] [3.7] [3.9] where [O2, (ads)] " and O2 ", (ads) are two different forms of superoxide adsorbed on the electrode surface. 3 Background and Literature Review 13 It is important to note the paramount importance of the initial adsorption mode of O2 (eqn. [3.1] and [3.4]) in determining the 2 or 4e" pathway of the overall reduction. The 2e" route on graphite and glassy carbon (eqns. [3.1] - [3.9]) requires that the O2 molecule adsorbs in an end-on position (the so-called Pauling model) whereas the 4e" pathway is obtained in the case of lateral adsorption (i.e. either Griffiths or Bridge model) (Fig. 3.1) [Kinoshita, 1992]. Pauling Griffiths Bridge ( 2 e ) ( 4 e ) ( 4 e ) Figure 3.1 Models for O2 adsorption and their influence on the reduction pathway. To illustrate quantitatively the effect of p H and electrode material on the 2e" reduction of O2, a few representative exchange current density values are given in Table 3.2. Table 3.2 Selected standard exchange current densities for O2 reduction to H 2 0 2 (25 °C) (based on [Kinoshita, 1992]). Electrode Electrolyte io (A m2) Carbon 0.1 M K O H 10-5 A u 0 . 1 M K O H 4x10"6 A u 0 . 5 M H 2 S O 4 IO"9 The beneficial effect of OH" on the O2 reduction is explained by the formation of a chain of hydrogen bonds between adjacent adsorbed species, i.e. O2 (Pauling model) and OH", respectively, and it is referred to as the 'joint pseudosplitting / peroxide' mechanism [Kinoshita, 1992]. 3 Background and Literature Review 14 The electroreduction of O2 to produce high concentrations of H2O2 (e.g. over 0.1 M ) at p H below ca 12 (and especially at p H < 7) presents interesting challenges for both fundamental and applied electrochemistry research. Pletcher and co-workers obtained up to only 20 m M H2O2 at p H ca 2, during O2 reduction on reticulated vitreous carbon (60 ppi) in a flow cell operated at atmospheric pressure. The current efficiencies were between 16 and 69% depending on the electrolytes used (e.g. NaCl , Na2S04) and the cathode potentials applied (i.e. -400 to -900 m V / S C E for superficial current densities in the range of 52 to 340 A m"2). The resulting peroxide solution was used for in-situ removal of organics from waste streams such as formaldehyde, phenols and related compounds [Alvarez-Gallegos and Pletcher, 1998 and 1999; Ponce de Leon and Pletcher, 1994]. To enhance the O2 electroreduction to H2O2 at low p H (i.e. < 7) previous research has mainly focused on either electro catalysis by transition metal macrocycles and surface adsorbed quinone derivatives or electrochemical mediation by bulk quinone compounds. Studies involving various transition metal macrocycle electrocatalysts for O2 reduction are abundant in the literature and a complete review of this subject is beyond the present scope. Generally, cobalt macrocycles are most likely to electrocatalyze the 2e" O2 reduction. Recent research looked at compounds such as electropolymerized cobal - tetra(o-aminophenyl)porphyrin (poly(CoTAPP)) [El Mouahid et al, 1997] and cobalt - tetrasulfophtalocyanine (i.e. vitamin B12) [Zagal et al., 1997]. Unfortunately, there are several difficulties associated with these electrocatalysts, e.g. loss of stability in the long term, strong p H and electric potential dependence (e.g. vitamin B12 switches from the 2e" to the 4e" pathway at acid pH) and challenges in preparation of the catalyzed electrode (especially for potential large scale use). Surface-bound quinone or anthraquinone derivatives represent a potential alternative to transition metal macrocycle electrocatalysts (eqn. 3.10) [Wrighton, 1986]. Q H 2 (surface) + 0 2 -* Q (surface) + H 2 0 2 . [3.10] Considerable research effort was directed toward the 'smart design' of various quinone based electrocatalysts, in order to improve their surface functionalization characteristics (e.g. surface attachement, long term stability, etc.) [Wilson, 1996]. Typically, the performance of diverse carbon electrodes modified by surface attached anthraquinone derivatives, was much 3 Background and Literature Review 15 better in alkaline (e.g. p H > 8) than in acidic solutions. For instance, the p H dependence of the rate constant k for eqn. [3.2], in the case of a polymeric anthraquinone film adsorbed on glassy carbon, was expressed as [Degrand, 1984]: log(k) = 0.62pH + 0.7, where k: (1 mol" 1 s"1). [3.11] In fact, according to Degrand [1984], the catalytic current for the 2e" reduction of O2 became negligible at p H < 2. ' Interestingly, an electrochemical process using ca 0.05 m M of an anthraquinone derivative to catalyze the O2 reduction in 2 M N a O H , has been patented and is employed industrially in China [Yuan and Yuan, 1996; Yuan and Qi , 1997]. According to the patent, the catalyst is deposited in-situ on the cathode (from the bulk of the N a O H solution), improving the current efficiency (e.g. > 90%) while lowering the energy consumption (i.e. lower cell voltage). The last approach for the enhancement of O2 reduction to H2O2, is based on bulk mediation by either organic or water soluble anthraquinone compounds. A separate section is devoted to the organic mediation due to its relevance to the present work (see Section 3.2.3). There are a few studies indicating the possibility of homogeneous aqueous mediation of O2 electroreduction to H2O2 by water soluble anthraquinones such as 2,6 or 2,7 - anthraquinone disulfonates [Keita and Nadjo, 1983; Kelsall and Thompson, 1993]. The solubility of 2,6-anthraquinone disulfonate at p H 2 is 0.1 M , therefore, fairly high current densities could be applied for the bulk reduction of the anthraquinone mediator. Unfortunately, this route was not adequately researched from the point of view of high (e.g. > 0.1 M ) peroxide concentration electrogeneration. One of the disadvantage of using water soluble mediators lies in the separation and recovery of peroxide form the anthraquinone solution (e.g. precipitation of H2O2 as Ca02 using Ca(OH)2 and subsequent recovery of H2O2 by reaction of Ca02 with CO2). 3 Background and Literature Review 16 3.1.2 The effect of surfactants on the electroreduction of O2 Classic polarography studies noted that several surfactants (e.g. gelatin, lauric acid, sulfonaphthylstearic acid) even in very low concentrations (e.g. 5x10"4 % w ) suppressed the polarographic wave corresponding to the 2e" reduction of O2 to H2O2 [Kolthoff, 1952]. More recent investigations showed that surfactants such as quinoline, inhibit selectively the polarographic reduction of O2 in 1 M N a O H , by blocking the second electron transfer to yield the superoxide ion (02~-) as the main product [Chevalet et al, 1972]. Also, it was reported that self-assembled monolayers of octadecylmercaptan on A u , are able to influence the overall O2 reduction mechanism at p H 8.3. A t electrode potentials more positive than -0.8 V S H E , the O2 reduction proceeded by the 2e" pathway. However, for potentials more negative than -0.8 V S H E , the 4e" pathway prevailed. Disorders in the monolayer structure at very negative potentials, was proposed to be the cause of change in the overall reduction mechanism [Vago et al, 1995]. Furthermore, in the case of O2 reduction in alkaline solutions on gas diffusion electrodes, it was found that certain non-ionic surfactants used to hydrophobize the carbon black cathode, can be detrimental to the overall peroxide electrosynthesis by increasing the rate of the electrochemical decomposition of H2O2 [Chaenko etal, 1996]. Based on the above review, it can be concluded that the existent literature on the effect of surfactants is relatively scarce and is not comprehensive enough, since none of the above studies carried out a systematic investigation of the main classes of surfactants (i.e. cationic, non-ionic and anionic). 3.1.3 Oxygen electroreduction in organic media In addition to aqueous electrolytes, the electroreduction of O2 in organic media can play an important role in one of the electrochemical systems under investigation (see Section 2.2). Therefore, a brief review of this subject is warranted. Usually, in aprotic media two O2 electroreduction products are detected, i.e. superoxide O2 * and peroxide O2 . There seems to be a consensus in the literature that the formation of superoxide (eqn. [3.12]) is a reversible process with a formal potential of ca -0.63 V S H E , which is generally independent of the electrode material (e.g. various carbons, Hg) [Radyushkina et al, 1992] whilst it is slightly dependent on the solvent used (e.g. acetonitrile, dimethylsulfoxide, 3 Background and Literature Review 17 dimethylformamide) [Vasudevan and Wendt, 1995]. Peroxide, however, is usually obtained as an irreversible wave (eqn. [3.13]), at potentials more negative than -1.5 V S H E [Nekrasov and Vykhodtseva, 1995]. 0 2 + le~<=>02~' 0 2 ~ ' + le~ 0 2 [3.12] [3.13] In addition to the electrochemical reactions [3.12] - [3.13], in aprotic media with supporting electrolyte present, both superoxide and peroxide can be involved in ion-pairing processes with the cations of the supporting electrolyte, such as quaternary ammonium ions (quat+). Furthermore, it was found that the size of the quaternary ammonium ion has a strong influence on the potential at which reaction [3.13], i.e. peroxide formation, occurs. The larger the size of the cation (e.g. tetrabutylammonium, T B A vs. tetramethylammonium, T M A ) the more negative the half-wave potential for 02~* / 0 2 2 ~, e.g. -2.4 V S H E for T B A vs. -1.5 V S H E for T M A , respectively [Nekrasov and Vykhodtseva, 1995]. The presence of a proton source in the organic solvent, plays a major role as well, in the mechanism of 0 2 reduction. In addition to superoxide (02~"), species such as H 0 2 ~ and H 0 2 * are involved in the overall reduction mechanism when protons are available [Radyushkina et al, 1992]. 3 Background and Literature Review 18 3.2 Electrochemical Phenomena involving Liquid / Liquid Emulsions This section provides an overview of certain aspects of electrochemical processing o f L/L emulsions, with regard to the emulsion mediated hydrogen peroxide electrosynthesis. 3.2.1 Emulsions and surfactants: General aspects The colloidal system produced by the dispersion o f two mutually insoluble (or little soluble) liquids, is called an emulsion. From a thermodynamic stability point o f view, one can distinguish between micro and macro emulsions. The former (i.e. microemulsion) is an isotropic dispersion of usually very small droplets {ca 10 nm), thermodynamically stabilized by amphiphilic additives (e.g. one or more surfactants). Macroemulsions on the other hand, are unstable (i.e. the phases separate or can be induced to separate), and typical droplet sizes are orders of magnitude larger than in the case of microemulsions. Surfactants might be added to enhance emulsification and improve the macroemulsion stability. Emulsions can be classifieds also based on their intimate structure, i.e. oil-in-water (O/W) where the organic phase is dispersed as droplets in the continuous aqueous phase; water-in-oil (W/O) characterized by a continuous organic phase with aqueous droplets dispersed in it and finally, so-called bicontinuous (or dual) emulsions, which display a complex structure composed of interconnected organic and aqueous conduits (Fig. 3.2). The latter type is usually observed in the case of microemulsions [Evans and Wennerstrom, 1994]. Figure 3.2 The bicontinuous emulsion structure [Evans and Wennerstrom, 1994J. 3 Background and Literature Review 19 The type of the emulsion formed when an organic and aqueous phase are brought into contact in the presence of an emulsifying agent, depends on several factors such as, phase volume ratio, surfactant type, architecture and concentration, characteristics of the organic and aqueous phases (e.g. ionic composition of the aqueous electrolyte), presence of impurities (e.g. finely divided solid particles), mixing characteristics, temperature etc. Therefore, the accurate prediction of the emulsification process based solely on theoretical considerations can easily lead to false conclusion and experimental verification (e.g. by conductivity measurements) is required. Among the factors that influence the emulsification process, the characteristics of the surfactant have a paramount importance. A n approximate, but nevertheless useful model, that allows the prediction of the type of surfactant aggregate and hence, the nature of the emulsion formed, is based on the surfactant parameter (or surfactant number) Ns [Evans and Wennerstrom, 1994; Rusling, 1994 II], i.e. where Vhc volume of the hydrophobic portion of the surfactant molecule (nm 3), Lhc length of the hydrocarbon chain (nm), ah effective area per head group (e.g. for N + in the case of quaternary ammonium salts), Nhc number of hydrocarbon chains (strands) in the surfactant structure, nc and ncH3 number of carbon atoms and methyl groups per chain, respectively. The surfactant number (eqn. [3.14]) relates the properties of the individual amphiphilic molecule to the preferred arrangement in solution of the aggregates (e.g. micelles*) and furthermore, to the most likely emulsion structure formed between two immiscible liquids with the surfactant present (Table 3.3). Figure 3.3 illustrates some representative surfactant aggregates as a function of Ns-V, ; with Vhc = OmiNhc(nc + nCH3)andLhc = 0.15 + 0 . 1 2 7 « c ; [3.14] * At surfactant concentrations above the critical micellar concentration (cmc) value. 3 Background and Literature Review 20 Table 3.3 Surfactant number, preferred surfactant aggregate and corresponding emulsion type (based on [Evans and Wennerstrom, 1994; Rusling, 1994 II]). Surfactant number Ns Aggregate type Preferred emulsion structure <0.33 Spherical micelles Oil-in-Water (OAV) 0.5 Cylindrical micelles Oil-in-Water (OAV) «1 Bilayers; Vesicles Bicontinuous (microemulsions) >1 Inverted micelles and cylinders Water-in-Oil (W/O) Figure 3.3 Representative surfactant aggregates: a) Spherical micelle; b) Inverted micelle; c) Planar bilayer; d) Bicontinuous structure. ® Head group; — Hydrophobic chain. [Evans and Wennerstrom, 1994]. 3 Background and Literature Review 21 In the surfactant number equation (eqn. [3.14]), the area per head group ay,, is the most sensitive variable (for ionic surfactants especially). It is strongly dependent on the electrolyte composition and concentration, since the balance of electrostatic forces involving the ionic surfactant and the counter-ions present in the electrolyte, wi l l ultimately determine whether the surfactant can or cannot form aggregates with closely packed head groups such as bilayers, vesicles or inverted micelles. 3.2.2 Surfactant adsorption at interfaces involving a solid phase For electrochemical systems, the surfactant adsorption on solid surfaces can play major roles in a large variety of cases such as corrosion, electrometallurgy, electroanalysis, electrosynthesis, electrocatalysis and electrokinetic phenomena [Mackay, 1992]. There are a number of excellent reviews on various issues of surfactant adsorption relevant to electrochemical phenomena, by Rusling [1994 I, II; 1991], Lipkowski [1992], Ingram and Ottewill [1991], Fuerstenau and Herrera-Urbina [1991], to mention only a few. Here, only those aspects wi l l be highlighted which are relevant to the systems under investigation. a) Surfactant Adsorption at the Electrode /Electrolyte Interface According to Traube's qualitative rule [Shaw, 1992] the longer the hydrocarbon chain length Lhc (eqn. [3.14]) the greater is the tendency of a surfactant to adsorb at various interfaces (e.g. solid / liquid, gas / liquid, etc.). However, for the electrode / electrolyte interface the situation is more complex. The adsorption of surfactants at electrified interfaces depends on a number of factors such as electrode potential, surfactant type and concentration, the ionic strength of the electrolyte and pH. A typical adsorption isotherm for an ionic surfactant on an oppositely charged surface is comprised of three regions (Fig. 3.4) [Rusling, 1994 I; Myers, 1988]. In region I, at very low bulk concentrations (i.e. much below cmc) the surfactants adsorb as individual ions, with no lateral interaction, replacing some of the native counterions from the electrode surface (so-called 'ion-exchange' mechanism). A s the bulk surfactant concentration is increased (region II), lateral, hydrophobic interactions among the chains become more significant, adsorption increases sharply with concentration and surface aggregate formation begins ('surface ion-pairing' 3 Background and Literature Review 22 mechanism). A s established by experimental studies, ionic surfactants adsorb on oppositely charged surfaces in a variety of arrangements (e.g. 'head-down', 'chain-down') accompanied by tail-to-tail interactions (especially for long and multiple chain surfactants), leading to the formation of planar bilayers or various spherical or cylindrical surface aggregates on patches of the surface [Marine and Gaub, 1995]. These surface structures are sometimes referred to as admicelles or hemimicelles. Increasing further the bulk surfactant concentration (region ELT, Fig. 3.4), ultimately a complete admicellar surface coverage is achieved, which can induce profound changes at the electric double-layer level, e.g. surface charge reversal, changes in surface p H and potential, with major consequences on the electrode reaction. 1 fj"5 ~q 1 — i — i i i | 1 — i — i i i | 1 — i — i i i | 1 — i — T T T 10-10 - | | 10-7 1 0 - 6 10-5 10-4 10-3 Surfactant bulk concentration / M Figure 3.4 Generic isotherm for ionic surfactant adsorption on oppositely charged surfaces (based on [Rusling 1994 I]). 3 Background and Literature Review 23 In the case of electrode surfaces the surfactant adsorption isotherm and correspondingly the physical state of adsorbed surfactant molecules are a strong function of the electrode potential. Generally, the surface concentration of non-ionic surfactants is the highest at the potential of zero charge (Epzc). Cationic surfactants on the other hand, are more strongly adsorbed at negative potentials vs. E p z c , whilst anionic surfactants adsorb at potentials positive to Epzc [Banetal., 1998]. Furthermore, Burgess et al [1999] presented direct visual evidence (i.e. atomic force microscopy and scanning tunneling microscopy images) for the potential dependent surface phase transition of an anionic surfactant (i.e. sodium dodecyl sulfate, SDS) on a single crystal gold electrode, i.e. from half-cylindrical admicell (predominant at potentials between -200 and + 300 m V vs. SCE) to condensed structureless film (occurring at potentials above +400 m V vs. SCE). The change of surfactant surface structure with the applied electrode potential can have profound consequences on the electrode kinetics of the reactive species of interest. For instance, following the example of SDS on A u electrode, at potentials where the half-cylindrical surface aggregate prevails, due to the extensive lateral adsorption of the hydrocarbon chain on the electrode surface [Burgess et al, 1999], the surfactant might induce the so-called 'blocking effect' by minimizing the access of the electrochemically reactive species to the active surface sites. A t very positive potentials on the other hand, a fluid-like film is formed on the surface, causing a different local environment for the electrochemical transformation of the reactive species. b) Wetting and Related Phenomena The good coverage of the electrode surface by the reacting fluid (i.e. liquid or gas) is important for all electrochemical reactions, but of particular significance in the electrochemical processing of multi-phase fluid systems (e.g. L / L , L / G or L / L / G ) . In the latter case, usually a selective (or preferential) coverage by one of the phases is desired. Surfactants, due their inherent interfacial adsorption properties, can be used to manipulate the fluid wetting and spreading characteristics of the electrode surface. 3 Background and Literature Review 24 a) Poor wetting by phase 1 b) Improved wetting by phase 1 c) Spreading of phase 1 Figure 3.5 Three-phase contact between a solid surface (S) and two fluid phases: 1 (e.g. organic) and 2 (e.g. aqueous). Schematic representation of the potentially beneficial effect on the surface wetting by phase 1 induced by surfactant adsorption. 9 -contact angle. -4 in Fig. 3.5 c, indicates the spreading pressure of fluid 1 at the S,2 interface. The three-phase equilibrium involving a solid surface (S) and two fluid phases 1 and 2, respectively, is described by Young's equation [Blake, 1984]: cos0 = r s ' 2 ~ r s - 1 , [3.15] where 9 contact angle (deg.), ys,2, Ys,i and y u interfacial tensions (N m"1) between the corresponding phases (see Fig. 3.5). Improved wetting of the solid surface by one of the phases (e.g. phase 1) means a lowering of the contact angle 6. From Young's equation, addition of a surfactant wil l lower # i f either ys,j or y u (or both) are reduced. Assuming that 1 is organic and 2 is aqueous phase, thus the surfactant must lower the solid/organic and organic/aqueous interfacial tensions. Such a condition could be realized by using a strongly organic-soluble surfactant (see further). The general relationship between the contact angle and surfactant concentration, is obtained by differentiating eqn. [3.15] with respect to lnc, and applying the Gibbs equation [(dy / d lnc)T,p = -RT^Twi th /"surfactant surface concentration)]. Thus the contact angle 6 -surfactant concentration c, expression is given by [Blake, 1984]: 3 Background and Literature Review 25 r,,2 sin e ( d9 \ RT(Fs,2 ~ TS,l ~ ri,2 C O S 0 ) , [3.16] \d\RCj where ris the surfactant surface excess (mol m"2) at the S,2; S , l and 1,2 interfaces respectively. Based on eqn. [3.16], improved wetting by phase 1 with increasing surfactant concentration corresponds to the condition (dQ I d lnc) < 0; which means that: i.e. the surfactant should preferentially accumulate at the solid/organic and organic/aqueous interfaces. The ultimate form of selective surface coverage is the complete wetting of the surface by one of the phases (i.e. zero contact angle), which implies the spreading of e.g. phase 1, and total displacement of phase 2 from the solid surface (Fig. 3.5 c). For porous materials, the presence of a surfactant can improve the selective penetration and wetting of the pores by one of the fluids, by lowering the required capillary pressure. Using the Laplace equation the pressure drop across the hemispherical meniscus within a capillary of radius R is [Blake, 1984]: Furthermore, the penetration depth by phase 1, xp, can be obtained by combining eqn. [3.18] with Poiseuille equation (i.e. for laminar flow within the capillary), giving Ts,2 <Ys,i + r u cos#, [3.17] Rtyx 2cos0 [3.19] where / time (s), /n dynamic viscosity of the fluid mixture (Pa s), whilst the rest of the variables have the same meaning as before. 3 Background and Literature Review 26 Eqn. [3.19], also known as the integrated Washburn equation, shows that the 'driving force' for capillary penetration by phase 1 is the term yijcos 0. Thus, taking into account Young's equation also (eqn. [3.15]), only surfactants which adsorb at the S , l interface wil l be effective in increasing the penetration depth of phase 1. Surfactants that adsorb only at the fluid/fluid (1,2) interface (i.e. reduce yli2), although they lower the contact angle according to Young's equation, wi l l not improve the capillary penetration of phase 1 since the term Xucos 6, remains more or less unchanged [Blake, 1984]. 3.2.3 Charge and mass transfer in emulsions: Applications for electrochemical synthesis In electrochemical processing of organic compounds an aqueous / organic (L /L ) emulsion is typically employed in two cases: a) to increase the ionic conductivity of the system and reduce the energy consumption (e.g. electroreduction of acrylonitrile [Baizer, 1979]; electrooxidation of benzene in 1 M H 2 S 0 4 [Oloman, 1980]) b) for mediated electrosynthesis involving inorganic, water soluble, redox couples (e.g. Ce(IV)/Ce(III) or Mn(III)/Mn(II) for redox catalyzed electrooxidation of aromatics [Simonet, 1991]). For both cases the addition of surface active agents to the emulsion can have diverse effects, besides the general aspects of emulsification and wetting, such as increased water solubility of otherwise sparingly soluble organics (e.g. in the presence of McKee salts), improved mediator transfer across the organic / aqueous interface (e.g. phase-transfer catalysts) possibility of micellar catalysis and/or favorable changes in interfacial properties (e.g. increase of surface p H for acrylonitrile electroreduction). Feess and Wendt [1981, 1980], studying the charge transfer in O / W emulsions, established a set of conditions that are useful to be followed in order to obtain adequate yields and current efficiencies for the electrochemical conversion of a given organic compound dissolved in the organic phase. The underlying concept was to find conditions, which favor the current passage through the organic phase instead of the aqueous solution (see Fig. 3.6 as well). These conditions are summarized below: 3 Background and Literature Review 27 a) The organic phase should posses a relatively high electrical conductivity, which could be accomplished by the addition of appropriate lipophilic and non-extractable supporting electrolyte (e.g. certain quaternary ammonium cations with CIO4" as counter ion). b) A s low as possible contact angle and aqueous / organic interfacial tension. In other words, good wetting of the electrode surface by the organic phase*. c) The charge from the electrode / organic phase boundary must be compensated by an efficient counter-ion transfer (e.g. H + , or other cation) across the organic / aqueous interface (Fig. 3.6). Due to the different ionic mobilities across the L / L interface, a potential difference develops at the aqueous / organic phase boundary. Figure 3.6 Schematic representation of the electrochemical reduction of an organic compound at the organic / electrode boundary in a two-phase L / L emulsion (modified after [Feess and Wendt, 1981]). The intricate mass transfer in multi-phase systems further complicates the electrochemical processing of emulsions, especially in the case of three-dimensional porous electrodes. Generally, there is very little quantitative information regarding the organic liquid to solid mass transfer in electrochemical reactors operating with L / L or L / L / G multi-phase For the special case of porous electrodes, the solid / organic interfacial tensions is more important (see Section 3.2.2). 3 Background and Literature Review 28 electrolytes. Furthermore, all of the existent information was obtained for O A V emulsions, which might not be valid for the other types of emulsion. Wendt and co-workers were the first to propose two different mechanisms for mass transport to electrodes from two-phase, OAV, emulsions [Dworak and Wendt, 1977; Dworak et al, 1979]. One of the mechanisms refers to the mass transfer of an organic species from the organic droplet directly to the electrode surface (i.e. 'wetting' mechanism), whilst the other one is concerned with the effect of the organic droplets on the aqueous liquid to solid mass transfer (i.e. 'boundary layer disturbance' mechanism). In the latter case the organic droplet is electrochemically 'inert'. A special case occurs when the mass transfer is related to a species that is soluble and 'reactive' in both phases (i.e. organic and aqueous), e.g. liquid to solid mass transfer of O2 or I 2 in aqueous / organic emulsions. This particular case is referred to as the 'extraction' mechanism [Fenton and Alkire, 1988]. The term 'extraction' mechanism reflects the idea that for a mutually soluble species, the organic droplets can enhance the overall mass transfer without a direct contact between the droplet and the solid surface, by injecting reactant molecules in the thin aqueous boundary layer very close to the electrode surface. In the emulsion system under investigation in the present work (Section 2.2), the liquid to solid mass transfer of 2-ethyl anthraquinone (which dissolves exclusively in the organic phase) is characterized by the 'wetting' mechanism, whilst the organic liquid to solid O2 mass transfer could be treated as an 'extraction' mechanism. Naturally, in addition to liquid - solid mass transfer, the multi-phase system involves other steps as well such as GPL and L / L mass transfer. For O A V emulsions, it was found that highly turbulent conditions are unfavorable for the 'wetting' mechanism. A t high emulsion velocity the probability and duration of direct contact between the organic droplets and the solid surface is diminished. Therefore, the organic liquid to solid mass transfer rate could be reduced at high emulsion velocity, especially when the organic phase is less dense than the aqueous component [Dworak and Wendt, 1977; Dworak et al, 1979]. In a mass transfer study of nitrobenzene electroreduction on mesh electrodes from emulsions formed with 1 M H2SO4, it was observed that, contrary to the case of one-phase flow, for the two-phase system the increase of the catholyte flow rate did not increase the nitrobenzene reduction current [Letord-Quemere, et al, 1988]. Unfortunately, there are no quantitative 3 Background and Literature Review 29 correlations for the direct organic liquid to solid mass transfer from L / L emulsions ('wetting' mechanism). For the 'extraction' mechanism, based on the reduction of I 2 on reticulated vitreous carbon ( R V C 45 ppi) from toluene / aqueous emulsions, Fenton and Alkire [1988] observed the following trends: a) The mass transfer rate was up to one order of magnitude higher than the rate corresponding to one-phase (aqueous) flow. b) The overall mass transfer rate increased with increasing catholyte flow rate. c) For given flow rates, the 'reactive' two-phase mass transfer rate increased sharply with increasing organic phase volume fraction. For instance, for the same Reynolds number and droplets smaller than the pore size of R V C , a volume fraction increase from 20% to 30%), brought about a five times increase of the Sherwood number. d) The smaller the droplets the higher the mass transfer rates. Furthermore, the same authors also investigated the effect of surfactants such as Triton X-100 (non-ionic) on the 'reactive' two-phase mass transfer. They found that the 'extractive' mass transfer rates were generally higher in the presence of surfactant, presumably due to reduced droplet size as a result of lower L / L interfacial tension. However, contrary to the case when Triton X-100 was absent, an increase of the superficial velocity brought about a slight decrease in the mass transfer coefficient with surfactant present (similar to the 'wetting' mechanism discussed earlier). 3 Background and Literature Review 30 3.3 Electroreduction of Anthraquinone Derivatives in L / L Emulsions Although the electroreduction of various quinones dissolved in diverse organic solvents has been extensively studied [Chambers, 1988], the two-phase, organic / aqueous, electroreduction has received far less attention. Knarr et al. [1992] were the first to report on the electrochemical reduction of a quinone derivative (i.e. 2-ethyl anthraquinone, E t A Q ) employing an emulsion. The goal of their study was to investigate the feasibility of replacing the catalytic step in the 'thermochemical anthraquinone' peroxide process (Chapter 1) with an electrochemical route. Consequently, this study did not aim at the simultaneous in-situ generation of H2O2. A n O A V emulsion was employed, composed of up to 52% by volume organic phase dispersed in either 1 or 2 M N a O H . However, a 40% organic volume ratio was considered to be the best for preparative purposes. In selecting the proper organic phase composition, Knarr et al. were targeting organic solvents with the following characteristics: high solubility of reactant and product, i.e. 2-ethyl anthraquinone and 2-ethyl anthrahydroquinone, low water solubility and good electrochemical stability. The solution chosen was a (85/15) v/ v mixture of diethylbenzene (DEB) and tributylphosphate (TBP), containing 0.42 M E t A Q (i.e. saturation concentration at room temperature). A N2 atmosphere was maintained in the cathode compartment to avoid any oxidation of the product E t A Q H 2 . The preferred cathode was Pb, employed either in a rotating cylinder arrangement or electrodeposited on reticulated vitreous carbon (20 ppi, R V C ) in a 'flow-through' cell. For both cells, the cathode and anode compartments were separated by a Nafion membrane. Generally, the applicable current densities were lower for the rotating cylinder than for the flow cell, i.e. up to 210 A m"2 vs. 1750 A m"2, respectively. Unfortunately, current efficiency data were reported only for the rotating cylinder experiments, where it was found that the current efficiency dropped significantly as the current density was increased to the mass transfer limited value of 280 A m"2. For instance, at 140 A m"2 the current efficiency was 99% while at 210 A m"2 (i.e. 75% of the limiting value) it dropped to 88% [Knarr et al, 1992]. Recently, Huissoud and Tissot [1999 I and II] extended the work of Knarr et al, by investigating the possibility of in-situ generation of H2O2 in 2 M N a O H using the same organic composition, i.e. DEB(85%) V / TBP(15%) V with 0.42 M E t A Q . The preparative experiments were performed in a divided flow cell with co-current upward gas (O2) - liquid (emulsion) flow, 3 Background and Literature Review 31 and continuous recycle of the emulsion catholyte. In this work the cathode was R V C (20 to 45 ppi) instead of Pb. The maximum H2O2 concentration was 0.8 M in 2 M N a O H , obtained after 6.5 hours of continuous catholyte recycle using the 20 ppi R V C cathode operated a superficial current density of 225 A m"2. It was concluded that in order to generate high H2O2 concentrations the volume fraction of the organic phase must be as high as possible while still the aqueous electrolyte must remain the continuous phase, therefore a 40% organic ratio was proposed Also, it was found that the current efficiencies for the one-step in-situ generation of H2O2 (i.e. with simultaneous 0 2 and liquid flow) were always higher than for the two-step case, where first EtAQH2 is obtained in a N2 atmosphere and then is subsequently oxidized in a separate step [Huissoud and Tissot, 1999 II]. Most likely the direct 2e~ O2 reduction in 2 M N a O H , contributes as well to the overall higher peroxide concentration and current efficiency for the on-step process. As outlined above, the work of Huissoud and Tissot proved the concept of in-situ E t A Q mediated H2O2 electrosynthesis according to the scheme given in Fig. 2.1. However, their emulsion process has several disadvantages, hindering its applicability. Some of the problems are: a) The process was tested only in 2 M N a O H , in which O2 can be reduced efficiently by direct 2e" reduction on carbon electrodes without an organic phase (e.g. the industrial Dow-Huron process, Fig. 1.2). Thus, using an emulsion system, which adds extra engineering and operation difficulties, in this author's opinion is unjustified for H2O2 electrosynthesis in very alkaline electrolytes such as 2 M N a O H . b) Ultimately, the peroxide concentration obtained with the emulsion process was lower than the one typically generated by the Dow-Huron technology in 2 M N a O H , i.e. 0.8 M H 2 0 2 vs. ca 1.25 MH2O2. c) The superficial current densities are very low and industrially impractical i.e., equal or less than 225 A m"2. d) There is very little information on energy requirements (i.e. cell voltage) which is crucial for the industrial success of an electrochemical process. e) The use of high quantities of diethylbenzene as the main component of the organic phase, an environmentally hazardous chemical (e.g. ethylbenzenes are carcinogenic and not 3 Background and Literature Review 32 biodegradable [seweb2.phillips66.com]), most probably renders this emulsion process unsuitable for on-site use by the pulp and paper industry. There are two other studies by the same authors, which looked at different organic phase compositions but in a homogeneous mixture with water. In one of them a mixture of D M E with 0.1 M tetrabuytlammonium hydroxide and different quantities of 0.2 M HCIO4 was employed [Tissot and Huissoud, 1996]. In the other study, the catholyte was a one phase mixture of 1,2-dimethoxyethane ( D M E ) with 0.1 M tetraethylammonium tetrafluoroborate ( T E A T F B , supporting electrolyte) and 5% H 2 0 at p H 10.6 [Huissoud and Tissot, 1998]. According to the authors, for preparative purposes the latter composition proved more successful, by giving up to 0.1 M H2O2 after 10 hours of batch electrolysis using a 60 ppi R V C at 5 A m"2. Mechanistic investigations were also performed to establish the role of proton sources in the electrochemical reduction of E t A Q . Generally, the mechanism for E t A Q electroreduction from an emulsion involves a number of successive electron transfer and chemical steps such as protonation and/or ion-pairing, depending on the composition of the aqueous and organic phases in contact. For the overall reaction: E t A Q + 2e" + 2H* -> E t A Q H 2 , the electron transfer and chemical steps (e.g. protonation) can be summarized based on the so-called 'square-scheme' concept, illustrated in Fig. 3.7 [Amatore, 1991]. 3 Background and Literature Review 33 E t A Q « 6 - E t A Q " " — ' E t A Q 2 ' E t A Q H E t A Q H E t A Q H E t A Q H . 2+ e" E t A Q H E t A Q H . Figure 3.7 The 'square scheme' concept applied to the electroreduction of 2-ethyl anthraquinone (EtAQ). Based on the characteristics of both the organic solvent (e.g. dielectric constant) and aqueous phase (e.g. pH) the E t A Q reduction mechanism can occur virtually along any of the routes depicted in Fig. 3.7. For instance, in aprotic organic solvent in contact with a highly alkaline aqueous phase (e.g. 2 M NaOH), the mechanism along the first line, leading to E t A Q 2 " (i.e. EtAQNa2), is the most likely pathway. In acidic emulsions on the other hand, different protonation steps accompany the electron transfer, generating protonated E t A Q species. The influence of protons on the E t A Q reduction mechanism, is also reflected in the p H dependence of the reduction potential. For a homogeneous mixture composed of H 2 0 and 5% dimethylformamide (DMF) , Revenga et al. [1994], constructed an E - p H diagram for anthraquinone (AQ) reduction (Fig. 3.8). Fig. 3.8 can be used for the estimation of the E t A Q reduction potentials as a function of pH. A s a general trend, the higher the p H the more negative the reduction potential of the anthraquinone compounds. 3 Background and Literature Review 34 5 6 7 8 9 10 11 12 13 14 PH Figure 3.8 E . / 2 - p H diagram for A Q reduction in a homogeneous mixture of H 2 0 - 5% v D M F . E1/2 is the half-wave potential, (adapted from [Revenga et al, 1994]). 3.4 Elements of Porous Electrode Theory Porous, three-dimensional electrodes are characterized by large specific electro-active surface area e.g., between 5 x l 0 2 to 5 x l 0 4 m 2 m"3. Therefore, they can considerably increase the space-time yield of an electrochemical process. The need to employ porous electrodes arises whenever the current density at which the electrochemical reaction can be efficiently operated on planar electrodes is low, e.g. less than 100 A m"2. Such a situation usually occurs when the reactant concentration is low (e.g. less than 0.1 M ) and/or the kinetics is slow. The 3 Background and Literature Review 35 electroreduction of O2 from acidic aqueous electrolytes (solubility ca 1 m M , slow electrode kinetics), is a very good example for the beneficial role of three dimensional electrodes. Based on the relative direction of electrolyte and electric current flow, the porous electrodes are classified in two categories (Fig. 3.9): a) 'flow-by' electrodes, the electrolyte and current flows are perpendicular to each other b) 'flow-through' electrodes, the electrolyte and current flows are parallel to each other. Electrolyte flow ' F l o w - b y ' a ) I mm i l i l l i l Electrolyte flow I ' F l o w - t h r o u g h ' b ) Figure 3.9 Classification of porous electrodes based on the relative direction of electrolyte and current flows. Figure 3.9 shows that in the 'flow-through' design the conversion of the reactant is along the bed thickness whilst in the 'flow-by' mode the conversion occurs along the length of the fixed bed electrode. While the thickness of the electrode is limited by the potential drop to maximum one or two centimeters, the length of the bed can be made, theoretically, as long as necessary to achieve a desired reactant conversion per pass. Therefore, to achieve high conversions per pass the 'flow-by' design is preferred for industrial electro synthetic applications [Walsh, 1993]. 3 Background and Literature Review 36 The porous electrode theory is based on the intricate interplay of electrode kinetics, electronic and ionic conductivity, fluid dynamics and mass transfer in porous media. These effects induce usually, non-uniform potential and current distributions in the 3-D matrix, yielding non-uniform reaction rates (Fig. 3.10). The rigorous quantitative analysis of porous electrodes requires mathematical modelling [Newman, 1991]. The extensive modelling endeavor produced a few easy-to-use design equations, which are useful in evaluating the effectiveness of a certain porous electrode for a given practical application. One such design parameter is the maximum electroactive bed thickness rmax, which is the maximum bed thickness that works at the mass transfer limited current density without significant contribution from secondary electrode reactions (e.g. H 2 evolution). Any physical bed thickness, which exceeds zmax is used less efficiently (Fig. 3.10). The TMAX is estimated as follows [Masliy and Poddubny, 1997]: a) Solid matrix conductivity KM » Electrolyte conductivity m x max V nFKmaC 0.5 [3.20] b) Solid matrix conductivity KM = Electrolyte conductivity KI T = max HK-l,eff^y max nFKaC [3.21] '\ ,.„ max In eqns. [3.20] - [3.21], Ki,eff effective liquid electrolyte conductivity (S m"), AVr> maximum allowable voltage drop to avoid significant side reactions (e.g., 0 . 1 - 0 . 2 V [Newman, 1991]), n number of electrons exchanged, F Faraday's constant (96,584 C mol"1), Km mass transfer coefficient (m s"1), a specific electrode surface area (m 2 m"3) and C reactant concentration (mol m"3). The effective liquid conductivity is given by the Neale - Nader equation [Oloman, 1979]: 3 Background and Literature Review 21 2K,S, heff ~ 3 - e , with eeff = smpl [3.22] where seff effective bed porosity, sm solid matrix porosity, /?/ liquid hold-up and m liquid (electrolyte) conductivity (S m"1). Once the maximum electroactive bed thickness rmax, is calculated, the superficial mass transfer limited current density is expressed by, In order to apply eqns. [3.20] - [3.23] the mass transfer coefficient Km, or the mass transfer capacity Km a, must be known. Unfortunately, for multi-phase (e.g. L / L ; L / G and/or L / L G ) flow in fibre bed electrodes (e.g. fibre diameter the order of tenths of jum), mass transfer correlations are scarce in the literature. The extrapolation of mass transfer correlations from the general chemical engineering literature obtained for packed beds with packing sizes the order of mm or cm is not warranted. For co-current upward gas (O2) - liquid (1 M NaOH) flow in a graphite fiber bed (fibre diameter ca 20 urn), Hodgson and Oloman [1999] established the following correlations for mass transfer capacity and liquid hold up, respectively: h =^m^nFKmaC. [3.23] Kma = 5.91° 3 7 2 G 0 2 3 2 , for 0.04 <G< 0.36 and 3 < L< 7.6 (kg nf 2 s"1) A = 1 ~ 0.907ZT0 3 6 2 G 0 3 0 1 , for 0 < G < 0.35 and 1.53 <L< 7.62 (kg i n 2 s_1), [3.25] [3.24] where G and L are the gas and liquid loads, Pi liquid hold-up. Analogous correlations for G / L / L flow are not available. 3 Background and Literature Review 38 feeder separator a) Potential distribution A E m - V l " J V^ - E m , 1 ~ E s , 1 V , = E m , 2 " E s , 2 - max b) Current distribution for the main reaction j A L c) Polarization curve main reaction i A secondary reaction i B v1 v2 - V Figure 3.10 Hypothetical potential and current distribution along the thickness of a porous electrode. Legend: Em, Es electric potentials of the matrix and electrolyte solution, respectively, V electric potential 'driving force' at the matrix / solution interface, h physical bed thickness, rmax maximum electroactive bed thickness, / local current density. Subscripts: A main reaction, B secondary reaction and L mass transfer limited. 4 Experimental Methods, Apparatus and Materials 39 C H A P T E R 4: Experimental Methods, Apparatus and Materials 4.1 Fundamental Techniques 4.1.1 Cyclic voltammetry For cyclic voltammetry experiments an Omni-90 (Cypress Systems Inc.) potentiostat was employed with the conventional three-electrode arrangement (Fig. 4.1). The working electrode was a 1 mm diameter glassy carbon (GC, Cypress Systems Inc.) disk. The counter electrode was a Pt wire and the reference a mini A g / A g C l electrode in saturated KC1. The working G C electrode was cleaned by polishing with 1 and 1/4 um diamond paste and 0.03 um alumina paste followed by sonication in methanol and double distilled water. Typically u A currents were measured on the working electrode (accuracy 0.15%), as a function of the applied potential (accuracy ±1 mV) and scan rate (maximum range 10"4 to 5 V s"1, accuracy 0.25%) [Cypress Systems, 1991]. In the aqueous electrolytes (i.e. 0.1 M Na2C03 and 0.1 M H2SO4) the cyclic voltammograms were recorded for O2 saturated conditions, maintaining also an O2 'blanket' above the electrolyte (at atmospheric pressure). For cyclic voltammetry studies involving the organic media, the glass cell o f the voltammetric equipment contained a bed of oven-dry molecular sieves (Davison® type 3A, Fisher Scientific) to minimize the water content of the solvent. A l l voltammograms were recorded at 295 K . 4.1.2 Spectrophotometry The absorbance spectra of various organic phase compositions were recorded using a Novaspec II® spectrophotometer (Pharmacia Biotech Inc.) using quartz Suprasil® cuves (Fisher Scientific Inc.). The spectrophotometer was connected to a P C and controlled by the Novascan® software (Pharmacia Biotech Inc.). The employed wavelength was between 400 and 900 nm with a scanning interval of 1 nm. A l l the absorbance spectra were recorded against distilled water as reference. 4 Experimental Methods, Apparatus and Materials 40 v F / X - Y Ref Working Counter 2 (or N 2 ) Figure 4.1 Cyclic voltammetry set-up. 1. Glass cell, 2. G C working electrode, 3. A g / A g C l reference electrode in satd. KC1, separated with a porous plug from the working electrolyte, 4. Pt counter electrode, 5. Potentiostat, 6. X - Y recorder, 7. two-way gas purging device. 4.1.3 Ionic conductivity measurements The ionic conductivities of both aqueous and organic electrolytes were measured using a Jenway® 4020 conductivity meter, equipped with a PCM141 Pt conductivity cell (cell constant 1.0). The conductivity meter and cell were calibrated against two commercial standard conductivity solutions, TDS 442-15 (23.4 | iS cm"1 / 20 °C) and T D S 442-30 (46.7 u\S cm"1 / 20 °C), respectively (Myron L Co.). The standard deviation of the measured vs. the standard conductivity was + 1 U.S cm"1 . 4 Experimental Methods, Apparatus and Materials 41 4.2 Electrosynthesis Apparatus For preparative electrogeneration of hydrogen peroxide, three different electrochemical reactors were employed, a batch glass 'FT- cell (150 ml total volume), and two continuous 'flow-by' electrochemical cells (trickle-bed), with 0.14 m and 0.5 m effective length, superficial area 33 cm 2 and 267 cm 2 , respectively. 4.2.1 Batch 'H'-cell The components of the 'H ' -ce l l (C-600, Electrosynthesis Inc.) and the corresponding experimental set-up are illustrated in Fig. 4.2. The glass 'H ' -ce l l is composed of the cathode, anode and reference electrode compartments. The cathode space (total volume: 150 ml; effective volume: 110 ml) is separated from the other two compartments by a porous frit (vs. the reference electrode side) and a cation exchange membrane (vs. the counter electrode compartment), respectively (Fig. 4.2). The tip of the reference electrode compartment protruded ca. 4 mm away from the back of the cathode. A saturated calomel electrode (SCE) immersed in KCl s atd., was used as reference. Oxygen gas was purged in the cathode compartment at atmospheric pressure during electrosynthesis (Fig. 4.2). The cathodes were different grades of reticulated vitreous carbon ( R V C ) (see Section 4.3), with an effective superficial surface area of ca 10 cm 2 (i.e. 4.25 cm x 2.35 cm). The cathodes were cleaned by sonication and washing in methanol and distilled water. A s shown in Fig. 4.2, the porous cathodes were equipped with a stainless steel (316) cap and rod acting as current feeder. To minimize the corrosion of the steel components in the acid electrolytes (e.g. 0.1 M H2SO4) the cap and the bottom part of the rod were wrapped with several layers of Teflon tape and only the R V C part was exposed to the catholyte. Furthermore, the applied cathode potentials provided a certain degree of cathodic protection to the steel components. Thereby, the presence and interference by Fe ions due to corrosion, was ruled out. The anode was a Pt cylindrical mesh (superficial surface area 2.4xl0" 3 m 2 , 5 cm x 0.75 cm) immersed either in 0.5 M H2SO4 (when the catholyte was acidic) or 3 M N a O H (for alkaline catholyte). Current to the cell was supplied by a D C power supply (Anatek), with a maximum output of 1 A and 50 V . The highest current applied to the cell was 0.8 A . The cathodic current 4 Experimental Methods, Apparatus and Materials 42 and potential were each monitored by a digital multimeter, while the quantity of electricity was measured with a digital coulometer (Model 379 P A R Inc.) (Fig. 4.2). A magnetic stirrer bar provided mixing to the cathode compartment with the mixing plate set always to the same speed. The temperature was maintained constant during electrolysis, 300 ±3 K , by placing the 'H- ' ce l l in a water bath, which was connected to circulating thermostat cooling bath. 4.2.2 Continuous ' f low-by' cells a) General Experimental Set-Up for Continuous Operation For both aqueous and L / L emulsion electrosynthesis of peroxide in 'flow-by' cells, the same general experimental set-up was used (Fig. 4.3). The feed tank was equipped with a mixer and the corresponding controller (Servodyne Model 5000-30, Cole-Parmer). For emulsion mediated electrosynthesis the catholyte feed tank contained the organic and aqueous phases in the desired phase volume ratio. To prepare the desired organic phase composition (e.g., dissolve both the supporting electrolyte and 2-ethyl anthraquinone mediator in tributyl phosphate) and to maintain the L / L emulsification, a bow tie coil paddle was used, which is ca 50% more efficient than the standard propeller*. In the case of direct 0 2 reduction in aqueous alkaline and acidic electrolytes, the feed tank mixer was equipped with a marine-type, three blade propeller (used at 700 to 1500 rpm) to keep the surfactant dispersed in the aqueous electrolyte. The maximum catholyte volumes stored in the feed tank were: 20 1 of alkaline or acidic electrolyte for the O2 / surfactant system (see Chapter 5.3) and 16 1 o f L / L emulsion for the E t A Q mediated electrosynthesis (Chapter 6.4). Typically, 30 1 of anolyte was prepared, i.e., 1.5 M N a O H for the alkali and 0.5 M Na2S0 4 at p H 3 (adjusted with glacial acetic acid), for the acidic catholyte, respectively. A n electronic diaphragm pump with polypropylene head and a flow range of 0.79 to 158 ml min"1 ( L E 9IT, L M I Mil ton Roy) was employed for pumping the L / L emulsion. For both the anolyte and aqueous catholyte (i.e. experiments without emulsion), a double headed controlled volume pump was used (Milroyal® D , Mil ton Roy) with a maximum output of 55 ml min"1. * The marine-type propeller generates an axial flow, while the bow tie induces strong centrifugal flow patterns. 4 Experimental Methods, Apparatus and Materials 43 Legend: 1. Mixing plate 2. CorBtantterrperatixewaitar^ 3. Glass *H-cell 4. Cation exchange membrane (Nation 117 or Nation 350) 5. 3-D cathode vvith current feeder 6. Anode (Pt cylindrical mesh) 7. Reference electrode (SCE) 8. Gas inlet 9. Digital mdtiireters 10. Digital coulometer 11. DC Power suppply 12. Magnetic stirrer bar Figure 4.2 Set-up and components of the 'H ' -ce l l for batch electrosynthesis. As shown by Fig. 4.3, the O2 gas flow was introduced into the catholyte liquid line before the reactor. Thus, the electrochemical cell was operated with co-current upward multiphase flow on the cathode side. In parallel with the catholyte, the anolyte was pumped through the anode compartment also in upward flow. The employed electrochemical cells and their components are described in detail in the next two sections. 4 Experimental Methods, Apparatus and Materials 44 Up to 60 A D C current was supplied to the electrochemical cell, via two identical power supplies (Sorensen, Raeyton Co.) connected in parallel, each of them having a maximum output of 30 A and 50 V . The desired pressure in the cell was set and controlled via a pressure control valve (Foxboro, Inc.). In the case of atmospheric pressure operation, e.g., emulsion mediated electrosynthesis, the pressure control valve was by-passed (Fig. 4.3). The temperature of the cell (e.g. for the 0.5 m effective length flow cell) was maintained constant by cooling the back of the cathode stainless steel feeder plate with cold tap water (see Section 4.3 c). The catholyte product after passed through a heat exchanger, was collected in a tank where, the gases (i.e. excess 0 2 and possibly electrogenerated H 2 ) were disengaged. In the case of complete recycle (see Section 6.4.1 e), the catholyte was returned to the feed tank (Fig. 4.3). The anolyte was usually recycled with a flow rate of 55 ml min"1. The p H and concentration of the anolyte were monitored from time to time and adjusted with fresh make-up solution as needed. The experimental set-up shown in Fig. 4.3 was cleaned from time to time by pumping through the system for several hours first a mixture of methanol and deionized water followed by deionized water only to remove the methanol traces. The water source for the experiments was deionized water (using an Ion-X cartidge, Cole-Parmer Inc.) with a specific conductivity of 2 u,S cm"1 *. * For reference, typical measured conductivities of tap and distilled water were: 22.5 uS cm"1 and 1.2 uS cm"1, respectively. 4 Experimental Methods, Apparatus and Materials 45 4 Experimental Methods, Apparatus and Materials 46 b) 'Short', 0.14 m effective length, 'flow-by' cell A s mentioned earlier, two different size flow cells were employed. The one referred to as the 'short' cell, had the cell body made of Plexiglass with the following active cathode dimensions: length 14.5 cm, width 2.4 cm and thickness 4.5 mm (Fig. 4.4). The latter thickness in the case of graphite felt corresponds to 62% compression (see Table 4.1). Taking into account the shape of the electrode (Fig. 4.4), the superficial cathode area perpendicular to the electric current was 33 cm 2 whilst the area perpendicular to the fluid flow was 1.1 cm 2 (in the fully developed region). The entrance and exit regions of the cell contained plastic fluid distributors. The 3-D cathode was embedded in three or four Durabla® gaskets (i.e. two 1/16 inch and one or two 1/32 inch gasket), with the back of the cathode in contact with the feeder plate (Fig. 4.4). The highest current applied to the 'short' cell was 22 A . The overall arrangement of all the cell components on both the cathode and anode sides are shown in Fig. 4.5. Basically the same arrangement was used for both the graphite felt and the reticulated vitreous carbon electrodes, with the only difference that in the case of R V C , a very thin graphite paper (GF-S4 0.3 mm thick) was interposed between the current feeder plate and the cathode, to improve the contact between R V C and the feeder (Fig. 4.3). The margins of the Nafion® 350 membrane were sealed with silicon glue (Dow-Corning, Inc.) on both sides, to assure a tight seal between the gaskets and the membrane upon compression. On the anode side the Nafion® 350 was supported by a plastic screen, which also acted as an anode chamber spacer and turbulence promoter. The anode for alkaline conditions was a porous N i felt (0.7 mm thick) whilst for acidic operation a 'dimensionally stable' (DSA) 0 2 anode plate (1 mm thick) was employed. The cell assembly was uniformly compressed using a torque wrench at ca 5 lbf-ft. The highest pressure applied on the cathode side was 700 kPa. 4 Experimental Methods, Apparatus and Materials 47 2.4 cm 5 cm 10 cm Figure 4.4 Front view and dimensions of the cathode compartment for the 'short' flow-by cell. 4 Experimental Methods, Apparatus and Materials 49 c) 'Long', 0.5 m effective length, 'flow-by' cell The second flow cell employed in the present work, sometimes referred to as the 'long' flow-by cell (effective length 0.5 m) had a superficial cathode area in the direction of the current flow ca eight times higher than 'short' cell, i.e. 267 cm 2 vs. 33 cm 2 . The cathode area in the direction of fluid flow was 2.5 cm 2 vs. 1.1 cm 2 for the 'short' cell. The porous cathode for the 'long' cell was always graphite felt. The front view and corresponding dimensions of the cell are presented in Fig. 4.6. The cell body was made of stainless steel. The graphite felt cathode was embedded in Durabla® gaskets (i.e. two 1/16 inch and one or two 1/32 inch gasket) giving a compressed graphite felt thickness of 4.5 mm (see Table 4.1 as well). To minimize the electrolyte by-pass at the edges of the cathode, the gasket had four 'wings' on each side to redirect the flow toward the centre of the cathode (Fig. 4.6). The back of the graphite felt was in contact with the stainless steel current feeder. Currents up to 60 A were applied to the cell. The general arrangement of the cell components is shown in Fig. 4.7. The major difference between the 'short' and 'long' flow cell set-up, is the presence for the latter of an extra compartment, delimited by a rubber gasket, for water cooling the cathode side (Fig. 4.7). The flow rate of the cooling water was 100 ml min"1, with an inlet temperature of ca 8 °C. A Nafion 350® membrane supported by a plastic screen on the anode side separated the anode and cathode compartments. The anodes were: stainless steel (316) screen for alkaline conditions and D S A plate for acidic operation, respectively. The cell assembly was uniformly compressed using a torque wrench at ca 20 lbfft. 4 Experimental Methods, Apparatus and Materials 8.9 cm 17.7 cm Figure 4.6 Front view and dimensions of the cathode compartment for the 'long' flow-by cell. 4 Experimental Methods, Apparatus and Materials 51 CM 00 CM CD LO CN CO CN v — ' CO aS 0) SO 2 o R - ° X3 <u o 1 «" « is i * l « g « - 3 1 1 >> M O CO CU - a O L.J — _ i sa oo CD 0 S bp H CD + J —1 G CD <D 1 a CD 5 TB ! O OH a BO CD 5 u f CD E c o d o o CD -G o CD « •a w « o 3 S OH ofj w  o o a « CD ^ o » OH CD CD b CO . a 2? -a CD fi CD O _o CD jg s ^ CO ? «" c a 03 h 3 fe N » CD m <; c £ 2 Q z° . a ±f | , C D G ^ CD « a 4 Experimental Methods, Apparatus and Materials 52 4.3 Porous Electrodes and Membranes This section provides a summary of the physico-chemical properties of the employed cathode materials (i.e. graphite felt, GF , different types of reticulated vitreous carbon, R V C ) and cation exchange membranes. 4.3.1 Cathode materials Table 4.1 Physico-chemical characteristics of the graphite felt (Type Grade GF, Carborundum Co.). See also Appendix E, Figure E - l . Property Value Source Initial porosity, eo 0.95 [Olomane^a/., 1991] Mean fiber diameter (urn), df 20 idem Fiber density (kg m"3) 1,500 idem Graphitization (°C / h) 2,400 / 2 idem Carbon content (%) 99 idem Uncompressed thickness(mm), t0 7.25 measured Compressed thickness (mm), t 4.5 measured Compressed porosity, s 0.92 e = i _ ' . o - o t Compressed specific surface area (m 2 m"3), S 16,000 Electronic conductivity of the compressed matrix (S m"1), Km 23 ( V'55 Km = 10 + 2800 1 - — [Olomane/a/., 1991] 4 Experimental Methods, Apparatus and Materials 53 Table 4.2 Physico-chemical properties of various ppi (pores per inch) reticulated vitreous carbon electrodes ( E R G Inc.). See also Appendix E, Figure E-2. Property 10 ppi 30 ppi 60 ppi 100 ppi Specific surface area (m 2 m"3)* 530 1800 3660 6560 Porosity**, em 0.64 0.78 0.94 0.95 Carbon density (kg m"3)*, pcarbon 1,489 idem idem idem Electronic conductivity of the matrix** — 121.5 — 62.2 (S m"1) * [ERG, 1996] ** Measured in the present work (see further) The porosities sm, reported in Table 4.2 were calculated from the water uptake (V) of R V C rectangles of known mass (m), immersed in a measuring cylinder of water (contains also a few drops of non-ionic surfactant (Makon®) acting as a wetting agent): carbon The electronic conductivity of the R V C matrix was determined according the method described by Oloman et al, which is based on the measurement of the voltage drop across the dry matrix as a function of current using a couple of different matrix thickness [Oloman et al, 1991]. Table 4.1 and 4.2, show significant differences of the physico-chemical properties among the investigated cathode materials. Generally, graphite felt had a higher specific surface area and lower electronic conductivity than the R V C s used in this work. The differences in physico-chemical characteristics wi l l play a major role in the overall performance of these porous electrodes. 4 Experimental Methods, Apparatus and Materials 54 4.3.2 Cation exchange membranes (Nafion , DuPont Inc.) Two membranes were used: Nafion® 117 for the 'H ' -ce l l and Nafion® 350 for the 'flow-by' cells, respectively. Both of them are copolymers of tetrafluoroethylene and perfluoro-(4-methyl-3,6-dioxa-7-octene-l-sulfonic acid) but the 350 type is also reinforced for greater mechanical stability and durability. The membranes were mounted wet, after soaking for a few hours in hot water or concentrated N a O H (for alkaline operation). 4.4 Chemicals and Analytical Methods 4.4.1 Surfactants Three surfactants were investigated as representative for their classes, i.e. cationic: tricaprylmethylammonium chloride (Aliquat® 336, Aldrich Inc.), non-ionic: t-octylphenoxypolyethoxy ethanol (Triton® X-100, Sigma Inc.) and anionic: sodium dodecyl sulfate (SDS, Sigma Inc.). The formulae, molecular weight, aggregation number and critical micellar concentration (cmc) in pure water for the above surfactants are given in Table 4.3. Table 4.3 Surfactant characteristics [Sigma, 1991]. Surfactant Molecular weight Aggregation number C M C in pure water (M) (298 K ) Aliquat 336 (Cationic) 404.17 — 1.2xl0" 4 a [ C H 3 ( C H 2 ) 7 ] 3 C H 3 N + C r Triton X-100 (Non-Ionic) 624.9 140 2.4x10"4 C H 3 C ( C H 3 ) 2 C H 2 C ( C H 3 ) 2 -- C 6 H 4 0 ( C H 2 ) 2 0 ( C H 2 ) 2 O H SDS (Anionic) 288.5 62 8.3xl0" 3 C i 2 H 2 5 O S 0 3 " N a + a calculated from log(cmc)=A - B r^  where A=1.25, B=0.27 and 1^ =8; cmc: mM [Myers, 1988]. 4 Experimental Methods, Apparatus and Materials 55 4.4.2 Organic solvent This section provides a summary of the physico-chemical properties for tributyl phosphate, which is the main organic solvent employed in the present systems. Table 4.4 Physico-chemical properties of tributyl phosphate (TBP) [Lange's, 1992; Merck Index, 1989] Formula Source Molecular weight Purity (%) Density (kg m"3) Boiling point (°C) Flash point (°C) Solubility in water (% v oi) Viscosity (Pa s) LD50 (orally in rats, g kg"1) Caution: (C4H 90) 3P(0) Aldrich Chemical Co. 266.3 98 976 (25 °C) 289 (760 mrnHg) 146 0.6 (20 °C) 3.39xl0" 3 (25 °C) 3 Irritating to mucous membranes 4.4.3 Hydrogen peroxide analysis Hydrogen peroxide was analyzed by titration with 0.1 N K M n 0 4 in 20% H2SO4 [Kraft, 1969], In an Erlenmeyer flask containing 20 ml of H 2 S 0 4 20%, typically 0.5 to 3 ml H 2 0 2 solutions were pipetted, followed by titration with permanganate to a faint pink end point. The peroxide concentration was calculated based on the following equation: 2 K M n 0 4 + 5 H 2 0 2 + 4 H 2 S 0 4 -> 2 K H S 0 4 + 2 M n S 0 4 + 8 H 2 0 + 5 0 2 . [4.2] Blank titration experiments showed no significant interference between the permanganate method and either the surfactants or the organic solvent employed. For qualitative detection of peroxide Merck test strips were used (Merck Co. , 1.10011), which gave a characteristic blue coloration in the presence of H 2 0 2 . 5.1 Cyclic Voltammetry Studies of O2 electroreduction to H2O2 56 C H A P T E R 5: Results and Discussion: Surfactant Modified Electroreduction of O2 to H2O2 The goal of the present study was to put forward a different approach for 0 2 electroreduction to H2O2, by exploiting certain surfactant induced interfacial effects on the electrochemistry of the present system (see Chapter 2). Cyclic voltammetry, constant current coulometry and electrosynthesis experiments were performed to evaluate the influence of surfactant type (e.g. cationic, anionic and non-ionic) and concentration on the electroreduction of O2 to peroxide in both acid and alkali electrolytes. 5.1 Cyclic Voltammetry Studies 5.1.1 Kinetics of 0 2 electroreduction on bare glassy carbon First, as a reference, O2 electroreduction was investigated in the absence of surfactants. Figures 5.1 a and b, show major features of the scan rate dependence of the O2 cyclic voltammogram in 0.1 M Na2C03 and 0.1 M H2SO4, respectively. The distinctive feature of the cyclic voltammetry in alkali (Fig. 5.1 a) is the presence of two peaks: a shoulderlike peak at E1 pn = -0.35 V 1 and a bell-shaped peak at Ep between -0.74 and -0.82 V (depending on the scan rate). The latter is usually referred to as the main peak. Taylor and Humffray [1975] showed that both peaks are associated with a single electrode process, i.e. O2 reduction to H 0 2 " , and not with two successive electrode processes as previously thought. This point of view is now accepted in the literature [Hossain et al, 1989; Yano et al, 1998]. 1 all electrode potentials are given vs. Ag/AgCl 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 58 The shoulderlike peak is due to the reduction of 0 2 to HO2" electrocatalyzed by surface functional groups (mostly quinone groups [Hossain et al, 1989]). Note that the reduction peaks of the surface oxide species (e.g. quinone groups) for various carbons are between -0.35 and -0.42 V in 0.1 M N a O H [Swiatkowski et al, 1997]. The main peak is given by the uncatalyzed two-electron reduction of O2 [Yano et al, 1998]. The shift of the corresponding peak potential Ep, toward more negative values with increasing scan rates v, e.g. from -0.72 V (at 0.05 V s"1) to -0.82 V (at 0.5 V s"1), indicates mixed (i.e. activation - diffusion) control (Fig. 5.1 a). Regarding the 0 2 electroreduction in 0.1 M H2SO4 on bare G C (Fig. 5.1b) only one peak was observed, with peak potentials between -0.85 and - 1 V . The shape of the peak was scan rate dependent, i.e. sigmoid at scan rates below 0.2 V s"1, while above 0.2 V s"1 the bell-shaped behaviour was revealed. This indicates a slow, irreversible, electron transfer process under mixed control. Pertinent electrode kinetic parameters for O2 reduction on bare G C (Table 5.1) were calculated from cyclic voltammograms recorded at nine different scan rates (i.e. Figs. 5 .1a and b present two representative cases only). Appendix A and B shows the detailed calculation procedure for the O2 concentrations and kinetic parameters, respectively. Table 5.1 Kinetic parameters for O2 electroreduction on bare G C at 295 K . Electrolyte Overall Electrode Reaction n b(V) k s (m s"1) 0.1 M N a 2 C 0 3 0 2 + H 2 0 + 2e" ->• OH" + H 0 2 " (pH 11.5) 1.90+0.1 0.16 1.2xl0"7 r 1 ,~-13 O . I M H 2 S O 4 0 2 + 2H* + 2e" -> H2O2 2.04+0.04 0.17 5.1x10" (pH 0.9) 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 59 In Table 5.1, the total number of electrons n, confirms the overall 2e" reduction of 0 2 on G C in both electrolytes. The Tafel slopes b, compare fairly well with literature data. For instance at p H between 0.3 and 4.2 a Tafel slope of 0.155 +0.015 V was reported for pyrolitic graphite whereas on various carbons in alkali media, Tafel slopes in the range of 0.104 - 0.150 V were found [Taylor and HumfFray, 1975]. Furthermore, Table 5.1 shows that the standard rate constant ks, for O2 reduction on bare G C is about 2 .4x l0 5 smaller in 0.1 M H 2 S 0 4 than in 0.1 M N a 2 C 0 3 , reflecting the electrode kinetic limitation of the 2e~ O2 reduction in acidic media. Also, the ks for 0.1 M N a 2 C 0 3 is 250 times smaller than the value obtained in 0.1 M K O H on G C (i.e., 3xl0~ 5 m s"1 [Yano et al, 1998]). However, in 0.1 M K O H on a diamond electrode a ks o f 6.6xl0" 7 m s"1 was reported [Yano et al, 1998], showing the importance of the electrode material on the rate of O2 electroreduction to peroxide. 5.1.2 Influence of surfactants on the O2 cyclic voltammograms Figures 5.2 and 5.3, show that all three surfactants (i.e. cationic (A336), non-ionic (Triton X-100) and anionic (SDS),Table 4.3) affect the 0 2 cyclic voltammetry in both 0.1 M H 2 S 0 4 and 0.1 M Na2C0 3 . A l l three surfactants suppress the shoulderlike catalytic peak in alkali, so that, in the presence of surfactant, only one peak is observedun both media. Increasing concentrations of the cationic surfactant (Aliquat 336) shifted the peak potential Ep, to more positive values, i.e. in 0.1 M Na2C0 3 , the peak shifted from -0.76 V to -0.49 V with 1.4 m M Aliquat 336 (Fig. 5.2), while in 0.1 M H 2 S 0 4 from -0.99 V in the absence of surfactant, to -0.69 V in the presence of 17 m M Aliquat 336 (Fig. 5.3). The peak potential in the presence of the non-ionic (Triton X-100) surfactant remained almost unchanged with surfactant concentration for both electrolytes (Figs. 5.2 and 5.3). The same is approximately valid in carbonate for the anionic surfactant (SDS), while in acid the effect of SDS was somewhat more complex, e.g. with increasing SDS concentrations Ep shifted significantly to more positive values. 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 60 Figure 5.2 The effect of surfactant type and concentration on the first scan of 0 2 cyclic voltammetry. Electrolyte: 0.1 M N a 2 C 0 3 . Electrode: G C . Scan rate: 0.3 V s"1. T: 295 K . 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 61 Figure 5.3 The effect of surfactant type and concentration on the first scan of 0 2 cyclic voltammetry. Electrolyte: 0.1 M H 2 S 0 4 . Electrode: G C . Scan rate: 0.5 V s"1. T: 295 K . 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H2O2 62 The dependence of the peak current Ip, on the surfactant type and concentration (Figs. 5.2 and 5.3) provides insights into the mechanism of surfactant adsorption and surface film formation. Generally, in both H 2 S 0 4 and Na 2CC»3, the presence of the cationic surfactant increased Ip while the non-ionic and anionic surfactant decreased Ip. This effect indicates differences in the adsorption mechanisms between the cationic and the other two types of surfactant. For Aliquat 336 (A336) the dependence of Ip on surfactant concentration in 0.1 M N a 2 C 0 3 (Fig. 5.4) resembles the shape of a typical adsorption isotherm for polar surfactant adsorption on charged surfaces [Rusling, 1994 I and II; Myers, 1988] (see Section 3.2.2 1). Therefore, there appears a direct relationship between the electroreduction of 0 2 and the adsorption of cationic surfactant on the cathode. Interactions between the adsorption mode of cationic surfactants and electrochemistry, was observed for other electrosynthesis reactions as well (e.g. acetophenone reduction [Mousty and Mousset, 1992]). From Fig. 5.4, three regions can be identified as a function of Aliquat 336 (A336) concentration. A t concentrations below 6x10"3 m M (region I, Fig. 5.4) the cationic surfactant had little effect on Ip. However, at concentrations between 6xl0" 3 m M and 1 m M (i.e. around the cmc, Table 4.3) the peak current increases by almost 50% (region II, Fig. 5.4), whereas a further increase of the A336 concentration above 6 m M led to the leveling ofIp (region III). 5. / Cyclic Voltammetry Studies of 02 electroreduction to H202 63 12 <^  11 CD 13 O CO CD C L u 6x10 3 2 4 6 8 10 A336 concentration / mM Figure 5.4 Peak current for 0 2 reduction obtained on the first scan vs. Aliquat 336 concentration. Electrolyte: 0.1 M N a 2 C 0 3 . Electrode: G C . Scan rate: 0.3 V s"1. T: 295 K . The effect of Aliquat 336 concentration on Ip obtained for 0 2 electroreduction in 0.1 M H 2 S 0 4 (Fig. 5.5); is similar to that in 0.1 M N a 2 C 0 3 except that i) the cationic surfactant concentration where Ip reaches a maximum, is ca 3 times higher for 0.1 M JL.SO4 than for 0.1 M N a 2 C 0 3 (i.e. 17 m M vs. 6 mM) and ii) once a maximum Ip is reached, a higher Aliquat 336 concentration (e.g. 30 mM) decreases the peak current. Taking into account the theory of surfactant adsorption on charged surfaces ([Rusling, 1994 I]; [Rusling, 1994 II]; [Myers, 1988]), it is proposed that the observed behaviour of the 0 2 reduction peak current and potential in the presence of the cationic surfactant, is due to the complex changes occurring at the electrode surface with regard to the electrical properties (e.g. Stern potential, surface charge density), transport phenomena (e.g. 0 2 diffusion and concentration in the surfactant aggregates) and surface pH. 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 64 CO (D C L 18 16 A 14 H 12 H 10 8 II / * \ III n - J - / — r — y — i i 0 6x10-3 10 20 30 A336 concentration / mM Figure 5 . 5 Peak current for 0 2 reduction obtained on the first scan vs. Aliquat 336 concentration. Electrolyte: 0.1 M H 2 S 0 4 . Electrode: G C . Scan rate: 0.5 V s"1. T: 295 K . Regarding the influence of Triton X-100 and SDS on the peak current (Figs. 5.2 and 5.3), both surfactants lowered the peak current, hence, they suppressed the electroreduction of 0 2 . This is presumably due to the so-called 'blocking effect' of these surfactants [Lipkowski, 1992]. Therefore, it is hypothesized that instead of forming highly-ordered surfactant aggregates with head-down facing the surface, the non-ionic and anionic surfactants might adsorb in 'trains' and 'L' s [Myers, 1988], where parts of the hydrocarbon chain are facing the electrode surface. In 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 65 alkali, (pH 11.5) the electrostatic repulsion between the cathode and the anionic head group contributes to the formation of a 'blocking' arrangement of the SDS film. The increase of the 0 2 reduction peak current with A3 3 6 concentration is probably due to an increase of surface p H induced by the cationic surfactant. The voluminous tetraalkyl ammonium ions of the cationic surfactant displace protons from the electric double layer, hence, the surface p H increases and promotes O2 reduction. Similar surface p H effects induced by quaternary ammonium ions were exploited in other electrochemical systems, most notably in the electroreduction of acrylonitrile to adiponitrile [Rieger, 1987], A theoretical estimation of the surface p H for the present case, is given in Appendix C. It was found that in 0.1 M H2SO4 (bulk p H = 0.9) with 17 m M A336 the surface p H (i.e. at the inner limit of the diffuse double layer) can be as high as 9.4. A local p H of 9 - 10, as calculated theoretically (Appendix Q , is supported by the experimental data, which indicates a shift of the peak potential toward more positive values with increasing concentrations of Aliquat 336 (Figs. 5.2 and 5.3). Since A336 was the only surfactant (of those investigated, Table 4.3), which increased the rate of O2 electroreduction, it was used in further cyclic voltammetry studies. 5.1.3 Effect of A336 on the 0 2 electroreduction kinetic and mass transfer parameters The scan rate dependence of successive cyclic voltammograms was studied to obtain quantitative information regarding the influence of surface film formation on O2 electroreduction. Aliquat 336 was used in concentrations, which gave the most significant changes in Ip and Ep (Figs. 5.4 and 5.5), i.e. 17 m M in 0.1 M H2SO4 and 1.5 m M in 0.1 M Na2C03, respectively. Figure 5.6 illustrates the effect of successive potential cycling on the O2 voltammogram in the presence of A336 for scan rates between 0.3 and 0.5 V s"1. In Fig. 5.6, there is a significant difference between the first and the subsequent scans with respect to the magnitude of the peak current. The same behaviour was also obtained for the scans in Na2CC>3, as shown by Fig. 5.7 for both cases. Furthermore, Figs. 5.6 and 5.7 illustrate that the dependence of the peak currents on 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 66 the square root of scan rate is linear, in the presence of A336, as expected from the cyclic voltammetry theory. In the absence of surfactant the difference between the first and n-th peaks, was much smaller than in the presence of A33 6. For instance, at 0.4 V s"1, without surfactant, the ratio Ipj I IPi„ was measured 1.35, while with 17 mMA336, this ratio is 2.34 (Fig. 5.7). Fig. 5.7 also shows that in the presence of A336, the slope of the line Ip vs. square root of scan rate, is much larger for the 1-st than for the n-th scan. Actually, for the n-th scan Ip leveled off at scan rates above 0.3 V s"1. These results could be rationalized by assuming that the first scan response is due to the reduction of 0 2 from the surface film (i.e. intra-admicelle 0 2 ) while in subsequent scans, as the O2 from the immediate vicinity of the electrode surface was depleted, the cyclic voltammogram is controlled by O2 diffusion through the surfactant film. Since the peak current for the scans > 1 was always smaller at scan rates above 0.1 V s"1, the complete replenishing of the surface film with O2 was not achieved in the time between two consecutive scans. Therefore, it is plausible to assume that the reaction plane for O2 reduction is located inside the film and diffusion effects (e.g. non-linear diffusion to a partially blocked surface) might play an important role (see further). Therefore, the O2 electroreduction in the presence of the A3 3 6 surface film can be described by the following sequence of steps (written here for acidic media): O2, bulk ^ O2, outer plane of the film [5.1] O2, outer plane of the film ^  O2, inside the film [5-2] 02, inside the film + 2e + 1 ^ 2 0 ^ ( H 0 2 + OFT), inside the film [5.3] H 0 2 " , inside the film , outer plane of the film [5.4] H 0 2 " , outer plane of the film + F T ^  H2O2, bulk • [5.5] 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 68 0.2 0.3 0 .4 0.5 0.6 0.7 0.8 V^2 /(V S- 1) 1 / 2 Figure 5.7 Peak current vs. square root of scan rate for the 1-st and n-th scans. Electrode: G C . A336 concentration: 17 m M (for 0.1 M H 2 S 0 4 ) and 1.5 m M (for 0.1 M N a 2 C 0 3 ) . The mathematical treatment of the cyclic voltammetry theory in the presence of surfactants is typically based on the 'blocking' effect assumption, i.e. the rate of the electrochemical reaction is diminished by the surface coverage (the electrode surface is divided in 'active' and 'inactive' sites, the latter ones being those occupied by the surfactant) [Gueshi, et al, 1978, 1979]. However, in the present case the sites occupied by the surfactant might not be 'inactive' since the quaternary ammonium ions are able to induce a significant increase of the surface p H at those sites (Appendix C). 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 69 Therefore, in order to estimate, albeit crudely, the kinetic parameters for 0 2 electroreduction in the presence of A336, the 1-st scan cylic voltammetry data (Fig. 5.7), was subjected to the classic cyclic voltammtery interpretation of mixed activation-diffusion control (Appendix B, based on [Gileadi et al, 1975]. The corresponding O2 concentration was assumed to be the one inside the surface film, Co2,fiim (see Appendix D). From the 1-st scan cyclic voltammetry data for 0.1 M Na2C03 and 0.1 M H2SO4 (Figs. 5.6 and 5.7), the O2 diffusion coefficient in the surfactant layer, the Tafel slope, and standard rate constant for O 2 reduction to H 2 O 2 , were estimated (Table 5.2). Table 5 . 2 Apparent kinetic and diffusion parameters for O 2 electroreduction on G C in the presence of Aliquat 336 surface film at 295 K . Results based on 1-st scan cyclic voltammetry. Electrolyte Overall Electrode Reaction Dmm bmm k S ; m,,, (V) (m s-1) 0.1 M N a 2 C 0 3 Aliquat 336: 0 2 + H 2 0 + 2e~ - » OFT + H 0 2 " (4 .3±0.4)xl0" ,-io 0.22 1.8x10" ,-6 1 .5x l0" 3 M 0.1 M H 2 S O 4 Aliquat 336: 1 .7x lO" 2 M 0 2 + 2 rT + 2e~ -> H 2 0 2 (6.5+0.3) x l O " 1 0 0.22 9.9x10 5.1 Cyclic Voltammetry Studies of 02 electroreduction to H202 7 0 Comparing the data from Tables 5.1 and 5.2, show that the Aliquat 336 had a significant effect on all the parameters. First, the presence of the cationic surfactant increased the standard rate constant k„, about 15 times in 0.1 M N a 2 C 0 3 and 1,900 times in 0.1 M H 2 S 0 4 . Also, the Aliquat 336 film formation decreased the 0 2 diffusion coefficient, i.e. 3 times in 0.1 M F f 2 S 0 4 and 4 times in 0.1 M Na 2CC>3 compared to the value in the absence of surfactant, which is 1.9xl0"9 m 2 s"1 [Kinoshita, 1992]. The observed Tafel slopes bf,im, are the same for both alkali and acid, i.e. 0.22 V (Table 5.2). Tafel slopes in the range of 0.21 - 0.30 V are commonly encountered for electrode processes influenced by strong adsorption [Bockris and Khan, 1993]. 5.2 Batch Electroreduction of 02 to H202 on RVC 71 5.2 Batch Electroreduction of O2 on Reticulated Vitreous Carbon 5.2.1 Constant current coulometry. Influence of surfactants. Hydrogen peroxide electrosynthesis experiments were performed in an 'H'-cell (see Fig. 4.2), to investigate the effect of the three classes of surfactants on certain figures of merit such as, accumulated peroxide concentration and peroxide current efficiency. The duration of these experiments was short (i.e. 15 min. in acid and 5 min. in carbonate) to minimize the interference of the secondary reactions such as electroreduction and / or chemical decomposition of the electrogenerated peroxide. Figures 5.8 a and b, show the current efficiency for O2 reduction to H2O2 as a function of surfactant type and concentration in acid and alkali, respectively. Corroborating the cyclic voltammetry studies, only the Aliquat 336 increased the current efficiency (Fig. 5.8). The anionic and non-ionic surfactants retarded the O2 electroreduction. For instance, in 0.1 M H 2 S 0 4 , 2 m M of both SDS and Triton X-100 lowered the current efficiency for H2O2 from 40% without surfactant to 10%. The cationic surfactant had the greatest effect in acid (Fig. 5.8 a), where 1 m M concentration of Aliquat 336 increased the current efficiency from 40% to 92%. Increasing the Aliquat 336 concentration above 1 m M had no additional effect, the current efficiency leveled off around 90%. In alkali on the other hand, cationic surfactant concentrations up to 1 m M increased the current efficiency from 67% to 92% while a higher surfactant concentration was less effective, e.g. at 8.5 m M surfactant the current efficiency was only 80%. It is hypothesized that for high concentrations of Aliquat 336 (i.e. about 10 mM) a thick film of surfactant formed on the electrode, which hindered the access of O2 to the electrode surface. This assumption is also supported by the low intra-admicelle O2 diffusion coefficient in the carbonate electrolyte (Table 5.2). 5.2 Batch Electroreduction of 02 to H202 on RVC 72 CD " O X o i _ (D CL \ o "o i t c (D «_ »_ Z3 o 60 H 40 * 20 H 0 100-£ 80 60 -40 -20 -0 * a) 0.1 M H S O , 2 4 * A l iquat 3 3 6 A Tr i ton X-100 • S D S " T — • — r -b) 0.1 M N a 2 C O s N o surfactant 9ji A l iquat 3 3 6 A Tr i ton X - 1 0 0 • S D S 0 6 8 10 12 14 Surfactant concentration / mM Figure 5.8 Current efficiency for 0 2 reduction to H2O2 on a 30 ppi R V C vs. surfactant type and concentration. O2 saturated electrolyte. Superficial current density: 300 A m"2. T: 300 K . a) 0.1 M H2SO4; b) 0.1 M N a 2 C 0 3 . Surfactant: * Aliquat 336; • Triton X-100; • SDS. 5.2 Batch Electroreduction of02 to H202 on RVC 73 5.2.2 Influence of surfactants on the electroreduction of H2O2 For efficient 2e' reduction of O2 and electrogeneration o f concentrated H2O2 solutions (e.g., > 0.1 M ) , the secondary cathode reaction involving H2O2 has to be suppressed. The reduction o f H 2 0 2 to OFT or H 2 0 (eqns. [5.6] and [5.7]) , although thermodinamically favored vs. the O2 / H2O2 reaction, generally requires large cathodic overvoltages on carbon based electrodes as a consequence o f slow electrode kinetics. [5.6] In alkaline solution: H 0 2 + F f 2 0 + 2e" -> 30FT, E ° = + 0 . 8 7 V S H E In acid solution: H 2 0 2 + 2 F f + 2e" - » 2 H 2 0 , E ° = +1.76 V S H E • t 5 J l The rate of the above electrochemical reactions is o f first order with respect to the undissociated peroxide concentration Cmoi and zero order vs. H + concentration [Vetter, 1967], i.e. lH202 - 2Fks,H202CH202 e X P ^ V RT j [5.8] where IH202 current density for peroxide reduction (A m"2), kStH202 standard heterogeneous rate constant corresponding to eqns. [5.6] and/or [5.7] (m s"1), /? symmetry factor, 77 overvoltage (V). The rest o f the variables have their usual meaning. Equation [5.8] implies that the rate of peroxide electroreduction is independent o f p H up to a value where the H2O2 dissociation becomes significant, hence, CH202 decreases. At p H > p K a , H 2 0 2 = 11.6 [Lange's, 1992] the rate of peroxide electroreduction diminishes, since only < 50% of the total H2O2 remains undissociated. Furthermore, the half-wave potential for the H 2 0 2 / H 2 0 couple decreases from -0:6 V S H E at p H 11 to -1 V S H E at p H 13.5 [Vetter, 1967]. In the context of the present work, it was of interest to evaluate the H2O2 electroreduction under relevant experimental conditions, with and without the. presence of surfactants. Constant coulometry experiments were performed in deoxygenated 0.1 M N a 2 C 0 3 and H2SO4 solutions containing initially ca 1 M H2O2. The cathode was R V C 30 ppi operated at 300 5.2 Batch Electroreduction of02 to H202 on RVC 74 A m"2 for 15 min. at 298 K . The peroxide concentration and cathode potential were followed as a function of surfactant type and concentration. 'Blank' experiments (i.e. without current applied), were conducted, to determine the rate of peroxide loss by homogeneous decomposition. The results are presented in Figs. 5.9 and 5.10. From Fig. 5.9, the peroxide electroreduction rate is given by the difference between the chemical and the total (cathodic and chemical) loss o f H2O2 over the 15 min. period. Without surfactant, the peroxide electroreduction rate was 2.1xl0" 3 M min"1 in 0.1 M H2SO4 (Fig. 5.9 a) and 2.6xl0" 3 M min"1 in 0.1 M N a 2 C 0 3 (Fig. 5.9 b), which confirms the weak dependence on p H in the range of 0.9 and 11.5. In 0.1 M H2SO4 all three surfactants investigated (A336, Triton X-100 and SDS) suppressed the loss of H 2 0 2 by electroreduction. About 4 m M concentration of both A336 and Triton X-100 reduced the H2O2 loss to levels which correspond to chemical decomposition only, i.e. 1% loss in 15 min. In 0.1 M N a 2 C 0 3 , on the other hand (Fig. 5.9 b), the anionic surfactant SDS, was ineffective (i.e. after 15 min. of electrolysis the peroxide concentration was the same as in the absence of surfactant). However, under similar conditions, both A336 and Triton X-100 suppressed the peroxide loss as shown by the higher residual peroxide concentrations in comparison to the absence of surfactants. 5.2 Batch Electroreduction of 02 to H202 on RVC 75 1.00 0.99 0.98 -A 0.97 -A 0.96 T T initial cone. | chemical decomp. A * -a) :l1 r r-y ^ A 3 3 6 • Tr iton X - 1 0 0 + Laury l Su l fa te 0.95 1.00 0.98 0.96 -A l\lo surTactanT eTectrored7+~cTiemlcal decomp. —' 1 r 0 1 - | 1 1-2 3 n — r 5 0.94 0.92 -A T b) initial cone. chemical decomp. -A 0.90 w • • A/o surfactant, electrored. + chemical decomp. ~ 1 — r 4 Surfactant concentration / M x 1 0 3 Figure 5.9 Influence of surfactants on the electroreduction of H 2 0 2 on R V C 30 ppi. Constant current coulometry. N 2 atmosphere. 300 A m"2, 15 min., 298 K . a) 0.1 M H 2 S 0 4 ; b) 0.1 M N a 2 C 0 3 . 5.2 Batch Electroreduction of 02 to H202 on RVC 76 The cathode potentials measured ~1 min. after the start-up of electrolysis at 300 A m"z superficial current density, show a strong dependence on both surfactant type and concentration (Fig. 5.10). The more negative the cathode potential at which the peroxide reduction occurs (for a constant current density) the more effective the surfactant is in protecting peroxide against reduction. In acid (Fig. 5.10 a) all three surfactants investigated decreased the cathode potential for H2O2 reduction regardless of the surfactant type, i.e. typically E shifted from -1.05 to ca -2.25 VAg/Agci with 1 m M surfactant. Hence, as discussed previously, they were almost equally efficient in minimizing the H2O2 electroreduction. In alkali (Fig. 5.10 b) 0.6 m M A336 depressed the cathode potential, Ec, to -1.78 V compared to -0.92 V without surfactant present. Sodium dodecyl sulfate (SDS) on the other hand, left Ec basically unchanged that is reflected in its inability to prevent the electroreduction of HO2". Afanas'ev et al. [1992] observed as well the inhibiting effect o f certain quaternary ammonium ions on peroxide electroreduction on dropping H g electrode at potentials ca -1 V S C E in 3.16% NaCl . It was found that the inhibiting effect increased with hydrocarbon chain-length, e.g. tetrabutylammonium ion had the strongest influence while the tetramethylammonium had no influence at all. Furthermore, it has been suggested that the inhibitory effect of the quaternary ammonium surface structures on peroxide electroreduction, is due to the hindered diffusion of peroxide to the electroactive sites [Afanas'ev et al, 1992]; Afanas'ev et al, 1991]. However, more investigations are needed to establish the diffusional and/or electrode kinetic nature of the this effect on the peroxide electroreduction. The difference between the acid and alkali electrolytes, regarding the effect of the anionic and non-ionic surfactants on H2O2 electroreduction, is intriguing. It is proposed that the observed differences are related to the modifications in the chemical structure o f the surfactants with pH. In acid (pH 0.9) the negative charge on the anionic surfactant (SDS) is neutralized by H + transforming the - S 0 3 - functional group into -SO3H. Therefore, in acid SDS wi l l act somewhat as a non-ionic surfactant, without being electrostatically repelled by the cathode surface unlike in alkali (pH 11.5). This allows for the formation of more tightly arranged, ordered SDS surface structures that are able to minimize access of H2O2 to the cathode surface. 5.2 Batch Electroreduction of 02 to H202 on RVC 77 In 0.1 M H2SO4 the polyether non-ionic surfactant, Triton X-100, can undergo protonation as well at the oxygen atom of the ether linkage. Thereby, Triton X-100 might acquire a slight positive charge that induces the possibility of electrostatic attraction toward the cathode, generating a more organized surface arrangement that is effective in blocking the access of H2O2 to the electrode. A detailed analysis, using surface science techniques such as atomic force microscopy [Manne and Gaub, 1995], would be required to establish conclusively the structure of the surface films under different p H and polarization conditions and how the molecular organization of the surfactant affects the H2O2 reduction. Such an analysis was beyond the objective of the present work. Based on the studies regarding the influence of surfactants on the 2e" reduction of both O2 and H2O2, it is concluded that the cationic surfactant A3 3 6 in concentrations ca I to 3 m M could improve the figures of merit for O2 electroreduction to H 2 0 2 over a wide range of pH. Therefore, electrosynthesis experiments were performed in batch and continuous 'flow-by' cells to observe the effect of A336 on peroxide concentration and current efficiency. 5.2 Batch Electroreduction of 02 to H202 on RVC -2.6 O «: CD < C O > > L U I 1 1 i — a) 0.1 M H 2 S 0 4 -2.2 -\ -1.8-44 -1.4 -\ -1.0 • A336 • T r i ton X - 1 0 0 • S D S No surfactant 0 -1.8 -1.6 H •1.4 H •1.2 -1.0 H -0.8 i 1 i 1 r b) 0.1 M N a 2 C 0 3 No surfactant 8 — i | i | i | i | i r — 0 1 2 3 4 5 Surfactant concentration / M x10"3 10 Figure 5.10 Cathode potentials after 1 min. of H 2 0 2 electroreduction. Other conditions idem Fig. 5.9. 5.2 Batch Electroreduction of 02 to H202 on RVC 79 5.2.3 The effect of A336 on the figures of merit for batch O2 electroreduction Ten-hour batch O2 electroreduction experiments were performed under galvanostatic conditions (300 A m"2) as a function of Aliquat 336 concentration using the 30 ppi R V C cathode (see Sections 4.2.1 and 4.3.1). The peroxide concentration, cathode potential and p H of the catholyte were followed over time. Figure 5.11 shows that in 0.1 M H2SO4 at a constant superficial current density of 300 A m"2, after 10 hours the H2O2 concentration was 0.26 M for a Aliquat 336 concentration of 1 -2.5 m M , while without surfactant only 0.07 M H2O2 was obtained. Correspondingly, the current efficiency for peroxide after 10 hours was around 55% with Aliquat 336 and 14% without it. The cathode potentials were in the range of -1.0 to -1.5 V vs. A g / A g C l . The p H of the catholyte varied slightly from the initial value of 0.9 to a value between 0.8 and 1.0 after 10 hours. The rate of H2O2 electrogeneration in 0.1 M H2SO4 was higher in the first 5 hours for 1 m M Aliquat 336 than in the case of 2.5 m M surfactant concentration. However, in the last 5 hours, when the H2O2 concentration in the cell exceeded 0.2 M , there was very little additional H2O2 accumulated in the cell with 1 m M surfactant, while with 2.5 m M Aliquat 336 the peroxide concentration kept increasing near the rate of the initial 5 hours (Fig. 5.11). The reason for this is the lower electroreduction rate of H 2 0 2 at higher surfactant concentrations. In 0.1 M Na2CC«3 at 300 A m"2 (Fig. 5.12), addition of Aliquat 336, improves the figures of merit for O2 electroreduction, e.g. after 10 hours with 2.5 m M Aliquat 336 the current efficiency is 61% (H2O2 cone, of 0.31 M ) while without surfactant the current efficiency levels after 3 hrs. at 7% ( H2O2 cone, of 0.014 M ) . The p H of the catholyte increased from 11.5 initially to 13.2 - 13.4 after 10 hours and the cathode potential became more negative over time. For instance, at 300 A m"2 for 0.8 m M Aliquat 336, the potential decreased from -1.1 V after 1 hour to -2.3 V vs. Ag/AgCl after 3 hours, leveling off at this value up to 10 hours. 5.2 Batch Electroreduction of 02 to H202 on RVC 80 0 2 4 6 8 10 Time / hr. Figure 5.11 The influence of Aliquat 336 concentration on the electrosynthesis of H2O2 in 0.1 M H2SO4 at 300 A m"2. Cathode: 30 ppi R V C . p H ^ a i 0.9. T: 300 K . A336 cone. (mM): • 0; • 1; and A 2.5. 5.2 Batch Electroreduction of 02 to H202 on RVC 81 0 2 4 6 8 10 Time / hr. Figure 5.12 The influence of Aliquat 336 concentration on the electrosynthesis of H2O2 in 0.1 M N a 2 C 0 3 at 300 A m"2. Cathode: 30 ppi R V C . pH m i t iai 11.5. T: 300 K . A336 cone. (mM): • 0; • 0.8 and • 2.5. 5.2 Batch Electroreduction of 02 to H202 on RVC 82 The cyclic voltammetry studies along with batch electrosynthesis experiments, showed the possibility of employing A336 to enhance the O2 reduction to H2O2. A s shown in Figs. 5.11 and 5.12 in the presence of 0.8 - 2.5 m M Aliquat 336, at 300 A m"2, 0.1 M P a 0 2 and 300 K , 0 2 was reduced fairly efficiently to peroxide in both acid and alkali. Peroxide concentrations up to 0.31 M in 0.1 M N a 2 C 0 3 (current efficiency 61%) and 0.26 M in 0.1 M H 2 S 0 4 (current efficiency 55%) were obtained, while without surfactant the maximum peroxide concentration in acid was about 0.07 M (current efficiency 14 %) and 0.014 M in alkali (current efficiency 7%), respectively. Therefore, further studies were performed under conditions closer to potential industrial applications, i.e., using a continuous 'flow-by' electrochemical cell operated at superatmospheric pressure and current densities above 300 A m"2. 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 83 5.3 Electroreduction of 0 2 to H 2 0 2 in 'Flow-by' Cells Parametric and factorial experiments were performed in continuous 'flow-by' cells to evaluate the effect of A3 3 6 on 0 2 electroreduction to H 2 0 2 under a wide range of operating conditions, e.g. superatmospheric pressure, various liquid and gas loads and superficial current densities above 300 A m"2. 5.3.1 Graphite felt vs. reticulated vitreous carbon cathodes In cyclic voltammetry and batch electrolysis, vitreous carbon was the working electrode of choice (Chapters 5.1 - 5.2). However, graphite felt is a very effective cathode material for 0 2 reduction to H 2 0 2 in trickle-bed electrochemical reactors. In addition to favorable electrocatalytic properties, the graphite fibre bed operated in co-current upward gas-liquid flow provides high overall mass transfer capacity (e.g. 3 to 9 s"1) and increased liquid hold-up [Hodgson and Oloman, 1999]. Graphite felt (Table 4.1) and reticulated vitreous carbon 100 ppi (Table 4.2), were tested as cathodes for 0 2 reduction in both alkali (pHj n 9.6) and acid (pHj n 3.1), in the presence and absence of the cationic surfactant (Figs. 5.13 and 5.14). The electrolytes were 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pH;„9.6) and 1 M N a 2 S 0 4 (pH i n 3.1 acidified with glacial acetic acid and 2 M H 2 S 0 4 ) . The 0.14 m effective length (i.e. 'short') flow cell (Chapter 4), was employed with a gas load of 0.16 kg m"2 s"1 at STP* and liquid load of 4.5 kg m"2 s"1. These values are in the 'percolating (surging)' flow regime [Storck et al, 1986]. The outlet pressure was kept constant at 350 kPa for alkali and 425 kPa for acid, respectively. Figures 5.13 and 5.14 show that in both media, regardless of the surfactant presence, the peroxide concentrations generated per one pass, are much higher for the graphite felt cathode than for R V C at superficial current densities above 300 A m"2. For instance in alkali, with 3 m M A3 3 6, at 5700 A m"2 using the graphite felt cathode, the peroxide concentration was 120 m M (current efficiency 51%), while for R V C _ 1 0 0 , starting at 800 A m"2 the H 2 0 2 concentration obtained per pass decreased, becoming zero at about 2000 A m"2 (Fig. 5.13). * Standard temperature and pressure: 273 K and 1 arm. 5.3 Electroreduction of O2 to H2O2 in 'Flow-by' Cells 84 140 —1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r 0 2000 4000 6000 8000 Current density / A m 2 Figure 5.13 Electroreduction of 0 2 in 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 ( p H i n 9.6). Comparison between graphite felt (GF) and reticulated vitreous carbon 100 ppi (RVC_100) cathodes. 'Flow-by' cell (0.14 m effective length). Acathode 33xl0" 4 m 2 . P o u t 350 kPa. Liquid flow rate: 4.67xl0" 7 m 3 s"1. Gas flow rate: 1.2xlQ-5 m 3 s"1 at STP. T a v e 308 K . 5.3 Electroreduction ofC>2 to H2O2 in 'Flow-by' Cells 85 0 1000 2000 3000 4000 Current density / A nrr 2 Figure 5.14 Electroreduction of 0 2 in 1 M Na2SC>4 (pH m 3.1) . Comparison between graphite felt (GF) and reticulated vitreous carbon 100 ppi (RVC_100) cathodes. 'Flow-by' cell (0.14 m effective length). Acathode 33xl0" 4 m 2 . P o u t 425 kPa. Liquid flow rate: 4.67xl0" 6 m 3 s"1. Gas flow rate: 1.2xl0"5 m 3 s"1 at STP. T a v „ 308 K . 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 86 The difference in performance between R V C and G F at superficial current densities above ca 500 A m"2, is due to the complex interdependencies between mass transfer, potential distribution and electrode kinetics in the porous, 3-D, electrodes. For the employed liquid and gas loads, based on the Hodgson - Oloman correlation (eqn. [3.24]) the mass transfer capacity of G F is 6.7 s"1. For the reticulated vitreous carbon cathode, due to lack of mass transfer correlations for relevant two-phase (i.e. 'reactive' gas and liquid) flow conditions, the mass transfer capacity was estimated based on one-phase flow studies, as -18 times lower than for G F [Walsh, 1993]. Consequently, the mass transfer capacity for R V C _ 1 0 0 is ca 0.37 s"1. Employing eqns. [3.20] and [3.22] - [3.25], the mass transfer limited superficial current densities IL, are: 370 A m"2 for R V C and 2200 A m"2 for GF, respectively (see Appendix E). Figure 5.13 shows that in alkali, O2 was reduced fairly efficiently even in the absence of surfactant (i.e. current efficiencies above 80%), for current densities up to the mass transfer limited values (i.e. 2200 A m"2 for G F and 370 A m"2 for R V C ) . The same conclusion, however, is not true, for the acid electrolyte (Fig. 5.14), where due to the O2 electroreduction kinetics limitation, the cell cannot be efficiently operated at superficial current densities as high as the mass transfer limited values. For instance, without surfactant, on G F at ca 2000 A m"2 the current efficiency is only 17%, while in alkali under the same conditions is close to 80% (Figs. 5.13 and 5.14). Regarding the effect of A3 3 6, it is important to note, that for both electrodes, the effect of the cationic surfactant in the alkali electrolyte was significant only above the mass transfer limited current density (e.g. /' > 2000 A m"2 for GF, Fig. 5.13), while in acid, the cationic surfactant had a pronounced effect even much below the mass transfer limited current density (e.g. starting at / > 500 A m"2 for GF, Fig. 5.14). Generally, under the employed conditions the cationic surfactant increased the peroxide concentration and current efficiency by up to 40% in alkali and up to 50% in acid, for both G F and R V C J 0 0 . To explore the role of A336 in conjunction with the other process variables (e.g. liquid load, pressure, current density), factorial experiments were performed using the graphite felt cathode in the 'short' (0.14 m effective length) flow-by cell. 5.3 Electroreduction of 02 to H2O2 in 'Flow-by' Cells 87 5.3.2 Factorial experimental design: Acid electrolyte Four variables at two levels were selected for the factorial experiments in 1 M N a 2 S 0 4 (pHi n 3.1). The variables and their levels are given in Table 5.3. Furthermore, a 'center point' was added to investigate the potential non-linear response (i.e. curvature) of the system. Table 5.3 Variables and their levels for the 2 4 +l factorial runs in 1 M N a 2 S 0 4 (pH; n3.1). T a v g : 303 K . A ^ 33xl0" 4 m 2 . G: 0.47 kg m"2 s"1 at STP. Graphite Felt Cathode. Nr. Variables Levels Low (-) Center (0) High (+) 1 Current density; I (Am" 2 ) 729 1000 1274 2 Catholyte Load; L (kgnfV) 1.6 2.9 4.3 3 Pressure; P* (kPa) 425 525 700 4 A336 cone; S (mM) 0 0.66 1.1 * absolute outlet reactor pressure According to the 2 4 +l experimental design created with the Jass 2.1® software [Jass, 1987], seventeen experiments were performed. The 'responses' were the peroxide concentration per one pass and the corresponding current efficiency. The experiments and the results obtained 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 88 are given in Table 5.4. Selected runs were replicated and the response errors together with the confidence intervals for the effects were calculated (see Appendix F). The outlet p H for the factorial experiments (Table 5.4) varied between 3.8 and 5.0. A p H of 5 was obtained when the cell was run at 'low' flow rate (i.e. ' low' liquid load) and 'high' current density (i.e. 1274 A m"2). It was assumed that the change of outlet p H in the above range, had not significant effect on the results. Table 5.4 shows that generally, the peroxide concentration obtained per pass was the highest for 'low' liquid loads (L) , e.g. runs nr. [5] and [15] in Table 5.4. This is due to the increased catholyte residence time in the cell. However, higher H2O2 concentration at low flow rate also tends to favour peroxide reduction, hence lowers the current efficiencies, as shown in Table 5.4, e.g. compare runs nr. [15] (CE: 35%) and [13] (CE: 62%). Davison et al. [1983] showed that for the O2 / H2O2 electrosynthesis, the lower the flow rate the more non-uniform the potential distribution along the length of a packed-bed cell, i.e., the potential becomes more negative from entrance to the exit. Therefore, at low flow rates (or low liquid loads) the cathode potential might fall into a region where H2O2 is reduced, thus, the current efficiency decreases. For a more complete analysis of the system under investigation, based on the experimental results o f Table 5.4, the main and interaction effects of the variables were calculated with the Jass 2.1® software (Table 5.5). 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 89 Table 5.4 Design matrix and 'responses' for the 2 4 +l factorial runs in 1 M Na2SC>4 (pEL 3.1). T a v g : 303 K . Ac: 33xl0" 4 m 2 . G: 0.47 kg m"2 s"1. Graphite Felt Cathode. Nr. Variable Levels Peroxide Cone. (mM)* Current Efficiency ' ( % ) " I L P s 1 0 0 0 0 30 ±1 53 +2 2 - + + + 22 80 3 - + - + 17 60 4 + + - + 20 40 5 + - - + 46 35 6 + + - - 15 31 7 - + + - 16+1 58 ±4 8 - - - - 28 38 9 - - + - 34 ±0.6 46+1.7 10 + + + - 23 49 11 - - - + 37 49 12 + - + - 36 28 13 + + + + 30 62 14 - + - - 12 44 15 + - + + 45 +1 35 ±1 16 - - + + 43 57 17 + - - - 32 25 Average factorial response: 28.6 mM. Pooled standard deviation: +0.8 mM. ** Average factorial response: 46.5%. Pooled standard deviation: ±1.8%. 5.3 Electroreduction of 02 to H2O2 in Elow-by' Cells Table 5.5 Main and Interaction Effects for the O2 / A336 system. Catholyte: 1 M N a 2 S 0 4 ( p H i n 3.1). Graphite Felt Cathode. E F F E C T S P E R O X I D E cone. (mM)* C U R R E N T E F F I C I E N C Y " (%) M A I N I 4.7 -15.9 L -18.3 13.8 P 5.2 11.4 S 8.0 12.6 Two-factor I N T E R A C T I O N I L 0.5 0.9 T P 0.0 -0.9 I S 0.7 -2.6 L P 1.5 6.9 L S -2.3 2.6 P S -0.3 0.8 Three-factor I N T E R A C T I O N I L P 2.3 2.4 E L S -0.4 -1.4 E P S -0.5 -0.6 L P S 1.0 1.6 * standard error of effect: ±0.4 mM; ** standard error of effect: +3.6% 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 91 Table 5.5 shows that, the main effects are larger than the interaction effects. Regarding the main effects, the liquid load (L) and the current density (I) had opposite influences on peroxide concentration and current efficiency. For instance, as discussed earlier, a 'low' liquid load (1.6 kg m - 2 s"1) and 'high' current density (1274 A m"2) increased the peroxide concentration per pass, whilst high current efficiencies were obtained for 'high' liquid load (4.3 kg m"2 s"1) and 'low' current density (729 A m"2). The main effects of surfactant (S) and pressure (P) on current efficiency are comparable in magnitude. A 'high' A3 3 6 concentration (1.1 mM) increased the current efficiency with 12.6% over the average of 46.5% and similarly, a 'high' pressure (700 kPa) caused an 11.4% increase of current efficiency (Table 5.5). It must be emphasized, however, that the surfactant main effect on peroxide concentration is 54% larger than the equivalent main effect of pressure (i.e. S = 8 m M and P = 5.2 mM) . A closer look at Table 5.5 shows that there is a positive interaction between pressure and liquid load, e.g. for current efficiency L P = 6.9%. The absence of statistically significant two-factor interaction between pressure and surfactant (PS, Table 5.5) suggests that increasing the pressure when 'high' level of A336 is present, wi l l not improve the figures of merit. Furthermore, taking into account that LPS = 1 m M , the PS interaction can be negative for ' low' liquid load. Indeed, comparing for instance runs nr. [5] and [15] in Table 5.4, shows that increasing the pressure from ' low' to 'high' with A3 3 6 present, did not improve the peroxide concentration. A n explanation of this effect, based on capillary flow conditions, is provided in Section 5.3.4. Additionally, the surfactant is involved in a statistically meaningful interaction with liquid load (LS), acting on the H2O2 concentration (Table 5.5). The negative interaction with the liquid load (LS = -2.3 mM) reflects the importance of A336 when the residence time is long (i.e. ' low' flow rate) and high peroxide concentrations are formed per pass. In the latter case, the inhibitory effect of A3 3 6 on peroxide electroreduction plays an important role, as discussed in Section 5.2.2. The curvature effect is 6.5% ±5.7 (based on the current efficiency), indicating a small non-linearity in the influence of I, F, P and S. 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 92 Eliminating the statistically insignificant effects, the results of the factorial design (Table 5.5) can be expressed in terms o f a regression model (Appendix F). In the present case is more useful to develop a regression equation for current efficiency only, since its value is always between 0 to 100% regardless of the cell dimensions, while the peroxide concentration per pass is strongly dependent on the physical dimensions of the electrochemical cell. For instance the same liquid load applied in a longer cell might give a high peroxide concentration that is outside the range of values for which the regression equation was developed based on the results of the factorial experiments (Table 5.5). For peroxide current efficiency, the regression equation is: CEreg (%) = 46.5 + 6.9xL -7.95*,. + 5.1xp +63xs +lA5xLxp, p y j where the coded variables x, are given by: _ L - 2 . 9 5 _ / - 1001.5 _ P- 562.5 _ S - 0 . 5 5 X l ~ 1.35 272.5 ' X p ~ 137.5 '** " 0.55 ; [510] with L: kg m"2 s'1; /': A m"2; P. kPa and S: m M . Alternatively, in the above equations the liquid load L and its coded variable xL can be replaced by either residence time r and xr or by the flow rate Fy and XF. Thus, _ r - 6 0 . 8 4 5 _ Fv - 3.085 x 10 ' 7 X t ~ 27.955 , X p ~ 1.415xl0" 7 ' [ 5 H ] where r : sec. and Fy: m V . Equations [5.9] - [5.11] are valid only for variable values within the range for which the above equations were determined and for cathode dimensions and residence times similar to those of the 'short' cell (see Table 5.3 and Fig. 4.4, respectively). 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 93 5.3.3 Experiments on the 'long' (0.5 m) flow-by cell: Acid electrolyte The factorial experiments performed on the 'short' (0.14 m effective length) cell showed that ca 1 m M concentration of A336 increases the current efficiency and peroxide concentration by 27% on average in acid, for current densities between 729 and 1274 A m"2. Next, experiments were performed on the 0.5 m effective length cell (Chapter 4.2.2, Figs. 4.6 and 4.7), to determine how the scaling of the reactor affects the figures of merit. It was also of interest to investigate the influence o f the cationic surfactant at concentrations above 1 m M , since in the factorial design the highest level of A3 3 6 was 1.1 m M , The 0.5 m long cell was operated at currents up to 54 A . The cathode was graphite felt, with a thickness of 4.5 mm under 38% compression. The catholyte flow rate was 7.17 xlO" 7 m 3 s_1 which corresponds to a liquid load of 2.9 kg m"2 s"1. This value is the same with the liquid load used in the 'centerpoint' experiment of the factorial design (i.e. entry nr. [1], Table 5.4). The O2 flow rate was 10 - 5 m 3 s_1 at STP giving a gas load of 0.058 kg m - 2 s"1. The outlet reactor pressure was 400 kPa. Fig. 5.15 shows that for 1 m M A336 the maximum H2O2 concentration per pass obtained in the 'long' flow cell was 80 m M , at 1050 A nf 2. The corresponding current efficiency is 38.5%. Under comparable conditions, the 'short' (0.14 m) flow cell, yielded 30 m M H2O2, at 53% current efficiency (Table 5.14, entry nr. [1]). Thus, the peroxide concentration does not increase proportionately with reactor length. The longer residence time (i.e., 2.9 min vs. 0.8 min) together with the less uniform potential distribution along the length o f the cell [Davidson et al, 1982], as well as the higher H2O2 concentration, increases the significance of secondary electrode reactions in the 0.5 m long cell, such as H2O2 electroreduction to H2O and/or H2 evolution. 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 94 0. 500 1000 1500 2000 2500 Current density / A rrr2 Figure 5.15 Ac id electroreduction of 0 2 in the 'long' flow-by cell. Influence of A336 concentration. 1 M N a 2 S 0 4 pHj„ 3.1. Ac: 0.027 m 2 . T: 300 K . P o u t : 400 kPa. A concentration of 1 m M A3 3 6 improved the figures of merit for superficial current densities above 500 A m"2 (Fig. 5.15). It is interesting to note that whilst in the 0.14 m ('short') flow cell 1 m M levels of surfactant accounted for ca. 27% improvement in peroxide concentration and current efficiency, in the 0.5 m ('long') cell, 1 m M A336 had a more pronounced influence, yielding for instance 62% higher H2O2 concentration and current efficiency at 1050 A m"2 (i.e. C E = 24% for A336=0; CE=38.5% for A336=l m M , Fig. 5.15). This finding is consistent with the results of the factorial design which showed the positive influence of A336 on peroxide concentration for long residence time (i.e. either ' low' liquid load or longer cell, L S = -2.3 m M , Table 5.5). Furthermore, increasing the A3 3 6 concentration to 3 m M had an interesting effect on the peroxide concentration (Fig. 5.15). A t superficial current densities below 750 A m"2, with 3 m M 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 95 A3 3 6, lower H2O2 concentrations per pass were obtained than even without surfactant. However, for 3 m M A336, the H2O2 concentration increased with current density up to 1500 A m"2, while the H2O2 concentration without surfactant started to decrease at 800 A m"2 or at 1100 A m"2 in the presence of 1 m M A336. The maximum H2O2 concentration with 3 m M surfactant, obtained at 1500 A m " 2 , was 87 m M . It is hypothesized that the surfactant concentration dependence of the current density corresponding to the maximum H2O2 concentration (Fig. 5.15), is due to the interplay between the cathode potential and the surface film formation. It may be that the higher the A 3 3 6 concentration, the more negative potentials (hence, higher current densities) are required to form the ordered surface arrangement that would bring about the desired kinetic effect on the O2 reduction. The outlet p H of the catholyte was higher in the long' (0.5 m) cell, e.g. with 1 m M A336 at ca. 1000 A m" pHout is 5.1, whilst in the 'short' cell pH 0 U t is only 4. Furthermore, due to higher current efficiencies, in the presence of the cationic surfactant the outlet p H was lower. For instance, without A336 at 1000 A m"2 pHout was ca. 9, while with 1 m M A336 is only 5.1. For the range of superficial current densities used (Fig. 5.15), the typical cell voltage values for the runs with 0 and 1 m M A336 were between: 1.1 and 3.5 V . However, in the case of 3 m M A336 the cell voltages were considerably higher, i.e. between 2.5 and 6.2 V . 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 96 5.3.4 Factorial experimental design: Alkaline electrolyte With a catholyte of 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pH = 9.6 / 20 °C), factorial experiments were performed using the 0.14 m long cell, equipped with graphite felt cathode. Again, four variables were selected at two levels, plus one centerpoint (2 4 +l, Table 5.6). The superficial current densities used in the factorial design, were chosen far apart, i.e. the 'low' value of 927 A m"2 is in the kinetic control region, while the 'high' 4650 A m"2 is in the mass transfer limited domain. The ' low' value of the outlet pressure was set to 200 kPa while the 'high' value _ 7 1 1 1 1 1 was 475 kPa. The flow rates were between 1.67x10" m s" and 7.5x10" m s" , respectively, providing a wide range of corresponding liquid loads and residence times (Table 5.6). The complete set of 17 factorial runs are given in Table 5.7. Several factorial points were replicated, usually three times. The standard deviation for peroxide concentration was ± 8 . 1 m M and for current efficiency ±2.1%. The outlet p H varied between 9.7 (for 'high' flow rate coupled with 'low' current density) and 12.5 ('low' flow rate - 'high' current density). The reactor temperature was 308 ± 7 K . Previous research showed that up to 350 K , the figures of merit for the electroreduction of 0 2 to peroxide in alkaline media, are not strongly affected by temperature [Oloman and Watkinson, 1979]. Therefore, it was assumed that the temperature variation in the present study did not influence the outcome of the factorial study. Typical cell voltages were between 2.6 V for 800 A m"2 and 5.7 V for 4000 A m"2. 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 97 Table 5.6 Variables and their levels for the 2 4 +l factorial runs in 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 ( p H i n 9.6). T: 308 ±7 K . Ac: 33x10' 4 m 2 . Gas Load at STP: 0.27 kg m"2 s 1 (linear velocity: 0.19 m s'1). Graphite Felt Cathode. Nr. Variables Levels L o w (-) Center (0) H i g h (+) 1 Current density; I (Am" 2 ) 927 2780 4650 2 Catholyte Load; L (kgrnV) 1.7 5.3 7.8 3 Pressure; P* (kPa) 200 350 475 4 A336 cone; S (mM) 0 1 3 * absolute outlet reactor pressure 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 98 Table 5.7 Design matrix and responses for the 2 4 +l factorial runs in 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pHj„9.6). T: 308 ±7 K . Ac: 33xl0" 4 m 2 G : 0.27 kg m ' 2 s"1 at STP. Graphite Felt Cathode. Nr. Variable Levels Peroxide Cone. ( m M ) Current Efficiency** (%) I L P s 1 - - + + 78 82 2 + + - + 62 59 3 - + - - 18 85 4 - - - - 80 85 5 + - - - 301 ±13 63 ±3 6 + + + - 72 68 7 + - - + 400 ±9 84 ±2 8 0 0 0 0 77 ±2 83 ±2 9 + - + + 350 ±7 74 ±3 10 - + - + 20.5 97 11 - - - + 83 88 12 + + - - 50 ±2 47 ±2 13 + + + + 81 77 14 - - + - 78 82 15 - + + + 21 100 16 - + + - 20 95 17 + - + - 364 ±5 77 ±1 * Average factorial response: 127.8 mM. Pooled standard deviation: ±8.3 mM. ** Average response: 79.2%. Pooled standard deviation: ±2.4%. 5.3 Electroreduction of 02 to H202 in Elow-by' Cells Table 5.8 Main and Interaction Effects for the 0 2 / A3 3 6 system. Catholyte: 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pH i n 9.6). Graphite Felt Cathode. E F F E C T S P E R O X I D E cone. (mM)* C U R R E N T E F F I C I E N C Y " (%) M A I N I 160.2 -20.6 L -174.4 -0.8 P 5.8 5.9 S 15.4 7.2 Two-factor I N T E R A C T I O N IL -113.8 -10.9 IP 7.3 4.9 IS 12.4 2.4 LP 4.6 7.1 LS -7.7 2.1 PS -14.9 -4.6 Three-factor I N T E R A C T I O N ILP 2.3 1.6 ILS -8.1 -1.4 IPS -13.9 -2.1 LPS 14.0 2.1 * standard error of effect: ±9 mM. ** standard error of effect: ±2.7%. 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 100 Table 5.7 shows that in 0.5 M Na2CC>3 / 0.5 M NaHC03 , peroxide concentrations per pass > 300 m M were obtained for ' low' liquid load and 'high' current density (e.g. runs nr. [5], [7], [9] and [17] Table 5.7). The corresponding current efficiencies were above 60%. The main and interaction effects are presented in Table 5.8. The 'high' current (I) and ' low' liquid load (L) both individually and in synergy, increased the peroxide concentration per pass typically with over 100 m M (e.g. I = 160 m M , L = -174.4 m M and EL = -113.8 m M , Table 5.8). Thus, 4650 A m"2 superficial density, associated with the high residence time due to ' low' liquid load (i.e. ca 1.5 min.) is beneficial for peroxide concentration. However, there is a trade-off for lower current efficiency due to the same effects. The cationic surfactant (S) had a significant positive main effect on both concentration and current efficiency, 15.4 m M and 7.2% respectively. Interestingly, the interaction effect between surfactant (S) and pressure (P) is negative for ' low' liquid load (L), e.g. for peroxide concentration PS = -14.9 m M and L P S = 14.0 m M (Table 5.8). Consequently, the highest peroxide concentration per pass, 400 m M , was obtained at 4650 A nf 2 with 3 m M A336 at ' low' (200 kPa) pressure (see run nr. [7] Table 5.7). Under the same conditions but at 'high' pressure (475 kPa), the peroxide concentration was lower, only 350 m M (run nr. [9] Table 5.7). The same P S interaction was observed for the acid electrolyte (Section 5.3.2). To explain the observed phenomena, one has to consider the capillary flow conditions at the G/L /S interface in the presence of surfactant. Both high external pressure and the surfactant, decreases the radius of the gas (i.e. O2) bubbles according to the Young-Laplace equation [Hunter, 1993]. For trickle-bed reactors the gas-liquid interfacial area increases with pressure when the linear gas velocity is greater than 0.03 m s_1 [Gianetto and Specchia, 1992]. In the present case the O2 velocity is 0.19 m s"1. Therefore, the G / L surface area wil l increase, and surfactant molecules wi l l transfer from the bulk liquid to the G / L surface to compensate for the local decrease of the G / L surface excess concentration of surfactant (i.e. local increase in surface tension), (Gibbs effect [Shaw, 1992]). The surfactant accumulation at G / L surface, coupled with a ' low' liquid load (i.e. poor liquid-solid mass transfer), may decrease the concentration of A3 3 6 at the S/L interface. Consequently, 5.3 Electroreduction of 02 to H202 in 'Flow-by' Cells 101 the ability of A3 3 6 to form organized and uniformly distributed S/L surface structures that are effective in enhancing the kinetics of O2 reduction (Chapter 5.1), is diminished. One way to validate experimentally the discussed hypothesis, is to compare the pressure drops through the graphite fiber bed for different conditions, with and without the presence of A336. There is experimental evidence that the pressure drop associated with the movement of gas bubbles in capillaries filled with liquid, increases when surfactant is present. The theoretical explanation relies on the fact that the pressure drop increase is due to the G / L surface tension spatial gradients, which in turn are a reflection of the surfactant surface excess concentration gradients at the G / L interface [Probstein, 1989]. Table 5.9 compares the measured overall pressure drops without and with 3 m M A336, for the same gas load and outlet pressure but for different liquid loads, i.e. 1.7 kg m"2 s"1 and 7.8 2 1 • kg m" s", respectively. Table 5.9 Influence of A33 6 concentration on pressure drop. Graphite felt (0.14 m effective length). 0 2 load. 0.27 kg m"2 s"1. P o u t : 475 kPa. Catholyte: 0.5 M N a H C 0 3 / 0.5 M N a 2 C 0 3 . Nr. Liquid load (kgrnV) Pressure drop, (kPa) N o A336 3 m M A336 1 1.7 25 50 2 7.8 50 85 Table 5.9 shows that the pressure drop for the same outlet pressure and gas load, was 25 and 35 kPa higher in the presence of 3 m M A3 3 6, for the liquid loads used in the factorial 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 102 experiments. Thus, the G / L surface tension gradient hypothesis and its influence on the peroxide electrosynthesis, as previously described, seem plausible. From the interaction effects (Table 5.8), it can be seen that the surfactant interacts positively with current density, IS = 12.4 m M . This means, that the effect o f A336 is more pronounced at high superficial current densities, where the inhibitory effect on peroxide electroreduction (Chapter 5.2.2) might come into play as well. The curvature effect (i.e. difference between centerpoint result and the average of the factorial points) shows a different trend for peroxide concentration than for current efficiency. The curvature on concentration is -50.8 m M ±18.6, whilst on current efficiency is statistically insignificant, i.e. -2 .2% ±5.5. Eliminating the statistically insignificant effects from Table 5.8, the regression equation for current efficiency for the set of conditions employed in the factorial design (Table 5.6), is: CEreg(%) = 7 9 . 2 - 10.3*,. + 2.95x P + 3.6xs - 5.45x,xL + 2.45x,x P +3.55xLxP -2.3xPxs; [5.12] where the coded variables x, xP, xs and xL are expressed as: / - 2788.5 P- 337.5 S-l.5 , Z - 4 7 5 with i: A m"2; P: kPa, S: m M and L: kg m"2 s"1. In the above equations, the coded variable for liquid load can be replaced by the variable for flow rate, whilst the coded variable for total current X/ can be used instead x,, i.e., X / = - — ^ ^ 7 where / : A . 6 1 2 5 [5.14] 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 103 5.3.5 Experiments on the 'long' (0.5 m) flow-by cell: Alkaline electrolyte The performance of the surfactant system was also studied on a longer cell, to evaluate the effect of reactor scaling on the figures of merit. The catholyte was a mixture of 1 M NaHC03 and 0.5 M NaiCCb (pHin 9.4). The reactor equipped with graphite felt cathode (superficial area: 0.029 m 2 , Section 4.2, Figs. 4.6 and 4.7) was operated with a liquid load of 1.55 kg m"2 s"1, which corresponds to a flow rate of 21 ml min"1. The O2 gas load was 0.034 kg r n 2 s"1, supporting a constant outlet reactor pressure of 375 kPa. Superficial current densities between 250 and 2000 A m"2 were employed, for currents of 7.5 to 58 A . Fig. 5.16 shows the dependence of peroxide concentration and current efficiency on superficial current density as a function o f A336 presence. The cationic surfactant improved the figures of merit over the entire range of explored current densities. However, as expected from the factorial experiments, the effect of A336 was more pronounced at high current densities (i.e. above 1000 A m' 2). For instance at 2000 A ni" 2 with 1 m M A336 the peroxide concentration was ca 40% higher, i.e. 465 m M vs. 330 m M . Furthermore, due to the corresponding higher current efficiency (55% vs. 39%), the outlet p H of the catholyte is lower in the presence of surfactant, i.e. 12.4 vs. 13.5, which is an additional benefit for potential subsequent pulp bleaching. 5.3 Electroreduction of 02 to H202 in Elow-by' Cells 104 0 500 1000 1500 2000 Current density / A m - 2 Figure 5.16 Alkaline electroreduction of 0 2 in the 'long' (0.5 m) flow-by cell. Influence of A336 concentration. 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 , p H i n 9.4. A . : 0.029 m 2 GF . T: 300 K . P ( 375 kPa. Liquid load: 1.55 kg r n 2 s"1. Gas load: 0.034 kg i n 2 s'1. 6.1 Exploratory Studies: Composition of the Organic Media 105 C H A P T E R 6: Emulsion Mediated Electroreduction of O2. Results and Discussion 6.1 Exploratory Studies: Composition of the Organic Media The proposed emulsion mediated system for O2 electroreduction to H2O2, was outlined in Chapter 2.2, and general principles of electrosynthesis in L / L systems were presented in Section 3.2.3. For efficient two-phase (L/L) electrosynthesis of H2O2, the physico-chemical properties of the organic solvent are crucial. Besides providing the proper electrochemical environment for reduction of the anthraquinone mediator, the organic solvent has to be chemically and electrochemically stable, it should emulsify easily with the aqueous phase of choice without being miscible and also, it should provide a high H2O2 partition coefficient for the aqueous over the organic phase. These aspects would also simplify the downstream processing, enabling the reuse of the organic phase. These 'Exploratory Studies' were aimed at defining the composition of the organic media to fulfill the above specifications. For this purpose the following experiments were performed: a) determination of the H2O2 partition coefficient between the aqueous phases (i.e. acid or alkali) and various organic solvents, b) measurement of the electrical conductivity of both the organic solvent and emulsion and c) emulsion mediated batch O2 reduction experiments to test the performance of the proposed organic phase. 6.1.1 Aqueous / Organic partition of H 2 0 2 The H 2 0 2 partition (distribution) coefficient between the aqueous and organic phases K, is defined as the peroxide concentration ratio between the two solutions, i.e., j£ _ CH2Q2,AQU [6.1] r ^H202,ORG 6.1 Exploratory Studies: Composition of the Organic Media 106 The distribution law as stated by eqn. [6.1], is strictly valid only i f the solute (i.e. H2O2) undergoes no dissociation or association in each of the two phases [Maron and Prutton, 1971]. Thus, for alkaline aqueous phases with p H e.g. > 11, causing dissociation of H2O2, eqn. [6.1] has to be expressed in terms of the concentration ratio of the species which is common to both phases, i.e. the undissociated H2O2. The partition coefficient was determined for five organic solvents: 1,2-dichloroethane (DCE) , tributyl phosphate (TBP), a 15/85 v mixture of T B P and diethylbenzene ( T B P / D E B ) , propylene carbonate (PC) and 2-undecanone, using three different total H2O2 concentrations, i.e. 0.59, 1.2 and 1.6 M (Appendix G). Figures 6.1 a and b, reveal important differences in H2O2 partition depending on the nature of the organic solvent, H2O2 concentration and aqueous phase composition. Ideally, the partition coefficient K, should be independent of peroxide concentration. Any deviation from single molecule distribution between the phases and/or peroxide consumption by the organic solvent, brings about the variation of the partition coefficient. For tributyl phosphate (TBP), in both acid and alkali, over the entire range of peroxide concentrations K is constant, i.e. 1.4 - 1.5 at 300 K (Fig. 6.1). The constancy of K shows that T B P is stable in the presence of 0.59 M to 1.6 M H2O2 and also, H2O2 is not associated into double (or multiple) molecular aggregates in this solvent. The high permanent dipole of both T B P (juD = 3.07 D [Lange's, 1992]) and H 2 0 2 (JUD = 2.2 D [Lange's, 1992]) supports the measured partition coefficient. For the organic solvent used in previous studies of 2-ethyl anthraquinone electroreduction in alkaline emulsions, i.e. a 15 to 85 volume ratio of T B P and D E B ( [Knarr et al, 1992; Houissoud and Tissot, 1999 (I and II)]), K was generally higher than for T B P alone and it varied significantly with total H2O2 concentration (Fig. 6.1). The higher K is simply due to the presence in the mixture of the non-polar D E B (JUD = 0.4 D [Lange's, 1992]) causing the 'rejection' of the highly polar H2O2. Furthermore, this mixture behaved quite differently when in contact with acid or alkali aqueous phases. In the case of 0.1 M H2SO4 (Fig. 6.1 a), K decreased with total H2O2 concentration from 2.7 (for 0.59 M H 2 0 2 ) to 1.0 at 300 K (for 1.6 M H 2 0 2 ) . This effect is typically the result of 6.1 Exploratory Studies: Composition of the Organic Media 107 association of the solute in the organic solvent [Maron and Prutton, 1971]. Due to the very low permanent dipole of D E B (JUD = 0.4 D [Lange's, 1992]), it is conceivable that the polar H2O2 molecules, especially at high concentrations, are able to increase their solubility in the non-polar organic solvent by forming double or multiple molecular associations*. In the case of 0.1 M Na2C03, on the other hand, K increased with peroxide concentration for the T B P / D E B mixture, from 10 to 21 (Fig. 6.1 b). It is hypothesized that the high aqueous to organic partition coefficient in alkali is due to both the dissociation of H2O2 and to the partial oxidation of D E B at the L / L interface by HC»2~ and other reactive short-lived species that are present in alkaline peroxide solutions (e.g. H O ' , H02*)- The same phenomena might be responsible for the unexpectedly high K values observed for 1,2-DCE and 2-Undecanone (Fig. 6.1). The reactions of H2O2 with a variety of organic compounds such as aromatics, chlorinated and oxygen containing hydrocarbons, is well documented in the literature [Kislenko and Berlin, 1991]. Regarding propylene carbonate (PC), at 1.6 M total H2O2 concentration no measurements could be made in either acid or alkali (Fig. 6.1). It was observed that the opaque emulsion after ca. 5 minutes of mixing turned translucent and homogeneous. This suggests that under the employed conditions at high concentrations of H2O2 (> 1 M ) , P C was oxidized. Ancillary experiments showed that propylene carbonate underwent also alkaline hydrolysis in the presence of 0.1 M N a 2 C 0 3 . Based on the experiments discussed above, it was concluded that among the investigated organic solvents, tributyl phosphate (TBP) had the most promising physico-chemical characteristics for peroxide synthesis. The constancy of K over the range of peroxide concentrations, regardless of the nature of the aqueous phase (i.e. acid or alkali), indicated an uncomplicated behaviour associated with good chemical stability of the solvent. Furthermore, there are no major environmental issues with T B P . This chemical does not significantly bioaccumulate since it is moderately biodegradable [jtbaker.com]. to reduce their overall permanent dipole 6.1 Exploratory Studies: Composition of the Organic Media 108 o CO 5 4 3 2 -1 -1 1 1 1 — a) 0.1 M H 2 S 0 4 103 1 0 2 - 4 101 10° • T B P • T B P / D E B - - 1 T 2 -DGE • P C * 2 - U n d e c a n o n e • 1^1 • • -A---1 1 1 1 1 1 r b) 0.1 M N a 2 C 0 3 A A 0 -| 1 1 1 1 1 1 — 1 1 1 1 1 1 r 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 ~~1 1 1 1 1 1 — r — I 1 I 1 I , — 0.4 0.6 0.8. 1.0 1.2 1.4 1.6 1.8 T o t a l H2O2 c o n e , in e m u l s i o n / M Figure 6.1 The aqueous / organic partition coefficient of H2O2 as a function of the total H2O2 concentration in the emulsion. (Aqueous / Organic) phase volume ratio: 3/1. 300 K . 6.1 Exploratory Studies: Composition of the Organic Media 109 6.1.2 Effect of the cationic surfactant and supporting electrolyte on the conductivity of the organic phase As outlined in Section 2.2, the proposed emulsion mediated system makes use of both surfactant and supporting electrolyte. The main role of the surfactant is to improve the wetting of the cathode surface by the organic media, while the supporting electrolyte provides the necessary electrical conductivity in the organic phase. The long-chain quaternary ammonium salt A3 3 6, was selected as the surfactant of choice for the emulsion mediated system, due to the positive influence of A336 on the figures of merit of the 2e" reduction of 0 2 (Chapter 5). As supporting electrolyte in the organic phase, tetrabutylammonium perchlorate ( T B A P ) was employed. From an electrochemical point of view, for an emulsion the conductivity of the organic phase ideally should be the same order of magnitude as the conductivity of the aqueous phase (i.e. > 1 S m"1). This would reduce the overall energy requirements (i.e. low cell voltage) and also, it would create a favorable current distribution for the electro-organic reaction [Feess and Wendt, 1981]. In the present system, the organic solvent (i.e. TBP) is virtually electrically non-conductive (Fig. 6.2 b), its dielectric constant s, is 7.96 [Lange's, 1992]. However, both the long-chain cationic surfactant A336 and tetrabutylammonium perchlorate enhance the electrical conductivity of T B P . Figure 6.2 b (inside figure), shows that A336 induces a small increase of the conductivity K, up to 0.48 mS m"1 for 0.06 M A336 at 296.5 K . Since for electrosynthesis, A3 3 6 is present in concentrations around 1 m M only, its effect alone on* the conductivity is not sufficient and the addition of the supporting electrolyte ( T B A P ) is required (Fig. 6.2). In the investigated range of T B A P concentrations (i.e. up to 0.2 M , which is near the solubility limit at 296 K ) , the conductivity increased linearly from 0.48 mS m"1 (with 0.06 M A336 only) to 18.2 mS m"1 (Fig. 6.2). A plot of the molar conductivity AM, vs. the square root of concentration reveals the mechanism of ionic conduction by the quaternary ammonium ions. For T B A P especially, the molar conductivity AM, decreased abruptly with concentration (Fig. 6.3). This is generally described as being due to incomplete dissociation of the solute and increase in viscosity (i.e. low ionic mobility) [Abott et al, 1997]; [Prentice, 1991]. 6.1 Exploratory Studies: Composition of the Organic Media 1 20 15 0 K = 0 . 6 4 + 9 0 C T B A P ; R 2 = 0.99 E / U 5 - i CO 10 -0.4 -E 0.3 -— 0.2 -/ E 0.1 -5 -0.0 -1 0.00 b) T T 0.00 0.02 0.04 0.06 A336 concentration / M T 0.25 0.05 0.10 0.15 0.20 T B A P concentration / M Figure 6.2 The conductivity of tributyl phosphate (TBP) as a function of supporting electrolyte ( T B A P ) and cationic surfactant concentration (A336; [inside figure b]).296 K . 500 E CO E < 400 300 -200 -100 -0 • A 3 3 6 > T * * ' f M — r -0.0 0.1 0.2 0.3 C 1 / 2 / M 1 / 2 0.4 0.5 Figure 6.3 Molar conductivities for T B A P and A336 in T B P vs. the square root of concentration at 296 K . 6.1 Exploratory Studies: Composition of the Organic Media 111 6.1.3 Electrical conductivity of the emulsion: Determination of the continuous phase In addition to the conductivity of the organic phase, it is of interest to investigate the conductivity of the emulsion, which is the catholyte for H2O2 electrosynthesis. Furthermore, the conductivity measurements could reveal the nature of the emulsion, i.e. 'oil in water1, 'water in oil 1 or bicontinuous emulsion. The organic phase was T B P with 0.2 M T B A P and 0.06 M A336. Also, 0.1 M 2-ethyl anthraquinone (EtAQ) was added to the above organic solution, to 'simulate', as closely as possible, the organic catholyte to be used during electrosynthesis. The aqueous phases were 0.1 M H2SO4 and 0.1 M Na2C03, respectively, in a 3 to 1 volume ratio vs. the organic media. The temperature was 296 K . The conductivities of the emulsion, and each of the two-phases, were followed over a 10 minute period. The results are shown in Figs. 6.4 a and b. The conductivity of'pure' 0.1 M H 2 S O 4 is more than 3 times higher than the conductivity of 0.1 M N a 2 C 0 3 , i.e. 5.48 S m"1 vs. 1.58 S m"1 at 296 K . Therefore, the acid emulsion had an overall higher conductivity than the alkali emulsion, i.e. 2.27 S m"1 vs. 1.07 S m"1, respectively, after 10 min. of mixing at 296 K (Fig. 6.4). The conductivity of the organic phase increased slightly during the first 30 seconds of emulsification from 0.0163 S m"1 to 0.028 - 0.032 S m"1, then leveled off. Presumably, at the start of the emulsification the organic phase picks up a small quantity of aqueous electrolyte, since the solubility of water in T B P is about 0.6% v o i (Table 4.4). The type of emulsion which is formed when two, relatively immiscible, liquids are brought into contact, depends on the volume ratio of the two phases and on the 'hydrophile-lipophile' balance of the emulsifying agent [Shaw, 1992]. The latter property can be estimated from the so-called surfactant number Ns (eqn. [3.14], Section 3.2.1). For A336, [CH3(CH2)7]3CH3N +], substituting the constants into eqn. [3.14] and using for the head group area ah a value of 0.6 nm 2 (based on studies of double-chain quaternary ammonium surfactants [Evans and Wennerstrom, 1994]), one obtains Ns 104. According to the correlation of Ns with the surfactant system architecture, for Ns ~ 1, the surfactant organizes preferentially into bilayers, which can lead to bicontinuous structures (see Table 3.3). However, using the 'phase volume ratio' criteria [Shaw, 1992], one should conclude that in the present case with a 3 to 1 volume ratio of aqueous to organic phases, the emulsion 6.1 Exploratory Studies: Composition of the Organic Media 112 nature has to be 'oil-in-water' (OAV), with the aqueous phase being continuous. The conductivity measurements presented in Fig. 6.4 can help in resolving the apparent contradiction in predicting the emulsion type, between the 'phase volume ratio' and the 'surfactant architecture criteria'. The knowledge of the emulsion type is important for a proper understanding and interpretation of the electrochemical data obtained for the two-phase system. If the aqueous phase is continuous and the organic is uniformly dispersed in it, the conductivity of the emulsion should respect the Maxwell equation, which for this case is more appropriate in its complete form [Sigrist et al, 1980; original reference therein]: K K. aqu em K'org + 2 K - a q u 2f(Kaqu rCorg) Korg + 2*f + f{^aqu ~ Korg) [6.2] where / is the volume fraction of the dispersed (organic) phase, K is conductivity (S i n 1 ) and the subscripts em, aqu and org refer to emulsion, aqueous (either 0.1 M H 2 S O 4 or 0.1 M Na2C0a) and organic phases, respectively. 6.1 Exploratory Studies: Composition of the Organic Media 113 E o Z5 ~o c o O 101 10° 10 ,-1 d Aqueous phase: acid A A A A Emulsion a) 1 Organic phase 10-0 101 ^ 200 400 600 10° 10-1 -4 Aqueous phase: alkali A A A A • • • • Emulsion b) 10-Organic phase 0 200 400 600 Time / s Figure 6.4 Conductivity of the emulsion and each of the two-phases for a typical catholyte composition: Organic phase: T B P , 0.2 M T B A P , 0.06 M A336, 0.1 M E t A Q . Aqueous phase: a) 0.1 M H 2 S 0 4 ; b) 0.1 M N a 2 C 0 3 . Aqueous / Organic phase volume ratio: 3 / 1. 296 K . 6.1 Exploratory Studies: Composition of the Organic Media 114 Table 6.1 shows a comparison between the calculated (eqn. [6.2]) and measured (Fig. 6.4) emulsion conductivities. Table 6.1 Measured and Calculated Emulsion Conductivities. Mixing time: 10 min, T: 296 K . Aqueous phase Composition / Kaqu (S m"1) (S m"1) Km (S m"1) Measured Calculated (eqn. [6.2]) 0.1 M N a 2 C 0 3 0.25 1.69 0.028 1.07 1.14 0.1 M H 2 S 0 4 0.25 5.09 0.032 2.27 3.41 In the case of the alkali aqueous phase there is good agreement between the measured emulsion conductivity and that calculated with the Maxwell equation (Table 6.1). Thus, the model of continuous aqueous phase with dispersed organic droplets, is valid for the alkali (0.1 M N a 2 C 0 3 ) emulsion (i.e. O A V type). With 0.1 M H 2 S04 , however, the calculated conductivity is 50% higher than the measured (Table 6.1), which shows that the O A V model does not hold for the acid emulsion. Furthermore, the acid emulsion conductivity (2.27 S m"1) is close to the average conductivity between the two phases, i.e. 2.56 S m' 1 . This means that both 'oil' and 'water' act as continuous phases, in other words, the acid emulsion can be considered bicontinuous or 'dual' (Section 3.2.1; Figs. 3.2 and 3.3). Thus, the structure of the acid emulsion is in accordance with the 'surfactant number' criteria. The difference in emulsion type between the 0.1 M H/2SO4 and 0.1 M N a 2 C 0 3 aqueous phases, is intriguing. A n explanation can be attempted based on the variation of surfactant number Ns and surfactant curvature with the counter-ions present in the electrolyte, i.e. C F , 6.1 Exploratory Studies: Composition of the Organic Media 115 and/or S042~, C032~. The OAV emulsion formed with an alkali aqueous phase, indicates that under those conditions Ns becomes less than 1 (see Table 3.3). Because in the Ns formulae (eqn. [3.14]) both Vhc and Lhc are constants, Ns can be lowered only i f the area per head group a/,, increases. A higher a/, occurs when there is strong electrostatic repulsion between the head groups (i.e. they cannot get close together). Therefore, the surfactant film curves toward the 'oil' phase incorporating the organic media inside the micelles, forming an OAV emulsion [Evans and Wennerstrom, 1994]. In 0.1 M Na2C03, the natural counter-ion C F of A336 is partially replaced by C032~ which has a larger radius of hydration than C F (4.5 A vs. 3 A , respectively [Lange's, 1992]. Hence, for C03 2 - the hydration shell is compact, tightly bound, thus C0 3 2 " is less effective in shielding the electrostatic repulsions between the head groups. Consequently, in the presence of carbonate, A336 curves toward the ' o i l ' phase, forming OAV micelles, thus the close packing in a planar bilayer is avoided. In 0.1 M H2SO4 the natural counter-ion C F is partially replaced by S042~ which has a hydration radius in between Cl~ and CO32"" [Lange's, 1992], Consequently, a close surfactant packing in a planar bilayer structure it is still possible, leading to the formation of 'oil' channels and a bicontinuous emulsion structure. The electrical conductivity studies revealed the intricate nature of the emulsifying process and its complex dependence particularly on the electrolyte composition and concentration, as well as the surfactant type. A n additional factor that was not investigated here is the effect of temperature on emulsification. Generally, the overall influence of temperature is determined by two competitive effects: a) raising the temperature causes an increase of the electrostatic free energy, stronger repulsion between the head groups, that leads to an enhanced curvature of the thin surfactant film toward the 'oil' phases, i.e. promotes OAV emulsification and b) the higher the temperature the more disorganized, coiled, the surfactant molecular structure, that causes a negative curvature of the surfactant film (oriented toward the aqueous phase) promoting W/O emulsification [Evans and Wennerstrom, 1994]. For ionic surfactants the electrostatic contribution to the overall temperature effect seems to dominate [Evans Wennerstrom, 1994], which means increasing the temperature wi l l favor OAV emulsification. 6.1 Exploratory Studies: Composition of the Organic Media 116 6.1.4 Preliminary electrosynthesis experiments Exploratory emulsion mediated H2O2 electrosynthesis experiments were performed in a batch, 'FT- cell (Chapter 4), using 0.1 M H2SO4, as aqueous phase. The aim was to compare the performance of the organic media proposed in the present work*, with the 'no surfactant, no supporting electrolyte' organic phase**, first put forward by Knarr et al. [1992] and recently used by Houissoud and Tissot [1999 I and II] for peroxide electrosynthesis in 2 M N a O H . The electrochemical mediator for both systems was 2-ethyl anthraquinone (EtAQ) but in different concentrations, 0.1 M and 0.42 M , respectively. The latter concentration is near the solubility limit at 300 K for the respective organic solvents. The organic to aqueous (0.1 M H2SO4) phase volume ratio was 1 to 3. Oxygen was continuously purged into the emulsion at atmospheric pressure. The cathode was a 100 ppi reticulated vitreous carbon ( R V C ) (with a superficial area of ca 10 cm 2), operated at a constant superficial current density of 300 A m"2. The temperature was kept constant at 300 K . The results are presented in Fig. 6.5. The peroxide concentration in the emulsion and the current efficiency were higher for the surfactant - supporting electrolyte organic phase proposed in the present study, despite more than four times lower mediator concentration. Without quaternary ammonium ions, i.e., D E B / T B P organic media, the maximum H2O2 concentration in emulsion was only ca 100 m M while the present system yielded 275 m M H2O2 (Fig. 6.5). For the latter organic phase, the measured cathode potentials (Ec) were less negative over time than in the case of the emulsion formed with D E B / T B P . For instance, in the presence of quaternary ammonium ions ( T B A P and A336), Ec was around -1.5 VAg/Agci, whilst for D E B / T B P the cathode potential dropped below -2 VAg/A g ci after 2 hrs. A t potentials more negative than -2 V the rate of the secondary cathode reaction of H2O2 reduction is significant, reducing the accumulation of H2O2 in the batch cell after 2 hrs. (Fig. 6.5). Ancillary experiments revealed that at superficial current densities > 500 A m"2 the D E B / T B P system cannot be used for H 2 0 2 synthesis in 0.1 M H2SO4, since the peroxide concentrations were extremely low, e.g. at 500 A m"2 after 2 hrs., only 8 m M H2O2 was obtained with 5% current efficiency. * i.e. TBP / TBAP 0.2 M / A336 0.06 M ** i.e. DEB (85% )^ / TBP (15%ol) 6.1 Exploratory Studies: Composition of the Organic Media 117 In a final assessment between the two organic phases under investigation, taking into account the electrochemical performance (Fig. 6.5), H2O2 partition behaviour between the aqueous and organic phases and the environmental aspects (Section 6.1), it is concluded that the T B P / T B A P / A 3 3 6 system is more suitable for a versatile, on-site emulsion mediated H2O2 electrosynthesis process, than the DEB(85% V ) / TBP(15% V ) mixture. Therefore, the former was retained for further studies of the mediation kinetics and peroxide electrosynthesis. 6.1 Exploratory Studies: Composition of the Organic Media 118 300 - i r 1 , 1 r 0 2 4 6 8 10 Time / hr Figure 6.5 Comparison between two organic phases for emulsion mediated H2O2 electrosynthesis in 0.1 M H 2 S 0 4 . O T B P / T B A P 0 .2M / A336 0.06 M / E t A Q 0.1 M • D E B / T B P / E t A Q 0.42 M . Cathode: R V C 100 ppi. 300 A m"2. 300 K . 0.1 M P a 0 2 . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 0, 119 6.2 On the Mechanism of the Emulsion Mediated O2 Electroreduction In spite of the impressive number of studies in the literature on the reduction of quinone compounds in various organic solvents, investigations that are aimed at the quinone mediated two-electron reduction of O2 are few, and not entirely applicable to the emulsion systems explored in the present work [Keita and Nadjo, 1983; Tissot and Huissoud, 1996; Huissoud and Tissot, 1998; 1999 (I and II)]. Therefore, spectrophotometric and cyclic voltammetry investigations were performed to study the mechanism of E t A Q reduction and to obtain pertinent kinetic parameters, under conditions that are relevant to the proposed peroxide electrosynthesis. 6.2.1 Cyclic voltammetry of 2-ethyl-9,10-anthraquinone in the absence of 0 2 To gain insights into the electrochemical mechanism of H2O2 formation by the E t A Q mediated system, cyclic voltammetry experiments were performed. First, the main components of the organic phase (i.e. E t A Q and O2) were studied individually and finally, the complete system, i.e. with both E t A Q and O2, was subjected to voltammetric analysis. The working electrode was a mini glassy carbon disk (1 mm radius), a platinum wire acted as counter-electrode whereas a mini Ag/AgCl s atd.Kci electrode was used as reference (see Chapter 4). Two main issues had to be addressed, i.e. the role of quaternary ammonium ions ( T B A P and A336) and the influence of 'strong' and 'weak' proton sources in contact with the organic phase (i.e. 0.1 M H 2 S 0 4 and 0.1 M N a 2 C 0 3 , respectively). a) The Role of Quaternary Ammonium Ions on the Electrochemical behaviour of EtAQ in Aprotic TBP Cyclic yoltammograms, for 1.1 m M E t A Q dissolved in T B P , were recorded at 295 K using two different supporting electrolytes, 0.1 M L i C 1 0 4 and 0.1 M T B A P / 9 m M A336, respectively (Fig. 6.6 a and b). L i C 1 0 4 was chosen for comparative purposes, since L i + was reported to have a strong tendency to form ion-pairs with the quinone radical-anions [Pletcher 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 120 and Thompson, 1998]. The cyclic voltammetry was performed in a deoxygenated medium ( N 2 purge for 2 hrs.), under N 2 atmosphere. Also, a bed of 'oven-dry' molecular sieves was added to the cell, to minimize the water content of the organic solvent. In the absence of quaternary ammonium ions (i.e. with 0.1 M L i C L C ^ ) three waves can be identified on both the anodic and cathodic scans (Fig. 6.6 a). There are two main peaks, b and c, at -0.86 and -1.15 VAg/Agci, respectively, and a shoulder-like wave, a, with a cathodic half-wave potential at -0.6 VAg/Agci (Fig. 6.6 a). On the anodic scan, with increasing scan rate, the wave a develops into a peak shape (curve a'; Fig. 6.6 a), which indicates that it is due to surface adsorbed species [Brown and Sandifer, 1986], most likely an adsorbed product of the E t A Q reduction [Pletcher and Thompson, 1998]. The two main peaks b and c, obtained with both supporting electrolytes (Figs. 6.6 a and b), are well documented in the literature (e.g. [Babaei, et al 1997]). They belong to the successive le" reductions: E t A Q / E tAQ - * (b) and E tAQ - * / E t A Q 2 " (c), respectively. The second step leading to the dianion, is reversible for both cases, since the anodic and cathodic peak separation is 60 m V (Figs. 6.6 a and b). Comparing Figs. 6.6 a and b, it can be seen that in the presence of quaternary ammonium salts instead of LiC104, the second reduction step (wave c) occurs at a more negative potential, i.e. -1.38 vs. -1.15 V . In other words, the cathodic peak separation between the second (c) and first reduction steps (b) is much larger with T B A + than L i + , i.e. -0.48 vs. -0.29 V . The difference in peak separation is due to the dissimilar ion-pairing capacity of L i + and T B A + , respectively, with the reduced quinone species [Pletcher and Thompson, 1998; Blankespoor et al, 1987]. Ultimately, the ion-pairing influences the disproportionation equilibrium given by: C • C 2 E t A Q - * <-> E t A Q 2 - + E t A Q ; JTD = ——^f-. [6.3] (^ EtAQ - ) It can be shown [Revenga et al, 1994] that the equilibrium constant KD is related to the standard reduction potentials for E t A Q / EtAQ"* , Eb,o (wave b, b') and EtAQ"* / E t A Q 2 - , Ec,o (wave c, c'), i.e. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 121 m ^ r > = - ~ ( E b 0 ~ E c 0 ) . [6.4] Based on eqn. [6.4], the more negative the peak potential difference between the second and first reduction steps, the smaller KD, which indicates greater stability for EtAQ"". Approximating the standard potential difference (Eb,o - Ec,o), with the difference in peak potentials (Eb,P - Ec,p), from eqn. [6.4] KD at 295 K for L i + is l . l x l O " 5 while in the presence of quat+ KD is equal to 6.3xl0" 9. Thus, the disproportionation of the radical-anion EtAQ" ' is greatly reduced by the presence of quaternary ammonium ions as opposed to L i + . In other words, the dianion E t A Q 2 " is not stable in the T B P / T B A P medium, therefore, the equilibrium [6.3] is shifted toward the radical-anion. Pletcher and Thompson reported also, that the extent of ion-pairing for various benzo- and naphtoquinone dianions increases from T B A + to L i + [Pletcher and Thompson, 1998]. Similar electrochemical behaviour to the one presented in Figs. 6.6 a and b for E t A Q , was reported for 1-methoxyanthraquinone as well, with regard to both LiC104 and T B A P [Blankespooretal., 1987]. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 122 Figure 6.6 Influence of supporting electrolyte on the E t A Q cyclic voltammetry in aprotic T B P . a) 0.1 M L i C 1 0 4 , scan rate (mV s"1): (1) 20, (2) 50, (3) 100; b) 0.1 M T B A P / 9 m M A336, scan rate: 20 m V s"1. E t A Q concentration 1.1 m M . N2 purge. G C electrode. 295 K . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 123 b) Influence of the acid and alkali aqueous phases on the electrochemical behaviour of The influence of 0.1 M H2SO4 and 0.1 M Na2C03 on the electrochemical behaviour of E t A Q in the organic phase, was investigated. The organic media was composed of 1.1 m M E t A Q , 0.1 M T B A P and 9.6 m M A336 dissolved in T B P . The aqueous and organic phases, in a 1/1 volume ratio, were brought into contact by purging N2 for about 30 min. After separation of phases, the cyclic voltammogram of the organic layer was recorded under N2 atmosphere (Figs. 6.7 a and 6.7 b, respectively). . Figure 6.7 a, shows that in the presence of 0.1 M Na2C03 the two successive main waves, discussed in the previous Section, are still detectable. However, a comparison of Figs. 6.6 b (i.e. in the absence of alkali) and 6.7 a, reveals that both reduction peaks are shifted to more positive potentials after contact with alkali, the first peak is at -0.82 V while the second one occurs at -0.98 V. Furthermore, the separation between the two reduction peaks, corresponding to the successive le" reductions, is smaller after contact with 0.1 M Na2C03 (Fig. 6.6 b vs. Fig. 6.7 a). Thus, the electrochemical behaviour of E t A Q in contact with 0.1 M Na2C03 resembles rather the L i C 1 0 4 case presented in Fig. 6.6 a. This means that the reduced E t A Q species wi l l preferentially form ion-pairs with Na+ and / or H + from H2O instead of quat+. The extent of ion-pair formation between the various quinone dianions and Na+ is very strong [Pletcher and Thompson, 1998] consequently Na+ stabilizes E tAQ 2 " . As a result, the disproportionation equilibrium of EtAQ"" (eqn. [6.3]) is shifted in favor of E t A Q 2 " formation, as indicated by the small peak separation and a higher KD than in the absence of alkali, i.e. 1.9xl0"3 vs. 6.3xl0" 9, respectively. Based on the cyclic voltammetry experiments, it can be concluded that the mechanism of E t A Q reduction in the presence of 0.1 M Na2C03 is described by eqns. [6:5] - [6.6]: EtAQ E t A Q + le" g " * * ^ ^ ' E t A Q * [ N a + ] + le" >EtAQ"*[Na + ] > E t A Q 2 [ N a + ] 2 . [6.5] [6.6] 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 124 Section 6.2.4 and Fig. H-3 (Appendix H), presents spectrophotometry data supporting the above mechanism. When the organic phase contacted 0.1 M H 2 S 0 4 , the electrochemical behaviour was altered significantly compared to the one obtained in the absence of any proton sources (compare Figs. 6.7 b and 6.6 b). Only a single, overall 2e", wave is obtained, with the cathodic peak shifted to a more positive potential, i.e. -0.44 V . The absence of a second reduction peak is due to the protonation of E t A Q " followed by the fast disproportionation of E t A Q H ' according to eqns. [6.7] - [6.9] [Babaei et al, 1997; Pekmez et al., 1993; Tissot and Huissoud, 1996]. The spectrophotometric studies presented in Section 6.2.4 and Appendix H, support also this mechanism. E t A Q + l e ' q u a t + > E t A Q " E t A Q * + H + -> E t A Q H * 2 E t A Q H * - > E t A Q H 2 + E t A Q The anodic peak (Fig. 6.7b) appears only at +0.20 V and it is typical for to the oxidation of the fully protonated quinone dianions [Pekmez et al, 1993], i.e. E t A Q H 2 for the present case. Similar behaviour was reported for the E t A Q in 1,2-dimethoxyethane with HCIO4 at p H 1.5 [Tissot and Huissoud, 1996]. Regarding H 2 evolution from the acid dissolved in the organic phase, it was observed that it occurs only at much more negative potentials than the E t A Q reduction, i.e. Em.evoi < -1.3 V A g/AgCl . [6.7] [6.8] [6.9] 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of O, 125 < -1.4 - 0 . 7 + 1 . 2 - 1 . 4 - 0 . 7 + 0 . 9 + 0 . 5 0 - 0 . 5 E / V v s . A g / A g C l - 1 . 0 Figure 6.7 Influence of the acid and alkali aqueous phases on the electrochemical behaviour of E t A Q in the absence of 0 2 . Organic phase after 30 min. contact with: a) 0.1 M N a 2 C 0 3 ; b) 0.1 M H 2 S 0 4 . Scan rate: 100 m V s"1. Organic phase composition: 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336 in T B P . N 2 atmosphere. G C electrode. 295 K 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 126 c) Kinetic parameters for EtAQ reduction in TBP / TBAP, A336 Pertinent electrode kinetic parameters for the two cases of E t A Q voltammetry presented in Fig. 6.7 were obtained, from the scan rate dependence of the cyclic voltammograms. For quasi-reversible charge transfer processes, the standard heterogeneous rate constant ks, can be calculated from the difference between the cathodic and anodic peak potentials, AEP =Epc - EPA, by Nicholson's kinetic parameter f i n conjunction with the Russel - Jaenicke approximation for AEP, [Bard and Faulkner, 1980; Posdorfer et al, 1991]: (D ID Y'2 ¥ = / ox r e d ) ,mks, [6.10] {DoxnvnF IRTf2 J and = 8 1 . 3 4 - 3 . 0 6 ^ +149VP~ 1 / 2 -US^1'3,(mV), [6.11] where Dox; Dred diffusion coefficients for the oxidized and reduced E t A Q species (m 2 s_1),v scan rate ( V s"1), a charge transfer coefficient, ks standard heterogeneous rate constant, (m s"1), AEP =Epc -EPia, (mV), n, nr. of electrons involved in the charge transfer step. For quasi-reversible and irreversible processes the charge transfer coefficient a, can be estimated from the shape of the cyclic voltammogram according to the equation [Brown and Sandifer, 1986]: EP,C-Ep,2,C= [6.12] where Epn,c cathodic half-peak potential. The diffusion coefficients in T B P at 295 K for the species involved (i.e. E t A Q , E t A Q H 2 , E t A Q , and E t A Q " ) were estimated using the Stokes-Einstein equation, from the diffusion 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 0: 127 coefficients in acetonitrile (AN) at 298 K for A Q , A Q 2 " and AQ"* reported in the literature [Posdorfer etal, 1991]*. Thus, D — D ^AN -'TBP •^i, TBP -^I , A N ' j, MTBP - " A N [6.13] where A , T B P and A , A N diffusion coefficient of a certain species /', in T B P and A N , respectively, / / T B P , / / A N dynamic viscosities for the two solvents and 7 V R P , 7 A N corresponding temperatures. For / / T B P = 4.8xl0" 3 Pa s (295 K ) and / / A N = 0.35xl0" 3 Pa s (298 K ) [Lange's, 1992], using DAQ, A N = 2.09xl0" 9 m 2 s\ DAQ~, A N = 1.55xl0"9 m 2 s"1 and DAQ", A N = 1.17xl0"9 m 2 s"1 measured at 298 K [Posdorfer et al, 1991] the relevant diffusion coefficients at 295 K in T B P are: DEIAQ -1.5xl(r 1 0 m 2 s"1; DE(AQ = 1.12xl0" 1 0 m 2 s"1 and DEUQ2' = 0.84xlCr 1 0 m 2 s"1. To apply eqns. [6.10] - [6.12] for the determination of ks, cyclic voltammograms were recorded at up to seven different scan rates between 0.02 V s"1 and 0.2 - 0.3 V s"1. Appendix / shows the cyclic voltammograms and the experimental data (i.e. AEP) employed in calculating ks. From the experimental AEP values obtained for different scan rates, ! f was calculated by solving the non-linear equation [6.11]. Once ^ w a s obtained, ks was calculated from eqn. [6.10]. Table 6.2 presents the standard heterogeneous rate constants for E t A Q reduction in the presence of both 'strong' and 'weak' proton sources, i.e. 0.1 M H2SO4 and 0.1 M Na2C03. * It was assumed that the diffusion coefficients of the EtAQ and AQ species are basically the same. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 128 Table 6.2 Standard heterogeneous rate constants ks, for E t A Q reduction in the presence of'strong' and 'weak' proton sources. Organic media: 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336 in T B P . Electrode: Glassy Carbon. T: 295 K . N 2 purge. Aqueous Electrolyte* k s , i x l O ' ^ m s " 1 ) k S ; 2 xlO^fms" 1 ) 0.1 M N a 2 C 0 3 1.42 ± 0 . 0 9 (a} = 0.67) 0.99 ±0.15 (a2 = 0.24) 0.1 M H 2 S 0 4 0.05 ±0.003 ( a =0.47) — * Organic / Aqueous phase volume ratio: (1 /1) Unfortunately, there are no literature data on E t A Q or A Q reduction in either aprotic or pro tic T B P to compare with the ks values from Table 6.2. Generally, the kinetic parameters for the electroreduction of quinones are very sensitive to the nature of the electrode material and the solvent employed [Nagaoka and Okazaki, 1985]. Table 6.2 shows that in the presence of 0.1 M N a 2 C 0 3 , ks for the reduction of E t A Q is about 30 times higher than in the case of 0.1 M H 2 S 0 4 aqueous phase. Thus, the presence of a 'strong' proton donor (i.e. 0.1 M H2SO4) slows the electroreduction of E t A Q to EtAQ" ' while it accelerates the disproportionation of the radical-anion, as discussed earlier. Furthermore, with alkali present, the first reduction step is faster than the second one, i.e. kSil = 1.42xl0"5 m s"1 and k,2 = 0.99xl0" 5 m s"1. The same behaviour was reported for the reduction of A Q in acetonitrile on a Pt electrode [Posdorfer et al, 1991]. ( 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of O. 129 6 . 2 . 2 The influence of proton sources on the electrochemistry of O2 in tributyl phosphate In addition to the E t A Q mediator, the electrochemistry of O2 in the two-phase system plays an important role in the H2O2 electrosynthesis. The electroreduction of O2 in aqueous electrolytes and the influence of surfactants were discussed in detail in Chapter 5. In the present Section, the electrochemical behaviour of O2 in the organic phase (i.e. T B P ) is described. The aim was to determine the influence of the aqueous phase on the O2 / T B P cyclic voltammograms. Three cases were considered: a) aprotic T B P , b) T B P / 0.1 M N a 2 C 0 3 and c) T B P / 0.1 M H2SO4. The supporting electrolyte was 0.1 M T B A P . For the aprotic solvent case, T B P was contacted with a bed of oven-dry molecular sieves (see Chapter 4). In all cases the solution was saturated with O2 and the experiments were performed under an O2 atmosphere. Following the separation of phases, the cyclic voltammogram of O2 was recorded from the organic layer on a mini glassy carbon electrode at 295 K . Fig. 6.8 shows the O2 cyclic voltammograms for the all three scenarios. In the absence of aqueous phase (Fig. 6.8 a), the cyclic voltammogram is characterized by a cathodic wave occurring at -1.16 V , and a well-defined anodic peak at -0.7 V . The cathodic wave resembles a typical, kinetically controlled (Tafel like) i-E polarization curve. The anodic peak is well documented in the literature [Sawyer, 1991] and it belongs to the superoxide ion O2"', formed by the one-electron reduction of O2 on the cathodic scan. Furthermore, the formation of ion-pairs between the quaternary ammonium ion of the supporting electrolyte ( T B A + ) and O2" was proposed, in order to explain the stability of the latter ion in organic media [Vasudevan and Wendt, 1995; Nekrasov and Vykhodtseva, 1995], When the O2 saturated organic phase was in contact with either 0.1 M H2SO4 (Fig. 6.8 b) or 0.1 M Na2CC>3 (Fig. 6.8 c), the cyclic voltammograms changed significantly. With acid, a well-defined cathodic peak appears at -1.03 V , but without the corresponding anodic wave. Furthermore, the cathodic peak is preceeded by a sigmoid wave with a half-wave potential of -0.58 V . Similar behaviour was reported for the O2 cyclic voltammogram in acetonitrile, in the presence of various concentrations of H2SO4 [Radyushkina et al, 1992]. The overall height of the cathodic peak is 100% larger than in the aprotic T B P , (compare Figs. 6.8 a with Fig. 6.8 b), 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 130 which is most probably due to the increase of the overall number of electrons transferred from one to two [Sawyer, 1991]. It is generally accepted that in protic organic solvents, with moderate or strong proton donors, the reduction of 0 2 yields H 2 0 2 as the major product [Sawyer, 1991]. However, the exact mechanism and the role of protons in facilitating the overall 2e" reduction of 0 2 instead of the le" process, is still not clearly established [Radyushkina et al, 1992; Nekrasov and Vykhodtseva, 1995]. Based on literature data [Radyushkina et al, 1992], the following mechanism is proposed for the reduction of 0 2 on glassy carbon from T B P after 30 min. contact with 0.1 M H 2 S 0 4 (Fig. 6.8 b): Sigmoid wave ( E i / 2 = -0.58 V ) : 0 2 + le" -> 0 2~* H + > H 0 2 ' ; [6.14] Cathodic peak ( E p =-1.03 V ) : H O / + le ' H + > H 2 0 2 . [6.15] Furthermore, the disproportionation of H 0 2 * has to be taken into account, i.e., 2 H 0 2 * - > H 2 0 2 + 0 2 . [6.16] In the case of 0.1 M Na2CC>3 aqueous phase (Fig. 6.8 c), a single, sigmoidal, cathodic wave was observed, with a plateau at -1.36 V . Also, the height of this peak is similar to the aprotic T B P case (compare Figs. 6.8 a and c). Therefore, it is hypothesized that with the alkali aqueous environment 0 2 is reduced in a one-electron process to superoxide CV*, as in the aprotic case. However, since there was no anodic peak corresponding to 02"" (Fig. 6.8 c), it is assumed that the superoxide ion disproportionated in the presence of water, according to eqn. [6.17]: 0 2~* + H 2 0 - > 1 / 2 H 2 0 2 + l / 2 0 2 + O H [6.17] 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of02 131 0 - 0 . 5 - 1 . 0 - 1 . 5 E / V v s . A g / A g C I Figure 6.8 Influence of 'strong' and 'weak' proton sources on the electrochemical behaviour of 0 2 in the organic phase, a) aprotic organic phase; b) after 30 min. contact with 0.1 M H 2 S 0 4 ; c) after 30 min. contact with 0.1 M N a 2 C 0 3 . Scan rate: 20 m V s"1. Organic phase composition: 0.1 M T B A P in T B P . 0 2 atmosphere. G C electrode. 295 K . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of02 132 6.2.3 Cyclic voltammetry of the 2-ethyl anthraquinone - O2 system In the previous two sections (i.e. Sections 6.2.1 and 6.2.2) the main components of the mediator system, E t A Q and O2 respectively, were studied individually. From the point of view of peroxide electrosynthesis, it is also of interest to understand the electrochemical behaviour of the overall system. The changes in the E t A Q voltammograms with increasing concentrations of 0 2 were investigated by cyclic voltammetry (Figs. 6.9 and 6.10) Fig. 6.9 shows the first-scan voltammogram for the E t A Q - O2 system when the organic phase was in contact with 0.1 M H2SO4. After a very short period of 0 2 purge (i.e. 15 sec, curve 2, Fig. 6.9 a) the cathodic peak current, increased ca. 2.4 times in comparison to the case without 0 2 (compare curves 1 and 2, Fig. 6.9). On the other hand, the height of the anodic peak at +0.2 V , given by the electrochemical oxidation o f E t A Q F b , diminished, as a result of the chemical reaction between EtAQH2 and O2. When O2 was purged for longer periods of time such as 15 min., the anodic peak of E t A Q H 2 electrooxidation cannot be detected due to the fast chemical oxidation of this species (Fig. 6.9 b). Furthermore, on the cathodic scan two waves can be distinguished, first a sigmoidal response with Em at -0.5 V followed by a peak at -1.27 V . The latter belongs to the 2e" reduction of O2 (see Section 6.2.2) whilst the sigmoidal cathodic current is the so-called 'catalytic' [Bard and Faulkner, 1981] (or 'redox regenerated') wave of E t A Q reduction. Therefore, the cyclic voltammograms presented in Fig. 6.9 are described by the combination of the homogeneous redox regeneration of E t A Q and the 2e" O2 reduction mechanisms, as shown by Fig. 2.1. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 133 Figure 6.9 First scan cyclic voltammogram for the E t A Q - 0 2 system when the organic phase is in contact with 0.1 M H 2 S 0 4 . a) 1: N 2 purge; 2: 0 2 purge for 15 sec, b) 0 2 purge for 15 min. Scan rate: 100 m V s"1. Organic phase: 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336 in T B P . G C electrode.295 K . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of02 134 In the case of 0.1 M N a 2 C 0 3 in contact with the organic phase, the cyclic voltammograms of the E t A Q - 0 2 system reflect similar phenomena to the acid media case (Fig. saturated organic catholyte. The height of the first reduction peak given by the E t A Q / EtAQ"* process, increased almost 4 times when 0 2 was purged for 15 seconds. Again, the anodic peaks are diminished by the chemical reaction between 0 2 and the reduced E t A Q species and after 15 min. of 0 2 purge the anodic response could not be detected (Fig. 6.10 b). Furthermore, on the cathodic response only a sigmoidal wave is obtained. The sole reduction wave, with a plateau around -1.3 V , incorporates both the E t A Q and 0 2 reduction processes as well (see Fig. 6.8 c). Although qualitatively it can be inferred from both Figs. 6.9 and 6.10 that the rate of reaction between 0 2 and the reduction products of E t A Q is 'fast' since the anodic peaks are absent in 0 2 enriched solutions, a quantitative assessment of the chemical rate constant kc, would be desirable. The homogeneous rate constant kc, can be evaluated from cyclic voltammetry measurements. From the cyclic voltammetry theory of 'redox regeneration' [Brown and Sandifer, 1986], it is known that the ratio of the limiting 'redox regeneration' current IL.r (i.e. corresponding to E t A Q reduction in the presence of 0 2 ) , to the diffusion current in the absence of regeneration IP4 (i.e. E t A Q reduction in N 2 saturated electrolyte), is related to kc according to eqn. [6.18]: 6.10). Fig. 6.10 a, shows the voltammogram when 0 2 was purged for 15 sec. in a previously N 2 [6.18] where kc (s"1) pseudo first-order reaction rate constant for the reaction of E t A Q H 2 with 0 2 , v scan rate ( V s"1), n number of electrons transferred per E t A Q molecule, whereas the rest of the variables in eqn. [6.18] have their usual meaning. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of02 135 0 -0.7 -1.4 E / V vs. Ag/AgCI Figure 6.10 First scan cyclic voltammogram of the E t A Q - 0 2 system when the organic phase is in contact with 0.1 M N a 2 C 0 3 . a) 1: N 2 purge; 2: 0 2 purge for 15 sec, b) 0 2 purge for 15 min. Scan rate: 100 m V s"1. Organic phase: 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336, in T B P . G C electrode. 295 K . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 136 To employ eqn. [6.18], a set of Ii,r and Ipj values were obtained by measuring the E t A Q reduction currents as a function of scan rate for both 0 2 and N2 saturated electrolytes. It must be emphasized that for the E t A Q - O2 system care must be taken to isolate the E t A Q response from that of O2 in evaluating the reduction currents. Therefore, at a given scan rate, the O2 reduction current was subtracted from the overall, E t A Q - O2, response. Appendix J shows the employed experimental data. A least-square fitting of the current ratios to the square root of scan rate yields the pseudo first-order rate constant kc (Table 6.3). Table 6.3 Pseudo first-order rate constant kc, for the homogeneous oxidation of the reduced E t A Q species (e.g. E t A Q H 2 ) . T : 295 K . Nr. Aqueous electrolyte in contact kc Pv2 with the organic phase (s-1) (for eqn. [6.18]) 1 0.1 M N a 2 C 0 3 16.5 ±3.4 0.96 2 0.1 M H 2 S 0 4 0.11 ±0.017 0.99 Table 6.3 shows that the intrinsic kinetic rate of E t A Q regeneration by homogeneous oxidation, is ca two-orders of magnitude faster when the aqueous part of the emulsion is 0.1 M N a 2 C 0 3 . Furthermore, it is important to compare the intrinsic kinetic rates of the homogeneous redox reaction rc, and that of E t A Q electroreduction rei respectively. Thus, assuming first-order dependence on concentration for both the homogeneous and the electroreduction intrinsic rates one obtains: 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of02 137 kC EtAQH 2 aks exp anF(E - E0EIAQ) [6.19] RT C EtAQ where a specific surface area of the electrode (m 2 m"3), ks standard heterogeneous rate constant (Table 6.2) (m s"1), a charge transfer coefficient (Table 6.2), n number of electrons exchanged in the electrochemical step per E t A Q molecule, kc homogeneous rate constant (Table 6.3) (s"1), E° estimated standard potential for E t A Q electroreduction under pertinent conditions (i.e. alkali (2-nd step) -1 V A g /Agci ; in acid -0.11 V A g / A g c i CEIAQ, Figs. 6.7 a and b, respectively) and CEIAQHI 2-ethyl anthraquinone and 2-ethyl anthrahydroquinone concentration, respectively (mol m"3). Considering that the concentrations of the quinone and hydroquinone forms are equal and assuming that a is around 10 4 m 2 m"3 (e.g. for the graphite felt or R V C 100 ppi, Section 4.3.1), with the ks and kc values from Tables 6.2 and 6.3 respectively, eqn. [6.19] implies that the rate of intrinsic chemical oxidation step of the mediating cycle is higher or equal to the rate of intrinsic E t A Q electroreduction up to overpotentials around -0.6 V in alkali and around -0.2 V in acid, respectively. Thus, at overpotentials exceeding the above values the overall intrinsic rate of mediated peroxide electrosynthesis is controlled by the chemical oxidation step (Fig. 2.1). However, for the 'real-case' scenario the various mass transfer effects of the multi-phase electrolyte system have to be taken into account for a judicious analysis (see further). 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 138 6.2.4 Determination of the reaction zone for 2-ethyl anthraquinone reduction in emulsion: Spectrophotometric experiments A n important issue is whether electroreduction of the E t A Q mediator takes place at the two-phase (organic / electrode) boundary or is confined at the three-phase (organic / aqueous / electrode) interface. To study this aspect, the following experiments were performed. The organic and aqueous phases, in a 1 to 3 volume ratio, were mixed for 10 min. under N 2 purge in an 'H ' -ce l l . The phases were left to separate, a N 2 'blanket' was maintained above the catholyte and a graphite rod electrode was inserted in the cathode compartment. The lower part of the graphite rod was in contact with the aqueous phase while the upper part contacted the organic phase. As soon as current was applied to the graphite cathode, the colour changes occurring at the interfaces and in the bulk phases were visually followed* (Fig. 6.11). After ca 10 minutes of electrolysis, a sample (4 ml) was withdrawn from the organic layer and the absorption spectra were recorded (Appendix H). For both the acid (0.1 M H 2 S 0 4 ) and alkali (0.1 M N a 2 C 0 3 ) emulsions, the organic phase was composed of T B P (solvent), 0.2 M T B A P , 0.06 M A336 and 0.1 M E t A Q . For the acid aqueous phase situation (Fig. 6.11 a), when current was applied (104 A m"2), the S /L /L interface became dark-red in less than ca 20 s. Eventually, the red coloration spread upward on the graphite surface, along the electrode / organic phase boundary. After ca 1 min. of electrolysis the graphite surface in contact with the organic phase became entirely red. The red coloration persisted only on the electrode surface, while the bulk of the organic became gradually dark yellow-orange, characterized by an absorbance peak at 468 nm (Fig. H-2 a, Appendix H). Based on literature studies of the spectroelectrochemistry of anthraquinone compounds in organic media (see Table H - l , Appendix H), the dark red-purple coloration is attributed to the radical-anion E tAQ - " , while the absorbance peak around 468 nm is due to the completely reduced species, the dianion E t A Q 2 ' (or its protonated form E t A Q H 2 ) , conferring the dark yellow-orange colour of the bulk. The fact that the red EtAQ"* layer was only observed on * The reduced EtAQ species are easily identifiable based on their characteristic colours (see Table H - l , Appendix H). 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 139 the cathode surface starting with the 'proton-rich' L / L interface, while away from the cathode (i.e. in the bulk) the dark red-purple species turned into the orange E t A Q H 2 , supports the le" reduction - disproportionation mechanism described by eqns. [6.7] - [6.9]. Furthermore, based on Fig. 6.11 a, the main reaction zone for the acid emulsion is situated at the three-phase, S /L /L interface. Since the graphite rod was inserted into the cell from top to bottom, a thin layer of organic adhered to the lower part, which contacted the aqueous solution. When current was applied a green film quickly covered the lower part of the graphite rod in contact with 0.1 M H2SO4 (Fig. 6.11 a). The green coloration could be also attributed to the fully reduced and protonated species E t A Q H 2 (e.g., according to Babaei, et al, [1997], A Q H 2 gives a green-yellow coloration corresponding to an absorption maxima of 383 nm). Turning to the alkali emulsion (Fig. 6.11 b), after current was applied the entire length of the graphite in contact with the organic was covered in ca 10 s by the dark red-purple, layer characteristic of E t A Q " . Furthermore, unlike the acid emulsion case, the purple coloration was not confined to the electrode surface, but it was present in the bulk organic phase as well, as shown by the characteristic absorbance peak at 521 nm for samples taken from the bulk (Fig. G-3 a, and Table Ff-1, Appendix H). This indicates that the radical-anion EtAQ" ' (very likely involved in ion-pairs with N a + ) is more stable when the organic phase is in contact with an alkaline aqueous phase, as discussed in Sections 6.2.1 a and b. The greater stability of E t A Q " (i.e., lesser tendency for disproportionation) enables the second eletroreduction step generating E t A Q 2 " as described by eqns. [6.5] - [6.6]. The absorption spectra of the organic phase after 2 min. of electrolysis (Fig. H-3 a, Appendix H), shows also a peak at 450 nm, which is characteristic to the orange EtAQ 2 " . Together the dark red-purple and orange species, give the observed brownish colour of the organic catholyte (Fig. 6.11 b). Based on Fig. 6.11 b, it was concluded that the entire electrode - organic phase boundary (including the three-phase S / L / L interface) constitutes the reaction zone for electroreduction of E t A Q in alkali emulsion. A n orange film covered quickly the lower part of the graphite rod in contact with 0.1 M Na 2 C03 . A s discussed above, the orange coloration is characteristic for E tAQ 2 " . 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 0, 140 The spectrophotometric experiments revealed that the main reaction zone for E t A Q reduction is different for the two emulsions under study. For the acid emulsion the reaction zone is situated at the three-phase S /L /L interface, whilst in the case of the alkali emulsion the entire electrode surface - organic phase boundary constitutes the active zone for E t A Q electroreduction. Chapters 6.1 and 6.2 were concerned with defining the composition of the emulsion system, studying the type of emulsions formed, and with investigating certain fundamental electrochemical aspects of the E t A Q - 0 2 system. The remaining parts of the present work deal with some of the practical, engineering and economic challenges of the multi-phase E t A Q mediated peroxide electrosynthesis. 6.2 Fundamental Studies of the Emulsion Mediated Electroreduction of 02 141 a) aqueous phase: 0.1 M H2SO4 b) aqueous phase: 0.1 M Na2CQ 3 organic EtAQ organic ©-aqueous aqueous No current i: 104 Am -2 Figure 6.11 Schematic representation of the colour changes observed during the electroreduction of 0.1 M E t A Q in T B P . Organic phase in contact with: a) 0.1 M H 2 S 0 4 ; b) 0.1 M N a 2 C 0 3 . Area of graphite rod: 7.3X10 - 4 m 2 . Time: 10 s to 5 min. 295 K . 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 142 6.3 Emulsion Mediated Electrosynthesis of H 2 0 2 in a Batch Cell The batch, 'H ' -ce l l (Fig. 4.2) is convenient for testing the feasibility of electrode materials, various electrolyte compositions and generally, to establish a set of operating conditions that might be suitable for larger scale experimentation. In the batch runs the cathode was reticulated vitreous carbon of different specific surface areas. Currents between 0.3 and 0.8 A were employed, corresponding to superficial cathode current densities from 300 to 800 A m"2. 6.3.1 Influence of the specific surface area of R V C Previous research of the E t A Q reduction in emulsion, showed that the specific surface area of the R V C cathode influences the figures of merit for peroxide electrosynthesis [Huissoud and Tissot, 1999 II]. It was found that the higher the number of pores per inch (ppi), i.e., the higher the specific surface area, the less efficient the mediated electrosynthesis of H 2 0 2 . This conclusion was rationalized based on the observation that for R V C with ppi > 45, the organic phase is almost completely held-up inside the porous cathode, causing the cathode potential to reach values at which H 2 0 2 is reduced. Therefore, a 20 ppi R V C was recommended over R V C types with higher specific surface area [Huissoud and Tissot, 1999 II]. To test the validity of the above finding for the emulsion mediated systems explored in the present work, experiments were performed with cathodes over a wide range of R V C surface area (see Table 4.2), i.e., between 530 and 6560 m 2 m"3, corresponding to 10 - 100 ppi. The peroxide concentration and the cathode potential at a constant superficial current density of 300 A m"2 were followed, over a 10 hour period, for each R V C cathode. The catholyte was a well-mixed emulsion of 0.1 M either N a 2 C 0 3 or H 2 S 0 4 aqueous electrolyte and T B P with 0.1 M E t A Q , 0.2 M T B A P and 0.06 M A336 at 300 K . The organic to aqueous phase volume ratio was 1 to 3. Oxygen was continuously purged into the catholyte, at atmospheric pressure. Figure 6.12, shows that, with both aqueous phases after 10 firs ca 0.5 M peroxide per emulsion was accumulated in the 'H'-cell, when the cathode was a 30 ppi R V C (i.e. specific surface area 1800 m 2 m"3). On the other hand, for ppi values between 60 and 100, the H 2 0 2 concentration was only between 0.2 and 0.3 M . 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 143 Significant, however not complete, retention of the organic phase inside the porous matrix was observed only for the 100 ppi R V C . Furthermore, when the number of ppi was smaller than 30, the performance suffered, yielding peroxide concentrations of only 0.23 M (in 0.1 M N a 2 C 0 3 ) and 0.4 M (0.1 M H 2 S 0 4 ) (Fig. 6.12 a). Thus, the results presented in Fig. 6.12 reveal a somewhat more complicated effect of the R V C specific surface area than the one reported in the literature. The cathode potential profiles (Fig. 6.12 b) can help explaining the observed results. Figure 6.12 b shows that the smaller the specific surface area of R V C the more negative the cathode potential for the same superficial current density. In other words, when the specific surface area is small the local current density is high, which drives the cathode potential to more negative values. For instance, in the case of 0.1 M N a 2 C 0 3 aqueous phase, by switching from 30 to 10 ppi for the same superficial current density of 300 A m"2, the cathode potential drops from -1.9 to -2.6 VAg/Agci,std.Kci (Fig. 6.12 b). A t the latter potential the secondary electrode reactions such as peroxide electroreduction and H 2 evolution are significant, which is reflected in lower peroxide concentration (Fig. 6.12 a). For high specific surface area on the other hand, (e.g. 6560 m 2 m"3, 100 ppi) with 0.1 M N a 2 C 0 3 the cathode potential is only -0.9 V (Fig. 6.12 b). However, the corresponding cyclic voltammetry wave for the E t A Q - 0 2 system has its plateau at -1.3 V (Fig. 6.10 b). Clearly, at 300 A m" the overpotential is not large enough on the 100 ppi R V C for effective electroreduction of E t A Q . The same is true for the acid aqueous phase, where based on cyclic voltammetry (Fig. 6. 9 b) potentials more negative then -0.9 V would be required to reduce E t A Q . In the batch cell, at 300 A m"2 the 30 ppi R V C (superficial area 1800 m 2 m"3) offered an 'optimum' cathode potential distribution, i.e., around -1.9 V in alkali and -1.35 V in acid, respectively (Figs. 6.12 a and b). Since the potential and current distribution in porous electrodes is determined by an intricate interplay of electrode kinetics, hydrodynamics and electric conductivity (both electronic and ionic), the effect of cathode specific surface area may change for different hydrodynamic conditions (e.g. in a continuous flow-by cell). 6.3 Emulsion Mediated Electrosynthesis ofH202 in a Batch Cell 144 c o TO C CD O C o o <D "D X o 1_ CD Q_ O <: < </>' > > LU 0.6 0.5 0.4 0.3 0.2 0.1 -0.5 1 i 1 i — r CE: 95% -1.5 -2.0 -2.5 -3.0 • 0.1 M Na2C03 O 0.1 M H 2 S O 4 ~ l I ' 1 ' 1 ' 1 1 1 1 1 r ~ 0 1 2 3 4 5 6 7 . 1 1 : b) 1 l 1 1 1 1 1 1 1 1 . - O -. O — • -- • -~: • -\ 1 1 0 1 1 l 2 1 1 1 1 3 4 1 1 1 1 1 5 6 / 7 Specific Surface Area / m 2 I I I m- 3 (x 103) I 10 30 60 100 R V C ppi Figure 6.12 Batch electrosynthesis: the effect of R V C specific surface area. 300 A m"2. 300 K . Organic: 0.1 M E t A Q , 0.2 M T B A P , 0.06 M A336. Org./Aqu. = 1 / 3 v / v . 0 2 0.1 M P a . Electrolysis time: 10 firs. 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 145 6.3.2 Emulsion mediated hydrogen peroxide electrosynthesis in the batch ' H * - cell To demonstrate the production of greater than 0.3 M peroxide in a variety of acid and alkaline electrolytes at superficial current densities > 500 A m"2, 10 hour batch electrosynthesis experiments were performed in an 'H'-cell with a 30 ppi R V C cathode. The performance of the mediated electrosynthesis was compared with direct 0 2 electroreduction and the effect of A3 3 6 was investigated. Typically, the catholyte was a 1 to 3 volume ratio (organic to aqueous) emulsion. Oxygen was purged at atmospheric pressure over the entire 10 hour run. The organic phase contained 0.1 M E t A Q , 0.2 M T B A P and various concentrations of A336, dissolved in T B P . The temperature was 303 ±5 K . a) Aqueous phase: 0.1 M H2SO4; pHin 1.1. Comparing the performance of the mediated emulsion system with that of direct O2 electroreduction in the presence of 10"3 M A336, Fig. 6.13 shows that at 500 A m"2 the H 2 0 2 concentration was almost 6 times higher for the E t A Q mediated electroreduction. The latter system generated 0.46 M H2O2 (per emulsion) with 58% current efficiency, while the 0 2 -surfactant system yielded only 0.085 M H2O2 at 25% current efficiency. These results show the importance of E t A Q mediation especially in the case of high superficial current densities, i.e. > 500 A m"2. Generally, at the employed superficial current density, the cathode potential for the O2 / A336 system was too negative for efficient peroxide synthesis, i.e. around -1.9 to -2.6 VAg/Agci- A t those potentials the secondary electrode reactions gain significance, giving low current efficiencies. In the case of mediated electroreduction on the other hand, the cathode potential was between -1.3 and -1.7 V , allowing for efficient electroreduction of E t A Q , as discussed in Section 6.3.1. 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 146 0 2 4 6 8 10 Time / hr. Figure 6.13 Comparison of L / L / G , E t A Q mediated and 0 2 / A 3 3 6 systems. Batch electrolysis. Aqueous phase 0.1 M H 2 S 0 4 . Organic phase: T B P , 0.2 M T B A P , 0.1 M E t A Q . A336 cone, per catholyte 10"3 M (for both cases). Organic / Aqueous: 1 / 3 v / v . 30 ppi R V C . 500 A m"2. p H i n , a q u 1.1, p H f , a q u 1.2 - 1.35. 0 2 0.1 M P a . 308 K . The effect of A3 3 6 concentration on the emulsion mediated electrosynthesis at 500 A m"2, was also investigated. Fig. 6.14 shows that increasing the A336 concentration increased the H 2 0 2 concentration and the current efficiency over the 10 hour period. For instance, without A336 after 10 firs, 0.36 M H 2 0 2 (46% current efficiency) was obtained while with 0.015 M A336 the peroxide concentration reached 0.53 M with 68% efficiency. The effect of A336 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 147 became noticeable after the first 2 or 3 hrs, when about 0.2 M H 2 0 2 was accumulated. The overall effect of the cationic surfactant is most likely due to a combination of several factors: such as its influence on both 0 2 and H 2 0 2 electroreduction (Sections 5 .1 - 5.3), H 2 evolution and improved wetting of the cathode by the organic (see further). Figure 6.14 Influence of A336 concentration on the emulsion mediated electrosynthesis of H 2 0 2 at 500 A m"2 for an aqueous phase of 0.1 M H 2 S 0 4 . Rest idem Fig. 6.13. 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 148 b) Aqueous phase: 2 MNa2S04; pHin 3.1. The aqueous electrolyte was 2 M Na 2S04 acidified with glacial acetic acid to p H 3.1.This electrolyte p H is relevant for certain acidic peroxide bleaching methods such as the 'GreenOx' process (see Chapter 1: Introduction). The effects of superficial current density and E t A Q mediator concentration were explored in the range of 500 to 800 A m"2 and 0.1 to 0.2 M E t A Q concentration, respectively. Fig. 6.15 shows that increasing the current density from 500 to 800 A m"2 while maintaining a 0.1 M mediator concentration has no benefits, since after 10 hrs. the peroxide concentrations are virtually the same, around 0.5 M , but the current efficiency for the 800 A m"2 case is much worse, i.e. 39% as opposed to 60% for 500 A m"2. A t high current density and low mediator concentration, the secondary reactions are significant and prevent further accumulation of peroxide in the batch cell. However, when the E t A Q concentration is increased to 0.2 M in conjunction with 800 A m"2 current density, the H2O2 concentration increased to 0.61 M after 10 hrs. (current efficiency 46 %). Note that 0.2 M E t A Q is near the solubility limit in T B P at room temperature. A s expected, the p H of the catholyte increases during the 10 hour runs, for 800 A m"2 the final p H was typically between 5.7 and 6.2, whereas for 500 A m"2 it was around 5. In acidic peroxide bleaching the desired p H is about 5 [Paren and Tsujino, 1998], Hence, the p H of the peroxide solution produced by electrosynthesis matches fairly well the p H required for acidic bleaching. 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 149 0 2 4 6 8 10 Time / hr. Figure 6.15 Influence of E t A Q concentration and superficial current density on H 2 0 2 electrosynthesis in 2 M N a 2 S 0 4 , pHi„, a q u 3.1. Batch electrolysis. Organic phase composition idem Fig. 6.13. A336 concentration 1.5xl0"2 M (per emulsion). Organic / Aqueous: 1 / 3 v / v . 30 ppi R V C . 0 2 0.1 M P a . 310 K . 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 150 c) Aqueous phase: 0.1 MNa2C03,' pHin 11.3. The last set of peroxide electrosynthesis experiments in the batch 'H ' -ce l l , were performed in alkaline conditions, using 0.1 M Na2C03 as aqueous electrolyte. Fig. 6.16 shows the peroxide concentration profile over time at 500 A m"2 superficial current density for both the emulsion mediated system and O2 reduction in the presence of A336. The average temperature was 310 K . After 10 hrs, with 0.1 M E t A Q at 500 A m"2, 0.72 M peroxide was obtained with a current efficiency of 89%. This is the highest peroxide concentration obtained in the batch cell for any of the three aqueous phases investigated. The p H of the aqueous catholyte (i.e. unbuffered 0.1 M Na2COa), increased after 10 hrs. from 11.3 to 13.9. A t p H around 14 the rate of the 2e" reduction of O2 is fast, which contributes to the high peroxide concentration. Similar to the H2SO4 case (Fig. 6.13), at 500 A m"2 the batch, non-mediated reduction of 0 2 at atmospheric pressure is inefficient on R V C , yielding only a maximum of 38 m M peroxide, with a current efficiency of 22% (Fig. 6.16). The main reason for the difference in performance between the non-mediated and mediated electrosynthesis is the different electrode polarization behaviour for the two cases. A t 500 A m"2 the cathode potential for the O2 / A336 system was around -3 VAg/Agci, whereas under the same conditions for the emulsion mediated system the potential was only ca -1.9 V . A t the latter potential E t A Q is efficiently reduced (see Sections 6.2.3 and 6.3.1), therefore enabling an effective mediation cycle. 6.3 Emulsion Mediated Electrosynthesis of H202 in a Batch Cell 151 0 2 4 6 8 10 Time / hr. Figure 6.16 Comparison of the emulsion mediated and ( V A 3 3 6 systems. Batch electrolysis. Aqueous phase: 0.1 M N a 2 C 0 3 . Organic phase: Organic phase: T B P , 0.2 M T B A P , 0.1 M E t A Q . A336 cone, per catholyte 8xl0" 4 M (for both cases). Organic / Aqueous: 1 / 3V/V. 30 ppi R V C , 500 A m " 2 . O 2 0 . 1 M P a . 310 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 152 6.4 Emulsion Mediated Electrosynthesis of H 2 0 2 in 'Flow-by' Cells The promising peroxide concentrations (e.g. > 0.5 M ) and current efficiencies (e.g. > 50%) obtained with the mediated electrosynthesis in the batch cell warranted investigation of the present system in a continuous 'flow-by' cell, under conditions which are closer to potential industrial applications. The objective of these experiments was to study the interplay between the electrochemistry, three-phase flow hydrodynamics (L /L /G) and the characteristics of the porous cathode (e.g. specific surface area) with regard to peroxide synthesis. Parametric and factorial experiments were performed with both acid (pH 3.1) and alkali (pH 9.3 - 9.6) aqueous electrolytes. The organic phase was composed of T B P , various concentrations of mediator (EtAQ), supporting electrolyte ( T B A P ) and cationic surfactant (A3 3 6). Emulsification was achieved by continuously mixing the two liquid phases in the feed tank. The three-dimensional electrode was operated with co-current, upward, flow of the L / L emulsion and 0 2 gas. A detailed description of the experimental apparatus and methods used, together with the corresponding flowsheet, is presented in Section 4.2.2. 6.4.1 Aqueous phase: 1 M Na2SC>4; pHjn 3.1. a) Cathode Selection In the batch experiments R V C was the preferred cathode material due to convenient handling and good performance. However, in the case of the flow cell, the issue of the porous cathode has to be revisited due to different mass transfer and operational constraints. Three cathodes were tested, i.e. R V C 30 and 100 ppi, and graphite felt (Section 4.3.2). The latter is the most convenient from the point of view of handling and assembling industrial scale cells (e.g. 1 m 2 superficial area). Furthermore, there are important electrocatalytic and structural differences between the three cathode materials that might affect the figures of merit for peroxide electrosynthesis (Table 4.1 and 4.2). 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 153 The peroxide concentration per pass and current efficiency for R V C 30 and 100 ppi and graphite felt were compared, using a catholyte emulsion composed of a 1 to 2 volume ratio of organic to aqueous phases. The flow rates were: 4.5xl0" 7 m 3 s"1 for the L / L emulsion and 1.98xl0~6 m 3 s"1 at STP for O2. These values correspond to a gas load of 0.026 kg m"2 s"1 and liquid load of 4.25 kg m"2 s"1, generating a bubble flow (liquid continuous) regime [Storck et al, 1986]. The outlet reactor pressure was 0.1 M P a and the temperature 303 K . The organic phase (TBP) contained 0.2 M E t A Q , 0.1 M T B A P and 1 m M A336. The three cathode materials were tested in the 'short' flow cell (i.e. 0.14 m length, superficial area 32.9xl0" 4 m 2 , Section 4.2.2), using superficial current densities between 300 and 3000 A m" . Figures 6.17 a and b, show that the best results were obtained with compressed graphite felt, followed by the 100 ppi R V C and lastly the 30 ppi R V C . Note that the specific surface area decreases from ca 16,000 m 2 m"3 for graphite felt to 1800 m 2 m"3 for R V C _ 3 0 (Tables 4.1 and 4.2). Therefore, it can be inferred that the overall mass transfer capacity for the three electrodes under study decreases in the same order, i.e. GF, R V C _ 1 0 0 and R V C _ 3 0 . Unfortunately, due to lack of mass transfer correlations for co-current upward three-phase flow through graphite fiber bed and/or R V C electrodes, an accurate calculation of the mass transfer limited superficial current density for E t A Q reduction, is impossible. Extrapolation of existing mass-transfer correlations for multi-phase flow over mm or cm size packing, to fiber beds with characteristic diameters the order of /mi, is unwarranted. However, an estimation of the liquid to solid mass transfer capacity can be attempted based on single-phase flow literature studies using graphite felt and R V C electrodes. Appendix K presents the estimation of the organic-liquid to solid mass transfer coefficient and capacity for the three electrodes under study (i.e., GF , R V C _ 1 0 0 and R V C _ 3 0 ) . Table 6.4 summarizes the mass transfer parameters together with the E t A Q superficial limiting current densities and the corresponding electroactive bed thickness (see eqns. [3.20] - [3.23]). 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 154 Table 6.4 Estimated mass transfer coefficient, mass transfer capacity, electroactive bed thickness and superficial limiting current density for E t A Q reduction in acid emulsion using a 'flow-by' cell. Emulsion load: 4.25 kg m"2 s"1. 0 2 load: 0.026 kg m"2 s'1 (at STP). 295 K . Porous Cathode Mass Transfer Coefficient x l O ' ^ m s " 1 ) Mass Transfer Capacity Electroactive Bed Thickness (mm) Limiting Superficial Current Density (A m"2) G F 2.4 0.13 0.5 933 R V C _ 1 0 0 0.7 0.015 1.4 275 R V C _ 3 0 0.3 0.002 4.4 111 Fig. 6.17 a, supports the mass transfer estimations presented in Table 6.4. Note that the experimental data from Fig. 6.17 includes also a contribution from direct 0 2 reduction (see further). Due to its higher mass transfer capacity, G F can be operated at 1000 A m"2 with a current efficiency of 75% whilst under the same conditions, the R V C electrodes are inefficient, with current efficiencies of ca 45% for R V C _ 1 0 0 and only 30% for R V C _ 3 0 (Fig. 6.17 b). Table 6.4 shows that for G F the maximum electroactive bed thickness capable of working at the mass transfer limited current density without significant secondary reactions, is only a small fraction of the nominal thickness (i.e. 0.5 mm vs. 4.5 mm). 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 155 c o (n E CD C d c o o CN o CN X > o c CD O it CD -I—' C CD i i_ o 40 30 20 10 0 0 — i — a) • Graphite Felt • RVC_100 • RVC 30 T T T 1000 2000 3000 4000 1000 2000 3000 4000 Current density / A rrr2 Figure 6.17 H2O2 concentration in emulsion and current efficiency for one-pass, E t A Q mediated, electrosynthesis in flow-by cells. Aqueous phase: 1 M Na2SC«4, p H ^ 3.1. Organic phase: T B P with 0.2 M E t A Q , 0.1 M T B A P and 1 m M A336. (Organic / Aqueous)v/v : 1 / 2.1. Emulsion load: 4.25 kg m"2 s"1. 0 2 load (at STP): 0.026 kg m' 2 s"1. 0.1 M P a . Superficial cathode area: 32.9xl0" 4 m 2 . T: 308 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 156 b) The Role of EtAQ Mediator To establish the individual contributions of the reduction of the mediator and the direct reduction of 0 2 to the overall peroxide synthesis in the multi-phase catholyte, two experiments were performed using a graphite felt cathode under the same conditions, with and without E t A Q present (Fig. 6.18). Fig. 6.18 shows that the two experiments (i.e., with and without E t A Q ) at 300 A m"2 generated similar peroxide concentrations. However, for superficial current densities over 300 A m' , the figures of merit for the E t A Q mediated system exceed those of direct O2 reduction. Thus, even the higher O2 solubility in the L / L emulsion and the enhanced overall O2 mass transfer rate*, are insufficient to efficiently support practical superficial current densities (e.g. > 500 A m"2) for direct O2 reduction in acid emulsion at 0.1 M P a . Figure 6.18 Comparison between E t A Q mediation and multiphase 0 2 reduction in a 'flow-by' cell equipped with a graphite felt cathode. Conditions idem Fig. 6.17. * In the presence of a second (i.e. organic) phase, the overall 0 2 mass transfer rate is expected to be higher than in the case of pure aqueous electrolyte, due to the so-called 'extraction mechanism' [Fenton and Alkire, 1988]. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 157 c) Factorial Experimental Design Using graphite felt cathode* a complete factorial experiment with four variables at two levels plus a centerpoint (i.e. 2 4 + 1) was performed, to better understand the characteristics of the acid emulsion mediated peroxide electrosynthesis. Table 6.5 Variables and their values in the factorial study of the emulsion mediated electrosynthesis of H2O2 using 1 M Na2SC<4 (pHi n 3 .1 ) as aqueous electrolyte. Variable / symbol (units) L O W level (-) •CENTER ' (° ) 'H IGH' level (+) 1. Superficial Current Density (I) (Am" 2 ) 1000 2000 3000 2. L / L Emulsion L o a d ( L ) (kgm- 2 s"1) 2.8 6.9 5.4 3. Organic / Aqueous Volume Ratio (R) 0.33 0.47 1.11 4. A336 concentration (S) (mM) p e r emulsion 0 0.5 1 * Superficial area: 32.9xl0"4m2, referred to as the 'short' electrode (Fig. 4.4). 6.4 Emulsion Mediated Electrosynthesis of H202 in 'Flow-by' Cells 158 In the factorial design the 0 2 load, outlet reactor pressure and the temperature were constant: at 1.45xl0' 2 kg m ' 2 s'1, 0.1 M P a and 313 K , respectively. Based on the gas and liquid loads, all the factorial runs were performed in the liquid continuous (bubble) flow regime. The organic phase contained 0.2 M E t A Q (mediator) and 0.1 M T B A P (supporting electrolyte). Table 6.6 shows the H2O2 concentration in emulsion per pass and the corresponding current efficiency obtained for all the 17 points of the factorial experiment according to the design created with the Jass 2.1® software [Jass, 1987]. A number of points (including the centerpoint) were repeated two or three times, and the standard deviations were calculated (e.g. see entry nr. 14, Table 6.6). The outlet p H for the aqueous electrolyte varied from 3.8 to 6. The latter value was obtained for the runs at 'high' current density and 'low' emulsion load. The change of p H in the above range is not expected to have a significant influence on the figures of merit. The factorial runs (Table 6.6) gave a wide range of current efficiencies, between 100 %, for 'low' level of current density and 'high' levels for the rest of the variables (entry nr. 12) and 15% for the opposite scenario, i.e. 'high' level current density whereas the rest of the variables are at 'low' level (entry nr. 5). From the experimental data of Table 6.6, the main and interaction effects of the variables are shown in Table 6.7. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 159 Table 6.6 Factorial experiment: Emulsion H 2 0 2 concentration and current efficiency per one-pass, for the mediated electrosynthesis on graphite felt, using 1 M N a 2 S 0 4 ( p H i n 3.1) as aqueous phase. T: 313 K . 0.1 M P a . For the meaning of symbols, see Table 6.5. N r . I L R s r '-'peroxide ( m M ) (%) 1 - + - + 12 87 2 - + + - 11 ±0.5 80 ±3 3 - + - - 10 72 4 + + - - 23 55 5 + - - - 26 15 6 - - + - 34 60 7 + + - + 32 77 8 + + + - 27 65 9 - - + + 40 70 10 + - - + 34 19 11 - - - + 38 63 12 - + + + 14 100 13 + + + + 35 84 14 0 0 0 0 31 ±0.9 66 ±1.8 15 + - + + 30 18 16 + - + 0 29 16 17 - - - - 35 62 * Average factorial response: 27.1 mM. Pooled standard deviation: ±0.9 mM. ** Average factorial response: 59.5%. Pooled standard deviation: ±2.3%. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 160 Table 6.7 Main and interaction effects for the emulsion mediated system with 1 M N a 2 S 0 4 (pHi„3.1) as aqueous electrolyte. Graphite Felt Cathode. 0.1 M P a . 313 K . P E R O X I D E C U R R E N T E F F E C T S concentration* EFFICIENCY** (mM) (%) M A I N I 5.3 -30.6 L -12.8 37 R 1.0 4.6 S 5.0 11.6 Two-factor I N T E R A C T I O N IL 12.3 16.1 FR 0.3 -1.1 IS 1.5 0.1 L R 1.3 4.2 LS 0.5 7.4 RS -0.5 1.1 Three-factor I N T E R A C T I O N 1LR 0.8 0.1 ILS 1.5 1.4 IRS -1.5 -2.4 LRS 0.5 -0.6 * standard error of effect: ±1.9 mM ** standard error of effect: ±4.8 % 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 161 Table 6.7 shows that the emulsion load (L) exerted the strongest main effect followed by the superficial current density (I) and the cationic surfactant concentration (S). Interestingly, the main effect of the organic/aqueous phase volume ratio (R) was negligible*. This result can be explained based on the spectrophotometric study (Section 6.2.1), which revealed that the reaction zone for E t A Q reduction in the acid emulsion starts at the 'proton rich' three-phase interface of the electrode with both the organic and aqueous phases (i.e. S /L /L boundary). Therefore, especially for the short liquid residence times encountered in the flow cell (i.e. 12 to 50 s), the electroreduction of E t A Q from the bulk organic phase contributes little to the overall electrosynthesis. In fact a 'high' organic to aqueous phase volume ratio is disadvantageous from the point of view of energy consumption, i.e. it gives higher cell voltages due to the decreased conductivity. For instance at 1000 A m' 2 for R = 0.33 the cell voltage was typically ca 6.5 V , whereas with a ratio R of 1.11, the cell voltage increased to 7.3 V . The main effects of the emulsion load (L) on both the H2O2 concentration and current efficiency, i.e. -12.8 m M and 37%, respectively, are a reflection of the different residence times. At ' low' emulsion load the residence time is high, ca 50 s, which allows for higher H2O2 concentrations per pass. The current efficiency, however, increased with 37% vs. the average at a 'high' load, because of the limited contribution from side reactions due to the low H2O2 concentration. A n additional aspect to the influence of the emulsion load, is expressed by its high interaction effect with the superficial current density (EL) (Table 6.7). EL for current efficiency is 16% whereas, for H2O2 concentration is 12 m M (Table 6.7). This finding is an indication for mass transfer control at 3000 A m - 2 . A t 'low' emulsion load, the limiting superficial current density calculated with the procedure described in the previous section, is ca 800 A m' 2 , hence, increasing the superficial current density from 1000 A m"2 to 3000 A m - 2 wi l l decrease the current efficiency, below 20%, as shown by entries nr. 5, 10, 15 and 16 in Table 6.6. For 'high' flow rate (or load), the E t A Q mass transfer is enhanced, the corresponding limiting current density almost doubles to 1400 A m"2, and at 3000 A m"2 the current efficiencies increase to 55 -84%. The magnitude of the effect is within the standard error for both H 20 2 cone, and current efficiency (Table 6.7). 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 162 The cationic surfactant A3 3 6 had statistically significant main effects on both H2O2 concentration (i.e. 5 m M ) and current efficiency (i.e. 11.6%). The positive interaction between emulsion load and surfactant concentration on current efficiency, L S = 7.4% (Table 6.7), suggests that at 'high' flow rate A336 is more effective. Indeed, comparing entries nr. 4 and 7 in Table 6.6, shows that with 1 m M A336 the current efficiency improves from 55 to 77%. It is hypothesized that under improved mass transfer conditions the surface coverage of the three-dimensional electrode by the surfactant film is more uniform. The interplay between fluid dynamics, electrode potential and porous structure on the cationic surfactant deposition process (i.e. thin film formation) should be investigated in future work, to 'optimize' the surfactant effect on the present system. A s mentioned in preceding sections, the influence of the surfactant on the system can be attributed to a combination of physico-chemical (e.g. wetting, spreading) and electrochemical (e.g. electrode kinetic) effects. The curvature effects, calculated as the difference between the centerpoint and the factorial responses, are positive, + 6.5 % and 3.9 m M , respectively. This shows a degree of non-linearity in the behaviour of the variables, with the 'optimum' conditions lying closer to the centerpoint values (Table 6.5). Based on the factorial experiments, neglecting the statistically insignificant values, the regression equation for current efficiency can be written as: C E r e g , a c i d l e m u l ( % ) = 59.5 - 15.3x, + 18.5xL + 5.%xs + %.QSxIxL +3.1xLxs, [6.20] where the coded variables x, are expressed by: /•-2000 1 -7 .25 , S-Q.S x,, = ;xL = a n d x ? = , [6 21] 1000 L 4.41 s 0.5 1 1 with /' current density (A m"2), L emulsion load (kg m ' 2 s'1) and S surfactant concentration per emulsion (mM). In the above equations the load L and its coded variable xL, can be replaced by total emulsion flow rate Fv(m3 s"1) and xp, i.e. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 163 i v -7 .67-10" 7 X ' = "4 .67 . !0- ' • ' 6 2 2 ' 4) Influence of the Flow Regimes Both the factorial and the parametric experiments presented so far were performed in the liquid continuous (bubble flow) regime. Therefore, it was of interest to study the influence of the O2 gas load, and implicitly the effect of different flow regimes, on the performance of the multi-phase electrosynthesis. The fixed bed cell, equipped with the graphite felt cathode, was operated at constant liquid (emulsion) load (7.57 kg ni" 2 s'1, 8xl0" 7 m 3 s"1) and various 0 2 gas loads ranging from 1.73xl0" 2 to 6.22x10"' kg m"2 s"1. The above values yield liquid to gas load ratios (L/G) between 12 and 438, thus, covering the entire range of regimes for upward co-current flow between the gas continuous and liquid continuous zones [Storck et al., 1986]. Fig. 6.19 shows the H2O2 concentration per pass as a function of O2 gas load for the emulsion mediated electrosynthesis at 3000 A m"2, 313 K , and 0.1 M P a . The organic to acid aqueous phase volume ratio was 0.3. Interestingly, the O2 load had virtually no influence on the H2O2 concentration over the entire range of gas loads (Fig. 6.19). Increasing the O2 load decreases the liquid hold-up in the graphite felt from ca 0.85 to 0.7 (eqn. [3.25]). Therefore, one would expect that due to the increased surface coverage by the gas phase, the electroreduction of E t A Q from the liquid phase (i.e. emulsion) would be less efficient at high gas loads (e.g. > 0.3 kg m~2 s"1). The constancy of H2O2 concentration as a function of O2 load (Fig. 6.19) might be due to the fact, the surface coverage by the organic phase, and especially the S / L / L contact area (i.e. principal E t A Q reaction zone, Fig. 6.11) might not decrease at high gas load, due to the A336, which helps maintain a very thin organic layer in contact with the electrode. Consequently, it is hypothesized that for the present system the very thin emulsion layer, which is in contact with the graphite felt surface, i.e. the boundary where the electroreduction of E t A Q takes place, is undisturbed even in the 'gas continuous' regime. This hypothesis is supported by G / L flow studies in packed beds, which indicate that for small pore sizes the gas flow through the 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 164 restricted channels is essentially laminar [Takahashi and Alkire, 1985], thus, producing little boundary layer disturbance. A s a result of the decreased liquid hold-up, the cell voltage increased slightly with gas load, e.g. from ca. 12.3 V for G < 0.12 kg m"2 s"1 up to 13.2 V for 0.62 kg m"2 s"1. co c CD O c o o CNI O CM X 40 £ 30 20 H 10 L cont. (Bubble flow) Surg ing f low G a s cont inuous (Pu ls ing flow) — i | i 1 1 1 1 1-0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Oxygen Load / kg nr 2 s 1 Figure 6.19 The influence of flow regimes on H 2 0 2 concentration for the emulsion mediated electrosynthesis with acid aqueous phase. Liquid load: 7.57 kg m"2 s"1. Graphite Felt. 3000 A m"2. 0.1 M P a . 313 K . (Org / A q u ) v / v 0.3. A336: 1 m M per emulsion. 6.4 Emulsion Mediated Electrosynthesis ofH202 in Elow-by' Cells 165 e) 0.5 m Long Cell, Complete Recycle of the Catholyte The last investigation of the mediated peroxide electrosynthesis in acid emulsions involved experiments using the 0.5 m long cell equipped with the compressed graphite felt cathode and a dimensionally stable 0 2 anode (DSA) separated by a Nafion®350 membrane (Section 4.2.2). The experimental conditions were selected based on the 'know-how' gained with the factorial and parametric experiments performed on the 'small' (0.14 m long) flow cell. Thus, the catholyte was composed of a 0.3 volume ratio of organic and aqueous (i.e. 1 M Na2SC>4, pHin 3) phases. The organic media contained 0.2 M E t A Q and 0.1 M T B A P dissolved in T B P . The cationic surfactant concentration per emulsion was 1 m M . The liquid emulsion flow was 2.47x10"6 m 3 s"1, the highest achievable flow from the employed pump. The corresponding liquid load was 10 kg m - 2 s"1. The gas flow rate was 1.52xl0"5 m 3 s_1 at STP or 8 .8x l0 - 2 kg nf 2 s"1, which assured a 'bubble' flow regime operation of the cell. The total recirculated emulsion volume was 3.5 1. Based on the flow rate in ca 24 minutes the entire 3.5 1 volume of emulsion passed through the reactor. The anolyte was 0.5 M N a 2 S 0 4 (pHin 2.9) operated in single-pass mode at a flow rate of 9.17x10 The reactor was operated in galvanostatic mode at a cathode superficial current density of 1000 A mf2 (i.e. current 26.7 A ) and constant temperature of 308 K (see Section 4.2.2). The H2O2 concentration, current efficiency and outlet p H of the aqueous phase were followed over time (Figs. 6.20 and 6.21). Fig. 6.20 shows that in the first hour of recycling {ca 3 complete passes of the catholyte through the reactor), the average current efficiency is over 85%, generating 0.12 M H2O2 per emulsion. A s the recycling of the emulsion progressed without H2O2 separation, the current efficiency decreased (e.g. 50%, 0.24 M H 2 0 2 after 3 hours). Continuing the experiment to 5 hours, the peroxide concentration leveled around 0.28 - 0.3 M corresponding to 39% current efficiency. Unfortunately, due to the low current efficiency for H2O2 in the last hour, the outlet p H of the aqueous phase increased dramatically from 5.5 to 10.5 (Fig. 6.21). Furthermore, the cell voltage increased from 5.6 to 6.5 V (Fig. 6.21). The possibility of membrane fouling due to the organic phase must be considered as a factor in the increase of cell voltage. Generally, the 6.4 Emulsion Mediated Electrosynthesis of H202 in 'Flow-by' Cells 166 separator performance is a very challenging and crucial issue particularly in the context o f electro-organic synthesis, as it was well described by Danly for the case of the electrochemical adiponitrile process [Danly, 1982]. Future work on the present system must address the problem of membrane performance considering even the possibility of eliminating the separator by developing a superior cell design (see also Chapter 8: 'Recommendations for Future Work ' ) . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 167 Time / min Figure 6.20 'Flow-by' cell with complete recycle of the catholyte: H2O2 concentration in emulsion and current efficiency for the E t A Q mediated electrosynthesis. Cathode: graphite felt ( 0.5 m long, 267x10"4 m 2). Current density: 1000 A m"2. Aqueous phase: 1 M N a 2 S 0 4 , pHin 3. Organic phase: T B P with 0.2 M E t A Q , 0.1 M T B A P . 1 m M A336 per emulsion. (Organic / Aqueous) v/ v : 0.3. Emulsion load: 10 kg m"2 s"1. 0.1 M P a outlet pressure. T: 313 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 168 ~j i i i i | i i i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — I 5.5 0 50 100 150 200 250 300 Time / min. Figure 6.21 'Flow-by' cell with complete recycle of the catholyte: Outlet p H and Cell voltage for the E t A Q mediated electrosynthesis. Cathode: graphite felt (0.5 m long, 267x10 - 4 m 2). Current density: 1000 A nf 2. Aqueous phase: 1 M N a 2 S 0 4 , p H i n 3. Organic phase: T B P with 0.2 M E t A Q , 0.1 M T B A P . 1 m M A336 per emulsion. (Organic / Aqueous) v / v : 0.3. 0.1 M P a . T: 313 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 169 6.4.2 Aqueous phase: 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 ; pHi„ 9.3. a) Cathode Selection Using the 'short' (0.14 m long) flow cell, the peroxide concentration and current efficiency for single-pass were compared for graphite felt, R V C _ 1 0 0 and R V C 3 0 (Fig. 6.22). The anode was N i felt, with 1.5 M N a O H as the anolyte, with a flow rate of 9.2xl0" 7 m 3 s'1. Nafion® 350 was employed as separator (Section 4.2.2). The flow cell was operated in the bubble flow regime [Storck et al., 1986], with atmospheric outlet pressure, at 308 K . Fig. 6.22 shows that up to about 400 A m~2 all three cathodes behaved similarly, e.g. current efficiency >50%. A t higher superficial current densities however, the peroxide concentration and current efficiency were the highest for the 100 ppi R V C . In contrast with the results obtained for the acid aqueous phase (Fig. 6.17), the performance of graphite felt was poor, e.g. the current efficiencies were below 30% for superficial current densities >1000 A m"2 (Fig. 6.22). Clearly, the behaviour of graphite felt in the case of alkali emulsion is not in accordance with the high liquid to solid mass transfer capacity of this porous electrode. The difference in G F performance between the acid and alkali emulsions, could be related to the structural dissimilarity between the two emulsions (i.e. bicontinuous and aqueous continuous O A V emulsion, respectively, Section 6.1.3). As shown in the previous section, the mass transfer capacity is approximately eight times higher for G F than for R V C _ 1 0 0 (Table 6.4), due to enhanced boundary layer turbulence. Although, this phenomena would be desirable for most cases, Wendt and co-workers [Dworak and Wendt, 1977; Dworak et al.,\919] showed that for O A V emulsions, where the 'wetting' mechanism should be operative (see Section 3.2.3), intense local turbulence minimizes the duration and extent of contact between the organic droplets and the electrode surface. This observation is relevant for the present alkali (OAV) emulsion because unlike in the acid emulsion case, the entire organic/solid interface represents the 'reaction zone' where the E t A Q reduction occurs (Section 6.2.4). Thus, the efficient and prolonged surface coverage by the organic droplets might be difficult under conditions, which are inducing intense boundary layer 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 170 turbulence. This hypothesis is supported by capillary flow studies [Probstein, 1989], showing that the thickness of a liquid film adhering to the capillary wall increases with the radius of the capillary and with decreasing liquid to gas velocity ratio. Such a scenario would be desirable for efficient E t A Q reduction in the case of the alkali (OAV) emulsion. Another aspect of the emulsion mediated electrosynthesis that must be discussed, is the possibility of organic phase retention inside the porous electrode. Huissoud and Tissot, studying the peroxide electrosynthesis in a flow-by cell using an alkaline (2 M NaOH) emulsion of diethylbenzene / tributyl phosphate, reported the increased retention of the organic phase inside the porous matrix with increasing specific surface area of R V C . For example, a complete retention of the organic was reported for 45 ppi R V C during continuous catholyte recycle at 0.9 m s"1 [Huissoud and Tissot, 1999 II]. Under the conditions employed in the present work (see Fig. 6.22), only at very low flow rates, < 1.67xl0" 7m 3 s"1 (i.e. <10 ml min"1) and for small organic to aqueous volume ratio (e.g. 1/10) a total hold-up of the organic phase was observed, for both R V C _ 1 0 0 and graphite felt cathodes. Therefore, the poor performance of graphite felt in comparison with 100 ppi R V C under conditions that did not lead to the retention of the organic phase (e.g. Fig. 6.22) was due to other effects, such as boundary layer turbulence, discussed above. The 30 ppi R V C behaved similarly to the acid emulsion case (compare Figs. 6.21 and 6.17), the mass transfer capacity is too small to sustain current densities > 400 A m"2. Based on the results presented in Fig. 6.22, for the alkali emulsion mediated electrosynthesis of peroxide, in flow-by cells the 100 ppi R V C was the preferred cathode. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 171 Figure 6.22 H2O2 concentration in emulsion and current efficiency for one-pass, E t A Q mediated, electrosynthesis in the flow-by cell. Aqueous phase: 1 M N a H C 0 3 / 0.5 M Na2C03, p H i n 9.3. Superficial area: 3.29xl0" 3 m 2 . Organic phase: T B P with 0.2 M E t A Q , 0.1 M T B A P and 1 m M A336. (Organic / Aqueous) v/ v : 1 / 1.9. Emulsion flow: 3 .83xl0 - 7 m 3 s~l. O2 flow (at STP): 2.34xl0" 6 m 3 s"1. 0.1 M P a . T: 308 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 172 b) Comparison between 'Direct' O2 Reduction and Emulsion Mediation on RVC J 00 The emulsion mediated electrosynthesis operated at atmospheric pressure was compared with the 'direct' electroreduction of O2 at 275 and 400 kPa, respectively, in the alkaline electrolyte composed of 1 M N a H C C h / 0.5 M Na2CC>3. The purpose of these experiments was to determine the importance of a complicated L / L emulsion system with mediator, relative to direct 0 2 reduction in 100 ppi R V C (Fig. 6.23). The liquid flow rate was 3.92xl0" 7 m 3 s"1 corresponding to a load of 3.5 kg m"2 s"1. The O2 flow rate at STP was 2.34xl0" 6 m 3 s"1 (load 0.028 kg m"2 s"1) for the runs at 100 and 275 kPa whilst for the 400 kPa run the gas flow rate was set to 1.22xl0"5 m 3 s"1 at STP (load 0.15 kg m"2 s"1). The organic-aqueous phase volume ratio for the emulsion mediated system was 0.52. Fig. 6.23 shows that the peroxide concentration per pass for the 'direct' O2 electroreduction on R V C _ 1 0 0 decreases for current densities exceeding 1000 A m"2 even at 'high', 400 kPa, O2 pressure. The peroxide concentration for the mediated system however, increased up to ca 2000 A m"2, yielding 35 m M H2O2 vs. 15 m M for 'direct' reduction at 400 kPa. Therefore, the presence of the mediator is important, especially to operate the flow-cell at current densities higher than 1000 A m"2. It must be emphasized that the figures of merit for the emulsion mediated system on R V C _ 1 0 0 are worse then those given by the direct O 2 / A 3 3 6 system on graphite felt under comparable conditions (see Figs. 6.27 and 5.13). 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 173 Figure 6.23 Comparison between 'direct' aqueous O2 reduction and emulsion mediation at atmospheric pressure. Cathode: 100 ppi R V C (area: 32 .9x l0 - 4 m 2). Aqueous electrolyte: 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 , p H i n 9.3. Organic phase: Organic phase: T B P with 0.2 M E t A Q , 0.1 M T B A P and 1 m M A336. (Organic / Aqueous) v / v : 1 / 1.9. Liquid load: 3.5 kg i n 2 s"1. 0 2 load (kg m - 2 s_1): 0.15 (for 400 kPa); and 0.028 (for 100 and 275 kPa). 303 K . 6.4 Emulsion Mediated Electrosynthesis of H202 in 'Flow-by' Cells 174 c) Factorial Experimental Design: Alkali Emulsion Mediation As can be seen from Fig. 6.23, the main advantage of using the emulsion mediated system vs. 'direct' alkali 0 2 reduction, lies in the fact that the peroxide concentration per pass increases for current densities up to ca. 2500 A m"2. Therefore, a factorial experiment was performed to explore the possibility of operating the emulsion mediated electrosynthesis at a constant superficial current density of 2500 A m"2 (current 8.2 A ) . The aqueous phase was composed of 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 (pHi„ 9.3), whilst the organic (TBP) solution contained 0.2 M E t A Q , 0.1 M T B A P and various concentrations of A336.Three variables were selected (Table 6.8): total emulsion load (L) , organic/aqueous phase volume ratio (R) and A336 concentration (S). The outlet reactor pressure (0.1 MPa) , the 0 2 load (0.022 kg m"2 s"1 at STP), temperature (318 K ±3) and current density (2500 A m"2) were constants. Table 6.8 Variables and their values used in the factorial study of the emulsion mediated H 2 0 2 electrosynthesis using 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 (pHj„ 9.3) as aqueous electrolyte. Cathode: 100 ppi R V C (area: 32.9xl0" 4 m 2). Variable / symbol (units) ' L O W level (-) ' C E N T E R ' (0) 'HIGH' level (+) 1. L / L Emulsion Load (L) (kgm- 2 s"1) 2.7 3.6 5.4 2. Organic / Aqueous Volume Ratio (R) 0.28 0.47 0.96 3. A336 concentration (S) (mM) per emulsion 0 0.5 1 6.4 Emulsion Mediated Electrosynthesis of H2O2 in Elow-by' Cells 175 Table 6.9 Factorial design: Emulsion H2O2 concentration and current efficiency per one-pass, for the mediated electrosynthesis on R V C _ 1 0 0 , using 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 ( p H i n 9.3) as aqueous phase. T: 318 K . 0.1 M P a . Current density: 2500 A m' 2 . For symbols, see Table 6.8. N r . L R s Cperoxide ( m M ) C E ** (%) 1 - - - 16 11 2 + + + 15 21 3 + + - 16 22 4 0 0 0 21 ±1.6 20 ±1.4 5 + - - 12 17 6 - + - 20 14 7 + - + 17 24 8 - - + 23 16 9 - + + 48 34 * Average factorial response: 20.9 mM. Pooled standard deviation: ±1.6 mM. ** Average factorial response: 19.9%. Pooled standard deviation: ±1.4%. The 2 3 +l factorial experiment presented in Table 6.9, revealed that generally at 2500 A m 2 the current efficiencies are low, between 11 and 34%. Under certain conditions (such as ' low' liquid load), the cationic surfactant improved markedly the figures of merit, for instance compare runs 6 and 9, where the current efficiency increased by 140% due to the presence of 1 m M A336 in the emulsion. Furthermore, the runs with 'high' organic/aqueous phase volume ratio (R), yielded better figures of merit, e.g. run 6 vs. 1 or run 3 vs. 5. This result is in contrast with the acid emulsion case where the phase volume ratio had virtually no influence (Table 6.7). The cell voltage for the experiments presented in Table 6.9, varied between 10.5 and 11.8 V . The 6.4 Emulsion Mediated Electrosynthesis ofH202 in 'Flow-by' Cells 176 outlet p H of the aqueous electrolyte was between 9.9 and 13. The p H increased with decreasing liquid flow. The main and interaction effects associated with the 2 3 +l factorial runs are shown in Table 6.10. Table 6.10 Main and interaction effects for the emulsion mediated electrosynthesis using 1 M NaHCOa / 0.5 M N a 2 C 0 3 ( p H i n 9.3) as aqueous phase. Cathode: R V C 1 0 0 . 2500 A m " 2 . 0.1 M P a . 318 K . E F F E C T S P E R O X I D E concentration* (mM) C U R R E N T EFFICIENCY** (%) M A I N L -11.8 2.2 R 7.9 5.7 S 9.7 7.8 Two-factor I N T E R A C T I O N L R -7.0 -4.7 L S -7.8 -4.8 R S 3.8 1.7 Three-factor I N T E R A C T I O N L R S -6.9 -5.7 * standard error of effect: ± 0.9 mM ** standard error of effect: +1% 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 111 Table 6.10 shows that in addition to the main effects all the interaction effects are statistically significant, unlike in the acid emulsion case (compare Tables 6.7 and 6.10). Another difference vs. the acid emulsion is the positive main effect of the organic/aqueous phase volume ratio R, on both peroxide concentration and current efficiency. Furthermore, the cationic surfactant had the strongest main effect on current efficiency, i.e. increasing the A336 concentration to 1 m M per emulsion, improved the current efficiency by almost 8% (Table 6.10). It is interesting to examine the interaction effects involving the liquid load (L) and the other two variables R and S, respectively. The negative two-factor interaction between L and the other two variables on both peroxide concentration and current efficiency, indicates that operating the present system at 'high' liquid load is not beneficial. Thus, the factorial experiments support the hypothesis put forward in Section 6.4.2 a, that factors which induce local boundary layer turbulence, such as a high liquid load, are undesirable since they minimize the duration and extent of contact between the organic droplets and the electrode surface in the case of an O A V emulsion. A s shown by the interaction effects L R and L S , the 'high' levels of R and S make a significant positive contribution only for ' low' L . For instance, in Table 6.9 compare runs 2 and 3 (at 'high' L ) and runs 6 and 9 (at ' low' L ) . The curvature effect was negligible. 6.4 Emulsion Mediated Electrosynthesis of H202 in Elow-by' Cells 178 d) The Influence of Flow Regimes The factorial experiments presented above were performed in the liquid continuous ('bubble') flow regime at a constant O2 gas load of 0.022 kg m"2 s"1. Based on the observation that in the case of alkali (OAV) emulsion the boundary layer turbulence must be minimized to favor the attachment of the organic droplets to the cathode surface, it is hypothesized that increasing the O2 load wil l lower the peroxide concentration per pass. To verify this hypothesis experiments were performed under the 'centerpoint' conditions (Table 6.8) with varying O2 gas loads (Fig. 6.24). Fig. 6.24 shows that in the 'bubble' flow regime the peroxide concentration is independent of the O2 load, however, the peroxide concentration per pass drops by about 25% when the O2 load is increased to values in the surging flow regime (0.1 to 0.3 kg m"2 s"1). The enhanced boundary layer turbulence created by the 'surging' and/or 'pulsing' flows is unfavorable for E t A Q reduction from the organic droplets of the O A V emulsion. A comparison of the O2 load effect on emulsion mediation in acid and alkali, (see Figs. 6.19 and 6.24) underlines the remarkable difference in behaviour between the two systems and also, the importance of the emulsion structure in determining the outcome of the electrosynthesis. The last section of the present chapter presents an evaluation of the economic feasibility of the emulsion mediated H 2 0 2 based on the 'best scenario' conditions determined so far. 6.4 Emulsion Mediated Electrosynthesis of H202 in 'Flow-by' Cells c o CO c CD O c o o CD •g o CD Q_ 25 20 H 15 H 10 L cont inuous (Bubble flow) Surg ing f low G cont inuous (Pu ls ing flow) 1 I 1 I 1 r 0.0 0.1 0.2 0.3 0.4 2 „-1 O x y g e n L o a d / kg m s Figure 6.24 The influence of flow regimes on the alkali emulsion mediated peroxide electrosynthesis. Cathode: 100 ppi R V C . Conditions idem 'centerpoint' (Table 6.8). 7 Process Synthesis and Economics 180 C H A P T E R 7: Process Synthesis and Economics. An Outlook 7.1 Introduction The previous chapters of the present thesis were concerned with fundamental and applied electrochemical studies related to the use of a mediator in a three-phase ( L / L / G ) system for peroxide electrosynthesis. The scientific merit, however, is only one aspect in determining the potential for larger scale application of a novel technology. This is especially true for electrochemical methods of synthesis, which ought to compete with corresponding thermochemical processes. Therefore, a number of issues, such as economic benefits, environmental friendliness and also, marketing of a novel on-site electrochemical technology to a traditional, natural resource based industry such as the pulp and paper, are of vital importance. A complete answer to all these challenges is out of scope and virtually impossible based only on the studies performed in the present work. However, a preliminary assessment of the economic potential via a level 2 process synthesis scheme, might serve as a useful guidance for future work in this area. 7.2 General Assumptions The level 2 process synthesis and the associated economic evaluation was performed for the acid emulsion (i.e. 1 M Na2S04 aqueous phase, p H m ~ 3) mediated H2O2 electrogeneration for the following reasons: i) From the point of view of peroxide bleaching technologies, acidic peroxide bleaching processes are in a much more advanced stage of development than the ones which are using Na2C03 as a partial or total replacement for N a O H in alkaline peroxide bleaching. In fact, the molybdate activated acidic (pH 5) peroxide bleaching (e.g. 'GreenOx' process of Kemira Chemicals, Finland), has already been tested and implemented in pulp mills [Paren and Tsujino, 1998]. Thus, the market for the end 7 Process Synthesis and Economics 181 product of the acid emulsion mediated electrosynthesis is already established, whilst for an alkali, Na2C03 / NaHCCb based peroxide solution, is not in-sight yet. ii) The acidic peroxide bleaching (e.g. DP c a t , GreenOx) is performed in relatively dilute H 2 0 2 solutions, i.e. 0.2% w (60 mM) for E C F and 0.7% w (200 mM) for T C F bleaching, respectively. The typical acidic peroxide bleaching conditions for an E C F sequence, are: 15 kg H2O2 per ton of pulp, 12% pulp consistency, 0.45 kg N a 2 M 0 4 catalyst per ton of pulp, 90 °C and 3 hrs bleaching time at p H 4 . 5 - 5 [Koshitsuka, 1998; Paren and Tsujino, 1998]. Taking into account the water content of the pulp, an electrochemical process generating ca 0.8 M (2.8%w) of H2O2 at p H 5, would suffice the requirements of the bleaching sequence. iii) The economic analysis is performed for a potential electrochemical plant located on-site of the pulp mill. The on-site and on-demand features could be very important for the end-user since it would eliminate the transportation and storage costs associated with merchant peroxide*. Furthermore, it is assumed that the mill uses 10 ton per day H2O2 (100%) basis). This value was chosen in order to match the typical production rate of the commercial alkaline O2 reduction D o w technology [Mathur and Daw, 1999; D o w Inc., 1999]. For the bleaching conditions mentioned above, 10 ton per day peroxide consumption means a production of ca 240,000 tons per year of E C F bleached pulp or ca 68,500 tons per year of T C F bleached pulp. iv) Regarding the bleaching liquor, it is assumed that both the electrochemical and the bleaching process is able to use a byproduct from the mill, the so-called 'spent acid', which is a mixture of N a 2 S 0 4 and H 2 S 0 4 , formed in the CIO2 generator of a kraft mill [Oloman, 1996]. Based on the above assumptions, the objective of the process synthesis is to evaluate the feasibility of the acid emulsion mediated system for on-site production of 10 t per day (12.3 kmol h"1) H2O2 (in p H 5 solutions of 2 .8% w (0.8 M ) H2O2 concentration), by performing a * Generally, pulp mills are remote located, close to the fibre source. 7 Process Synthesis and Economics 182 sensitivity analysis of the net economic potential (NEP) and return on investment (ROI) as a function of superficial current density. 7.3 Operational Conditions The following operational conditions are employed in the sensitivity analysis of the economic indicators: Cell components: Graphite felt cathode, Nafion® 350, D S A 0 2 anode Superficial cathode current density: 600 to 5600 A m ' 2 Current efficiency: 85% (assumed constant for simplicity) Cell voltage: 2 to 7 V (assumed based on typical operating voltage ranges of commercial electrosynthesis reactors, [Oloman, 1996]) Operation mode: Monopolar, galvanostatic, one-pass co-current upward G / L flow Pressure and temperature: 0.1 M P a , 313 K . Phase volume ratio: Organic/Aqueous = 1/3V. Aqueous catholyte: Inlet composition: 1 M Na2S04 at p H 3 (it is assumed that the acidity is obtained by diluting the CIO2 generator effluent, which contains H2SO4). Outlet composition: 1 M N a 2 S 0 4 , 0.8 M (i.e. 2.8%w) H 2 0 2 , p H 5. Flow: 15 m 3 h ' 1 Organic catholyte: Composition: T B P with 0.2 M E t A Q , 0.1 M T B A P and 4 m M A336 (i.e. 1 m M per emulsion) Flow: 5 m 3 h"1 Oxygen: Flow: 120 m 3 h"1 at STP (based on the ratio of L / G flow rate employed in Fig. 6.20). 7 Process Synthesis and Economics 183 Anolyte: Composition: 1.5 M Na 2 S04, pH;„ 3 Flow: 30 m'h" 1 . 7.4 Level 1 Analysis : Gross Economic Potential ( G E P ) The conceptual flowsheet corresponding to level 1 is given in Fig. 7.1. The net stoichiometry is: H 2 0 + ^ - 0 2 - > F f 2 0 2 . [7.1] Based on Table 7.1 and eqn. [7.1], the gross economic potential (GEP) associated with the production of 12.3 kmol h"1 peroxide is: G E P = [Value of Products] - [Value of Feeds]. G E P = 12.3x34x0.8 - (6.1x32x0.05+12.3x18x0.01) 323 $ h" 1 [7.2] The targeted price of 0.8 $ kg" 1 for H 2 0 2 (100% basis; Table 7.1), is about the current market price of commercial peroxide including the associated transportation and storage costs [manufacturing.net, 2000]. TBP (1) EtAQ (2) TBAP (3) A336 (4) Na 2 S0 4 H 2 0 0 2 (5) (6) (7) REACT A >. TBP (11) • EtAQ (12) • TBAP (13) >• A336(14) H202 H 2 0 Na 2 S0 4 (8) (9) (10) Figure 7.1 Conceptual flowsheet for Level 1 analysis 7 Process Synthesis and Economics Table 7.1 Chemical costs and molar flow rates for Level 1. 184 T B P (M: 266.3) E t A Q (M: 236.3) T B A P (M: 342) A336 (M:404.2) N a 2 S 0 4 (M: 142) H 2 0 (M: 18) o2 (M: 32) H 2 0 2 (M: 34) Price ($ kg'1) 4.6 6.6 5 3 0.11 0.01 0.05 0.8 I N L E T (kmol h"1) 72.8 4 2 0.08 , 75 5,000 21.5 0 O U T L E T (kmol h"1) 72.8 4 2 0.08 75 4987.8 15.3 12.3 7.5 Level 2 Analysis : Net Economic Potential (NEP) and Return of Investment (ROI) The level 2 flowsheet gives a more realistic description of the process by specifying the type of equipment used for various operations plus the recycled streams. Therefore, capital and operational cost estimates can be obtained based on a preliminary design and sizing of the process equipment. Fig. 7.2 presents a conceptual flowsheet for the present process. The most complex and cost intensive process equipment is the electrochemical reactor unit (6), which usually is composed of a large number of individual cells (cells) fed in parallel with electrolyte (see further). In addition to the electrochemical cells, the process uses a number of unit operations (Fig. 7.2). For the L / L emulsion preparation and feed, a baffled mixing vessel with mechanical agitation is provided (2). The peroxide containing emulsion from the cathode compartment is passed through a G / L separator (7), followed by a decanter (8), to separate the two liquid phases. The organic phase from the decanter could be reused as such or after minor refreshment. The loss of organic phase due to the water solubility of T B P (0.59% w at 293 K ) is 0.38 kmol h"1 (i.e., ca 100 kg h"1 o f T B P , 4.8 kg h ' 1 E t A Q , 3.5 kg h"1 T B A P and 0.2 kg h ' 1 A336). 7 Process Synthesis and Economics 185 Since the presence of organic phase in the 'crude' aqueous might interfere with downstream pulp bleaching operations, and also to recover the costly organic, the 'crude' acid peroxide liquor is passed through an adsorption column (8) packed with an organic polymer adsorbent such as polyacrylic ester or polystyrene [Perry's, 1984]. After purification, the acid (pH 5) peroxide solution can be employed for bleaching as discussed earlier. The economic evaluation of the present process was based on preliminary design calculations for the equipment shown in Fig. 7.2. 7 Process Synthesis and Economics 186 t 1 Emulsion containing H 2 0 2 ^ cooling water ± 'crude' aqueous H 2 0 2 Organic phase (Recycle) Organic solvent residues acid H 2 0 2 bleaching liquor (PH 5) e-HUM* mm cooling water W W Figure 7.2 Level 2 flowsheet for the acid emulsion mediated electrosynthesis. Legend: 1. 0 2 feed tank; 2. L / L emulsion feed tank; 3. Pumps; 4. Anolyte feed tank; 5. Coolers; 6. Electrochemical reactor unit; 7. G / L Separator; 8. Decanter; 9. Adsorption column. 7 Process Synthesis and Economics 187 A s shown in Fig. 7.2, the most important component of the flowsheet, is the electrochemical reactor. The total installed capital cost of the electrochemical reactor unit CIE, is given by the sum of the installed capital cost of the cells Cmc and the installed capital cost of the power supply system, Cps. Thus, CIE = Cmc + Cps. [7.3] For the electrochemical cells, the installed capital cost Cmc, is expressed as [Oloman, 1996]: CIEC=Nc-Gc-(Ac)m, [7.4] where Nc number of cells, Gc cost constant expressed in $ m"2, Ac electrode area per cell (m 2) and m scale exponent (0.8 - 0.9). For the present case G is ca 24,000 $ m"2. This value was obtained by assuming a Lang factor of 3 [Janson, 1984], for an 'off the shelf cell cost of 8,000 $ m"2, which includes the cost of membrane (e.g. -1,000 $ m"2), cathode (graphite felt), and the D S A anode. The cathode area and the number of required cells are interconnected through Faraday's law, i.e. zFPr i-CE-A' NC = . : ; R , , [ 7 . 5 ] where z nr. o f electrons involved in the process (2), F Faraday's number (96,485 k C kmol" 1), Pr peroxide production rate (kmol s"1), / superficial cathode current density ( k A m"2) and CE current efficiency. The upper limit for the superficial electrode area in trickle bed electrochemical cells, is determined by the extent of axial and radial dispersion effects, pressure drops and other operational constraints [Oloman, 1979]. Here, a superficial cathode area per ce l iac , of 1 m 2 was employed. 7 Process Synthesis and Economics 188 The installed capital cost of the power supply system CPS, is [Oloman, 1996]. C = P ^PS 1 w B'+ C with t J P^Nc-i-Ac-Vc mdVt=Nc-Vc, [7.6] [7.7] where Pw total power consumption (kW), B' and C" cost constants (100 and 20,000 $ kW" 1 , respectively), Vt total voltage (V) and Vc voltage per cell (V). The cost of dc electrical energy associated with the electrochemical reactor unit (part of the operating costs), Cei in $ h"1, is [Oloman, 1996]: C P c h « £ w [7.8] el RE where CAC cost of ac electricity (0.03 $ kWh" 1 [Mathur and Dawe, 1999]) and RE rectifier efficiency (0.96). A n additional utility cost with the electrochemical unit, is for cooling water required to balance the Joule heating (see Appendix L). The maintenance and labour cost associated with the electrochemical plant CEML ($ year"1), is a function of the total applied current and the number of cells Nc, [La Roche, et al, 1994], i.e.: C ^ = 0 . 2 . t f c - i ^ + 7 0 0 - t f c . [7.9] The installed capital and operating costs for the auxiliary equipment, (see Fig. 7.2, i.e., mixing / feed tanks, pumps, heat exchangers, G / L separator, gravity decanter and adsorption column) were estimated based on their respective design equations and the associated cost relationships (see Appendix L). Basically, these costs are mostly dependent on the plant capacity (equipment size). Therefore, in the sensitivity analysis of the economic indicators as a function of current density, the costs corresponding to the auxiliary equipment are treated as constants. 7 Process Synthesis and Economics 189 As shown in Appendix L, the total installed capital cost for the auxiliary process equipment is ca 725,000 $( y e a r 2000 basis)- The labour and maintenance cost per year for basic chemical engineering processes, is typically 5% of the capital cost [Oloman, 1996]. Therefore, 30,000 $ year"1 was considered for the present case. The total operating cost o f the non-electrochemical equipment (i.e. utilities plus labour and maintenance) is ca 7.8 $ h"1 (see Appendix L). Substituting the pertinent numerical values given in Section 7.3, in the capital and operating cost calculations for the both the electrochemical unit and the auxiliary equipment, the net economic potential (NEP) and the yearly return on investment (ROI) were calculated, N E P ($ h"1) = G E P - Operating Costs (i.e. utilities + labour and maintenance), [7.10] R O I ( % ) = . ^ P " 8 0 0 0 100. [7.11] V peryear; T o t a l Totalled Capital COSt Generally, investment considerations are given to processes characterized by an ROI > 30% per year [Oloman, 1996]. Figures 7.3 and 7.4 summarize the dependence on superficial current density and individual cell voltage of four major economic factors, i.e. N E P , ROI, total installed capital cost (CIC) and total operating cost (COP), respectively. 7 Process Synthesis and Economics 190 600 1600 2600 3600 4600 5600 i / A m - 2 1 1 i 1 i 1 i 1 i 1 i 2 3 4 5 6 7 U / Volt Figure 7.3 The Net Economic Potential (NEP) and Return o f Investment per year (ROI) for acid emulsion mediated peroxide electrosynthesis, as a function of superficial current density and corresponding cell voltage. Current efficiency: 85%. Operational conditions: Section 7.3. 7 Process Synthesis and Economics 191 — I — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | ' — 140 600 1600 2600 3600 4600 5600 i / A i n - 2 1 i | i | i | 1 1 1 1 2 3 4 5 6 7 U / Volt Figure 7.4 Installed Capital Cost (CIC) and Total Operating Costs (COP) for the electrochemical peroxide plant, as a function of superficial current density and corresponding cell voltage. Current efficiency: 85%. Operational conditions: Section 7.3. 7 Process Synthesis and Economics 192 Figure 7.3 shows that the highest ROI , i.e., 19% p e r year, was obtained at ca 4000 A m" with a cell voltage of 5.6 V . However, N E P reached a maximum of 175 $ h"1, at a current density of 1600 A m"2 (Fig. 7.3). The difference between the N E P and R O I maxima is due to the interaction between the installed capital (CIC) and the operating (COP) costs, as shown by Fig. 7.4. The installed capital cost of the electrochemical unit decreases significantly with current density from over $30 million at 600 A m"2 to ca $5 million at 4000 A m"2, since fewer cells are required (e.g., 1288 at 600 A m"2 and 193 cells at 4000 A m"2, see eqns. [7.4] and [7.5]). The operating cost on the other hand, has its minimum at 1600 A m"2 (Fig. 7.4). A t a higher current density the operating cost is dominated by the electricity consumption while at lower than 1000 A m"2 the operating cost is high due to the extensive labour and maintenance cost associated with the large number of cells (see eqn. [7.9]). Therefore, the lower installed capital cost at 4000 A m"2 increases the R O I (eqn. [7.11]) whilst the low operating cost at 1600 A m"2 maximizes the N E P . These two indicators, N E P and ROI, offer two different economic viability models for the present process. A high ROI (i.e. preferably above 30%Per year) wi l l recover the initial investment but from the end user's point of view, i.e. pulp mill, the N E P should be high to maximize long-term profits. Consequently, the 'best' strategy would be to increase the R O I at the maximum N E P (i.e., 1600 A m"2, 2.4 - 2.6 V per cell). The most important technical improvement of the process that could lower the installed capital cost is related to the electrochemical reactor design. For instance, lowering the uninstalled reactor cost by 50%, i.e., 4000 $ m"2 instead of the present 8000 $ m"2, gives a maximum R O I of 3 1 % p e r year at 3600 A m " 2 . A n undivided reactor design might accomplish this objective. B y eliminating the need for a cation exchange membrane and possibly the anolyte circuit as well, in addition to lower uninstalled capital cost, significant savings would be achieved on the operating costs due to lower cell voltage and simpler maintenance. 8 Conclusions and Recommendations 193 Chapter 8: Conclusions and Recommendations for Future Work 8.1 Conclusions Two novel methods for the electroreduction of 0 2 to H2O2 in a variety of acidic and alkaline solutions have been conceived and investigated in this work. The first method, unreported before in the literature, is based on the use of surfactants to enhance the direct, 2e" reduction of O2, by modifying certain interfacial properties, such as surface p H and O2 concentration (Chapter 5). The second method (Chapters 6 and 7), uses a three-phase L / L / G system composed of: a) an organic media (composed of the redox mediator, E t A Q ; supporting electrolyte, T B A P ; and cationic surfactant, A336; dissolved in a solvent, TBP) , b) an aqueous electrolyte of choice and c) oxygen gas. Although there are a few literature reports on emulsion mediated 2e" reduction of O2, published while the present study was in progress [Huissoud and Tissot, 1999 I and II], the novelty of the study described here lies in: • the different composition of the aqueous and organic phases (e.g., emulsion mediation was not tested before in acidic electrolytes) • the use of a surfactant in conjunction with the L / L / G system • the in-depth study and synergistic analysis of several phenomena such as H2O2 partition, emulsification, electroreduction mechanism and multi-phase flow dynamics, and • the economic analysis for in-situ H 2 0 2 production by acid emulsion mediation. 8.1.1 Surfactant modified electroreduction of 0 2 to H 2 0 2 The influence on the 2e" reduction o f O2 o f three surfactants representative of their classes (cationic, non-ionic and anionic), was investigated using vitreous carbon and graphite felt cathodes. Cyclic voltammetry at 295 K in 0.1 M Na2CC>3 and H2SO4, respectively, provided an insight into the effects of surface film formation on both the electroreduction kinetics and O2 transport. It was found that a triple-Cs chain cationic surfactant, Aliquat 336 (i.e., tricaprylmethylammonium chloride, [CH3(CH2)7]3CH3N+C1") increased the rate of O2 reduction 8 Conclusions and Recommendations 194 to H2O2 in both electrolytes. This was attributed to an increase of the surface p H induced by the Aliquat 336® surface structures. The non-ionic and anionic surfactants (Triton X-100 and sodium dodecyl sulfate, respectively) retarded the electroreduction o f 0 2 presumably by forming less organized, entangled, surface aggregates, which blocked the access of O2 to the cathode (Sections 5.1 -5.8). In addition to the effect on O2 reduction, separate experiments showed that surfactant film formation on the electrode surface, especially the Aliquat 336® film, could reduce the rate of H2O2 electroreduction (Section 5.2.2). The combination of beneficial effects on both O2 and H2O2 electroreduction warranted further electrosynthesis experiments involving this cationic surfactant. Batch electrosynthesis on a 30 ppi R V C cathode corroborated the cyclic voltammetry data (Section 5.2.3). In the presence of 0.8 - 2.5 m M Aliquat 336®, at 300 A m"2, 0.1 M P a 0 2 and 300 K , peroxide concentrations up to 310 m M in alkali (current efficiency 61%) and 260 m M in acid (current efficiency 55%) were obtained. Without cationic surfactant present, the maximum peroxide concentration in acid was ca 70 m M (current efficiency 14%) and 14 m M in alkali (current efficiency 7%), respectively. Based on the encouraging results from batch electrosynthesis, further studies were performed in 'flow-by' cells operated in one-pass mode, under conditions closer to potential industrial applications, i.e., superficial current densities up to 6500 A m"2, and O2 pressures up to 700 kPa. A comparative investigation of three cathodes R V C 30 and 100 ppi, and graphite felt, showed a direct correlation between mass transfer capacity and performance. Therefore, graphite felt (with the highest mass transfer capacity, i.e. 6.7 s"1) was selected as cathode for further experiments. The factorial experiments using the 0.14 m effective length 'flow-by cell' showed the statistically significant positive main effect of Aliquat 336® for both acidic (i.e., 1 M Na 2 S04 p H 3) and alkali (i.e., 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 pH9.6) solutions. A n interesting aspect revealed by the statistical design, is the negative interaction between pressure and surfactant (e.g., Section 5.3.4). In other words, the surfactant is more effective at atmospheric pressure than at high pressures. A n explanation based on capillary flow phenomena is provided. 8 Conclusions and Recommendations 195 Experiments performed in the acidic electrolyte using the 0.5 m effective length 'flow-by' cell indicated ca 62% improvement in H2O2 current efficiency, from 24% without A3 3 6 to 39% with 1 m M A336, at 1050 A m"2. Finally, for the surfactant modified 2e" reduction of 0 2 , based on the experimental evidence obtained in the present study it can be stated that modifying the double-layer structure and interfacial properties via the adsorption of long-chain cationic surfactants represents an interesting and novel approach for peroxide electrosynthesis, which opens up new venues for further research in this area and might lead to industrial applications (see Section 8.2.1, 'Recommendations for Future Work ' ) . 8.1.2 Emulsion mediated electroreduction of 0 2 The second method for peroxide electrogeneration investigated in the present work, made use of an organic redox mediator (i.e., 2-ethyl anthraquinone, E t A Q ) and an emulsion catholyte, to affect the 2e" O2 reduction for in-situ H2O2 electrosynthesis in various aqueous electrolytes. The complex relation among several factors, such as physico-chemical and electrochemical properties of the organic media, emulsification characteristics, mechanism of E t A Q electroreduction and multi-phase flow dynamics in porous media, has been studied by both fundamental and applied electrochemical techniques using parametric and statistically designed experiments (Chapter 6). Furthermore, an economic outlook is provided extrapolated from the experimental information available to date. Taking into account environmental factors, physico-chemical and electrochemical stability and the peroxide partition coefficient between the organic and various acid and alkaline aqueous phases, tributyl phosphate (TBP) was selected as the organic solvent for mediation. To lower the overall energy consumption and improve the current distribution between the organic and aqueous phases, tetrabutylammonium perchlorate ( T B A P ) was dissolved in the organic solvent as supporting electrolyte. Both the E t A Q mediator and T B A P concentrations in T B P were between 0.1 and 0.2 M . The latter concentration value is close to the solubility limit at room temperature (295 K ) for both compounds. The cationic surfactant Aliquat 336® was employed in the mediated system as well, where in addition to the beneficial influence on the direct O2 reduction, it might promote the selective wetting of the cathode by the organic phase. 8 Conclusions and Recommendations 196 The emulsification process between each relevant aqueous phases (i.e., 0.1 M Na2CC>3 and 0.1 M H2SO4, respectively) and the organic media was studied by conductivity measurements. It was found that in the case of alkali aqueous phase the emulsion is of 'organic-in-water (OAV) ' type, whilst, interestingly for the acid aqueous phase the overall emulsion conductivity was about half of the two 'pure' phase values, suggesting a complex, perhaps bicontinuous, emulsion structure. The nature of the emulsion had profound implications on the mass transfer characteristics in porous electrodes, impacting the overall figures of merit for electrosynthesis. The electroreduction mechanism in aprotic and protic organic media of both E t A Q and O2, individually and in conjunction, was studied by cyclic voltammetry. It was found that in the presence of the alkali aqueous phase (0.1 M Na2COs) the electroreduction of E t A Q occurs by two consecutive one-electron transfer steps (i.e., EE mechanism), generating E t A Q 2 " [Na+]2. For acid aqueous phase (0.1 M H2SO4) on the other hand, the first electroreduction step is followed by the fast disproportionation of E t A Q H * yielding EtAQH2 (i.e., EC mechanism). Both the standard heterogeneous rate constants for E t A Q reduction and the pseudo-first order homogeneous intrinsic rate constant for E t A Q H 2 oxidation were more than an order of magnitude higher for the alkali vs. the acid aqueous electrolyte. Separate spectrophotometric experiments confirmed the proposed electroreduction mechanism by identifying the major products of E t A Q electroreduction. Furthermore, when current was applied it was observed that for the acid emulsion the three-phase organic/aqueous/solid interface was the zone where the fastest colour change occurred (indicative of E t A Q reduction), while in the case of alkali aqueous phase the entire organic/solid boundary was active. Thus, the main reaction zones are: the S /L /L interface for the acid aqueous phase and the electrode/organic boundary for the alkali emulsion. Batch peroxide electrosynthesis experiments using a 30 ppi reticulated vitreous carbon ( R V C ) electrode at 500 A m"2, showed that by emulsion mediation more than an order of magnitude higher peroxide concentrations was obtained in comparison to the direct, surfactant modified, 2e" 0 2 reduction. For instance, at 500 A m"2 in 0.1 M N a 2 C 0 3 , emulsion mediation yielded 720 m M H2O2 after 10 hrs (current efficiency 83%) , while only a maximum of 38 m M H2O2 (current efficiency 20%) was obtained by direct O2 reduction. 8 Conclusions and Recommendations 197 Mediated experiments performed in 'flow-by' cells revealed significant differences in the behaviour of the acidic (i.e., 1 M N a 2 S 0 4 , p H 3) and alkali (i.e., 1 M N a H C 0 3 / 0.5 M N a 2 C 0 3 , p H 9.3) emulsions. The observed phenomena could be partly explained based on the different emulsion structures. For the alkali emulsion, using the 0.14 m effective length 'short' cell, the R V C 100 ppi cathode gave the highest peroxide concentration per pass, e.g., at 2000 A m"2 38 m M peroxide was obtained (current efficiency 40%), while graphite felt under the same conditions yielded 17 m M peroxide (current efficiency 20%). Factorial experiments at a constant current density of 2500 A m"2 revealed that ' low' level of liquid load (i.e. 2.7 kg m"2 s"1) and 'high' levels o f both organic / aqueous phase volume ratio (i.e. 0.96) and surfactant concentration (i.e. 1 mM), improved the figures of merit (i.e., peroxide concentration and current efficiency) both individually (as a main effect) and synergistically (interaction effect). However, for all the combinations of independent variables (i.e., liquid load, phase volume ratio and surfactant concentrations) the current efficiencies for alkali mediation at 2500 A m"2 in the 'flow-by' cells were below 40%. Increasing the 0 2 load above 0.1 kg m' s" at STP had a negative effect on the figures of merit. The above results were interpreted taking into account the O A V emulsion structure and the position of the reaction zone. Regarding the acid emulsion, in the 0.14 m effective flow-by cell, graphite felt gave better figures of merit than the R V C s (100 ppi and 30 ppi, respectively). Factorial experiments with four variables (i.e., current density, liquid load, phase volume ratio and surfactant concentration) indicated that the liquid load followed by the surfactant concentration had the strongest positive main effect on figures of merit, while the effect of the phase volume ratio was statistically insignificant. Furthermore, the 0 2 load in the range of 0.017 and 0.62 kg m"2 s"1, had no influence on the peroxide concentration. Exploiting the favourable main and interaction effects of the dependent variables, high current efficiencies (e.g., up to 84%) were possible even at 3000 A m" . The behaviour of the acid emulsion can be explained based on the proposed bicontinuous structure and the location of the reaction zone at the three-phase S / L / L interface. The experimental data obtained in the present study, suggests that the figures of merit for emulsion mediation were better for acidic than for alkali peroxide generation. B y acid emulsion mediation using the 0.5 m long cell operated at 1000 A m 2 , with 10 kg m"2 s"1 liquid load and 0.088 kg m"2 s"1 0 2 load, 280 m M H 2 0 2 was obtained after 5 hrs of 8 Conclusions and Recommendations 198 continuous catholyte recycle. However, the performance of the system deteriorated after about 2 hours of continuous operation, (e.g., the current efficiency dropped below 70%) due to membrane failure. The economic calculations (Chapter 8) for a potential in-situ electrochemical peroxide plant producing 10 t/day H2O2 and employing acid emulsion mediation showed that the lowest operating costs and thereby, the highest net economic potential (175 $ h"1) was obtained for a current density of 1600 A m"2. However, the return of investment reached a maximum of \9%per year at ca 4000 A m"2, due to the correspondingly low total installed capital cost (i.e., 5 million $ y e ar 2000 basis)- Although, for the present stage of process development the return of investment falls short of economic viability, improvements of the electrochemical cell design could increase the return of investment above 30%per y e a r , thus offering adequate economic prospects for this process (see Section 8.2.2, 'Recommendations for Future Work: Emulsion Mediation'). 8 Conclusions and Recommendations 199 8.2 Recommendations for Future Work 8.2.1 Surfactant modified O2 electroreduction to H2O2 The main goal of further research in this area should be to 'optimize' the cationic surfactant effect through a better understanding of surfactant adsorption and film formation on carbon electrodes. Furthermore, the interaction between cationic surfactant adsorption, hydrodynamics and, potential / current distribution in three-dimensional electrodes, such as graphite felt, must be looked at more closely. The following specific recommendations are made: • Investigate the Aliquat 336® film formation phenomena on carbon substrates by more elaborate surface science techniques. • Study the electrode potential dependence of Aliquat 336® adsorption by recording pertinent capacitance curves and calculating the surfactant adsorption density. • Develop an analytical method for quantitative determination of Aliquat 336® (e.g., volumetric analysis [Cross, 1994]) • Based on the quantitative analysis of the cationic surfactant, determine the surfactant adsorption density (i.e., surfactant concentration per electrode area) under different liquid and O2 flow conditions in 'flow-by' cells operated at various applied current densities of interest. • Perform electrode potential measurements at several points along the length (and possibly along the width) of the 'flow-cell ' as a function of inlet Aliquat 336® concentration, fluid dynamics and applied current density. • Investigate the spatial distribution of Aliquat 336® in the porous matrix. • Try to interpret the 'macroscopic' figures of merit for peroxide electrosynthesis using the information obtained in the previous tasks. Attempt the mathematical modelling of the phenomena under study. • Perform surfactant modified electroreduction experiments in 'flow-by' cells with complete recycle of the catholyte. Study the potential surfactant 'build-up' in the porous matrix and follow the performance of the membrane. • Depending on the outcome of the above studies, search for and test other cationic surfactants with performance superior to Aliquat 336 . 8 Conclusions and Recommendations 200 8.2.2 Emuls ion mediated electrosynthesis of H 2 0 2 Future work in this area should mainly concentrate on improving the cell design in order to eliminate the need for a cation exchange membrane and hence, increase the prospects of economic viability. Furthermore, the in-depth knowledge regarding the surfactant effect from Section 8.2.1, could be extremely useful for emulsion mediation. The following specific recommendations are made: • Using the 0.5 m effective length cell, test the performance in continuous recycle mode operated at higher liquid loads and current densities than in the present work, e.g., 20 kg m"2 s"1 and 3000 A m"2, respectively. • Explore the 'differential area' electrochemical cell design concept, to attempt the elimination of the cation exchange membrane. One idea would be to use a cylindrical cell design. The anode would be a rod (or a wire) in the centre, with plastic spacer and graphite felt (cathode) wrapped around it. The outer layer of graphite felt wi l l be in contact with a cylindrical stainless-steel current feeder. In order to reduce the contact between the anode and the organic media, the cell could be fed from two separate streams, an aqueous anolyte inlet around the bottom of the anode and the emulsion, L / L / G catholyte inlet, situated at the lower part of the outer cylinder contacting the graphite felt cathode. Efficient water cooling could be also provided all around the outer cylinder. • Investigate the possibility of using a bipolar, capillary gap cell. • Test the long-term stability of graphite felt in contact with the organic phase. • Investigate the homogeneous kinetics between E t A Q H 2 and 0 2 under conditions which are closer to the those encountered in larger-scale flow cells, i.e., E t A Q H 2 concentrations of ca 0.1 M . • Evaluate the 0 2 mass transfer capacity in the 'flow-by cell ' in relation to the electroreduction rate. • Perform emulsion-mediated experiments at superatmospheric 0 2 pressures. • Test various organic phosphates as potential solvents with water solubilities lower than T B P . References 201 R E F E R E N C E S Abbott, A.P. , Griffith, G . A . and Harper, J.C., 'Conductivity of Long Chain Quaternay Ammonium Electrolytes in Cyclohexane', J.Chem.Soc., Faraday Trans., 93, 577-582 (1997). Amatore, C , 'Basic Concepts', in 'Organic Electrochemistry', 3-rd ed., edited by Lund, H . and Baizer, M . M . , Marcel Dekker, N Y , (1991). Anderson, J .R . and Amini, B . , 'Hydrogen Peroxide Bleaching', in 'Pulp Bleaching: Principles and Practice', Dence, C .W. and Reeve, D . W . editors, T A P P I Press, Atlanta, Georgia, (1996). Afanas'ev, B .N . , Akulova, Yu .P . , and L . V . , Bykova, 'Predicted Kinetic Effects of Alkylammonium Ions on H30+ and Oxygen Electroreduction at the Dropping Mercury Electrode', Russ.J. o f Electrochem., 28, 1524-1526 (1992). Afanas'ev, B . N . , Skobochkina, Yu .P . , and L . V . , Bykova, 'Influence of Tetraalkylammonium Salts on Hydroxonium Ion Electroreduction Kinetics at the Dropping Mercury Electrode. Possibilities for Estimating the Inhibition Effect', Russ.J of Electrochem., 27, 412-418 (1991). Alcaide, F . , Brillas, E . , Cabot, P -L . , and Casado, J., Electrogeneration of Hydroperoxide Ion Using an Alkaline Fuel Cell', J.Electrochem.Soc. 145. 3444-3449 (1998). Alvarez-Gallegos, A . and Pletcher, D . , 'The Removal of Low Level Organics via Hydrogen Peroxide formed in a Reticulated Vitreous Carbon Cathode Cell.Part 2. The Removal of Phenols and Related Compounds from Aqueous Effluents' Electrochim.Acta., 44, 2483-2492 (1999). Alvarez-Gallegos, A. and Pletcher, D. , 'The Removal of Low Level Organics via Hydrogen Peroxide formed in a Reticulated Vitreous Carbon Cathode Cell.Part 1. The Electrosynthesis of Hydrogen Peroxide in Aqueous Acidic Solutions', Electrochim.Acta, 44, 853-861 (1998). Babaei, A., Connor, P .A . and McQuillan, J .A., 'UV-Visible Spectroelectrochemistry of Reduction Products of Anthraquinone in Dimethylformamide Solutions', J.Chem.Ed. 74, 1200-1203 (1997). Baizer, M . M . , 'Discovery, Development and Commercialisation of the Electrochemical Adiponitrile Process: Part 1', Chemistry and Industry, 7, 435-439 (1979). Ban, A. , Schafer, A . , and Wendt, H . , Fundamentals of Electrosorption on Activated Carbon for Wastewater Treatment of Industrial Effluents', J.Appl.Electrochem., 28, 227-236 (1998). Bard, A . J . and Faulkner, L . R . , Electrochemical Methods: Fundamentals and Applications', John Wiley & Sons, New York, (1980). Blake, T.D., 'Wetting', in 'Surfactants', edited by Tadros, Th., F., Academic Press, London, (1984). References 202 Blankespoor, R . L . , Schutt, D . L . , Tubergen, M . B . and De Jong, R . L . , 'The Influence of Ion Pairing on the Electroreductive Cleavage of Substituted 9,10-Anthraquinones in DMF Solutions', J.Org.Chem., 52, 2059-2064 (1987). Bockris, J . O ' M . and Khan, S.U., 'Surface Electrochemistry. A Molecular Level Approach', Plenum Press, New York , (1993). Brown, E . R . and Sandifer, J.R., 'Cyclic Voltammetry, AC Polarography and Related Techniques', in 'Physical Methods of Chemistry', 2-nd ed., vol. II: Electrochemical Methods, edited by Rossiter, B . W . and Hamilton, J.F., John Wiley & Sons, N Y (1986). Burgess, L , Jeffrey, C . A . , Cai, X . , Szymanski, G. , Galus, Z . , and Lipkowski, J., 'Direct Visualization of the Potential-Controlled Transformation of Hemimicellar Aggregates of Dodecyl Sulfate into a Condensed Monolayer at the Au(lll) Electrode Surface', Langmuir, 15, 2607-2616 (1999). Chaenko, N.V., Kornienko, G . V . , Vasil'eva, I.S. and Kornienko, V . L . , 'Influence of Surfactants on Characteristics of Hydrogen Peroxide Electrosynthesis from Oxygen on Hydrophobized Carbon Black Electrode', Russ. J .Appl. Chem., 69, 706-708 (1996). Chambers, J .Q. , in 'The Chemistry of the Quinonoid Compounds', edited by Patai, S and Rappoport, Z . , vol 2, 719-757, John Wiley & Sons, N Y , (1988). Chemical Engineering, 107(5), 164 (2000). Chevalet, J . , Rouelle, F., Gierst, L . and Lambert, J.P., Electrogeneration and some Properties of the Superoxide Ion in Aqueous Solutions', J.Electroanal.Chem., 39, 201-216 (1972). Cross, J . , 'Volumetric Analysis of Cationic Surfactants', in 'Cationic Surfactants: Analytical and Biological Evaluation', edited by Cross, J. and Singer, E .J . , Marcel Dekker Inc., N Y , (1994). Cypress Systems, Omni-90 Potentiostat, Instruction Manual, K S (1991). Danly, D. and Campbell, C.R., 'Experience in the Scale-up of the Monsanto Adiponitrile Process', in 'Technique of Electroorganic Synthesis. Part III: Scale-up and Engineering Aspects', edited by Weinberg, N . L . and Tilak, B . V . , John Wiley & Sons, N Y , (1982). Davison, J.B., Kacsir, J . M . , Peerce-Landers, P.J. and Jasinski, R., 'A Voltammetric Investigation of Oxygen Reduction in a Trickle Bed Cell using Graphite Chip and RVC Cathodes', J.Electrochem.Soc. 130. 1497-1501 (1983). Degrand, C , Influence of the pH on the Catalytic Reduction of Oxygen to Hydrogen Peroxide at Carbon Electrodes Modified by an Adsorbed Anthraquinone Polymer', J.Electroanal.Chem., 169. 259-268 (1984). References 203 Dong, D.F., Mumby, T .A . , Jackson, J.R. and Rogers, D.J . , 'Co-Generation of Ammonium Persulfate Anodically and Alkaline Hydrogen Peroxide Caihodically with Cathode Products Ratio Control', U S Patent, 5.643.437. July 1, (1997). Dow Inc., 'On-site Alkaline Peroxide Generation System', advertising literature D o w Chemical Co., (1999). Drackett, T.S., 'Electrolytic Production of Hydrogen Peroxide using Bipolar Membranes', U S Patent 5,358,609 (1994) [To: Chemetics International Ltd.] Dworak, R. and Wendt, H . , 'Investigation of the Mass Transfer through Turbulent Boundary Layers in Liquid-Liquid Emulsions', Ber.Bunsen-Ges. Phys.Chem., 81_, 728-732 (1977). Dworak, R., Fees, H . and Wendt, H . , 'How to Process Electrolysis with Two Phase Electrolytes', A I C h E Symp.Ser., No: 185, 75, 38-43 (1979). El Mouahid, O. , Coutanceau, C , Belgsir, E . M . , Crouigneau, P., Leger, J . M . , Lamy, C , Electrocatalytic Reduction of Dioxygen at Macrocycle Conducting Ploymer Electrodes in Acid Media ' , J. Electroanal. Chem., 426, 117-123 (1997). E R G Inc., 'Reticulated Vitreous Carbon', Technical Literature, (1996). Evans, D.F. and Wennerstrom, H . , 'The Colloidal Domain: Where Physics, Chemistry, Biology and Technology Meet', V C H Pub., 1994. Feess, H . and Wendt, H . , 'Performance of Electrolysis with Two-Phase Electrolyte. The Trickle Electrode for the Electrochemical Conversion of Substances with Low Solubility in Aqueous Electrolytes with Special Respect to Electrochemical Extraction Processes', J.Chem.Tech.Biotechnol., 30, 297-312 (1980). Feess, H . and Wendt, H . , 'Mass and Charge Transfer across the Organic/Aqueous Phase-Boundary in Electrolysis processes using Two-Phase-Emulsion Mixed Electrolytes', Ber.Bunsen-Ges. Phys.Chem., 85, 914-919 (1981). Fenton, J . M . and Alkire, R .C . , 'Mass Transfer in Flow-Through Porous Electrodes with Two-Phase Liquid-Liquid Flow', J.Electrochem.Soc, 135, 2200-2209 (1988). Foller, P.C. and Bombard, R , 'Processes for the Production of Mixtures of Caustic Soda and Hydrogen Peroxide via the Reduction of Oxygen', JAppl.Electrochem., 25, 613 (1995). Fuerstenau, D.W. and Herrera-Urbina, R., 'Adsorption of Cationic Surfactants and the Flotation of Minerals', in 'Cationic Surfactants. Physical Chemistry', edited by Rubingh, D . N . and Holland, P . M . , 407 - 447, Marcel Dekker, New York, (1991). G & P Engineering, Physprop®, http://gpengineeringsoft.com/. (1999). References 204 Gianetto, A . and Specchia, V . , 'Trickle-Bed Reactors: State of Art and Perspectives', Chem.Eng.Sci., 47, 3197-3213 (1992). Gileadi, E . , Kirowa-Eisner, E . , and Penciner, J., 'Interfacial Electrochemistry. An Experimental Approach', Addison-Wesley, Reading, M A , (1975). Gueshi, T., Tokuda, K . and Matsuda, H . , 'Voltammetry at Partially Covered Electrodes. Part I. Chronopotentiometry and Chronoamperometry at Model Electrodes', J.Electroanal.Chem., 89, 247-260 (1978). Gueshi, T., Tokuda, K . and Matsuda, H . , 'Voltammetry at Partially Covered Electrodes. Part II. Linear Potential Sweep and Cyclic Voltammetry', J.Electroanal.Chem., 101, 29-38 (1979). Hermann, C , Dewes, I. and Schumpe, A . , 'The Estimation of Gas Solubilities in Salt Solutions', Chem. Eng. Sci., 50, 1673 (1995). Hoare, J .P. 'Oxygen' in 'Standard Potentials in Aqueous Solution', Bard, A . J . , Parsons, R. and Jordan, J., editors, 56 - 58, Marcel Dekker, N Y , (1985). Hodgson, I. and Oloman, C , 'Pressure Gradient, Liquid Hold-Up and Mass Transfer in a Graphite Bed with Co-Current Upward Gas-Liquid Flow', Chem.Eng.Sci., 54^ 5777-5786, (1999). Hossain, M . S . , Tryk, D . and Yeager, E . , 'The Electrochemistry of Graphite and Modified Graphite Surfaces: The Reduction of O2', Electrochim. Acta, 34, 1733-1737 (1989). Hua, Y . J . and Qi , Z . C . , 'Cutting Back on Bleach Costs the Chinese Way', Pulp and Paper International, 10, 40-41 (1997). Huissoud, A . and Tissot, P., Electrochemical Reduction of 2-Ethyl-9,10-Anthraquinone (EAQ) and Mediated Formation of Hydrogen Peroxide in a Two-Phase Medium. Part I: Electrochemical Behaviour of EAQ on a Vitreous Carbon Rotating Disc electrode (RDE) in the Two-Phase Medium', J.Appl.Electrochem., 29, 11-17(1999). Huissoud, A . and Tissot, P., 'Electrochemical Reduction of 2-Ethyl-9,10-Anthraquinone (EAQ) and Mediated Formation of Hydrogen Peroxide in a Two-Phase Medium. Part II: Production of Alkaline Hydrogen Peroxide by the Intermediate Electroreduction of EAQ in a Flow-by Porous Electrode in Two-Phase Liquid-Liquid Flow', J.Appl.Electrochem., 29, 17-25 (1999). Huissoud, A . and Tissot, P., 'Electrochemical Reduction of 2-eihyl-9,10-anthraquinone on Reticulated Vitreous Carbon and Mediated formation of Hydrogen Peroxide', J.Appl.Electrochem., 28, 653-657 (1998). Hunter, R . J . , 'Introduction to Modern Colloid Science', Oxford University Press, Oxford, U K , (1993). References 205 Ibl, N . and Vogt, H . in 'Comprehensive Treatise on Electrochemistry', vol. 2, Bockris, J . O ' M Conway, B E . and White, R . E . , editors, p. 169-208, Plenum Press, New York , (1981). Ioanides, G. , Network of Centres of Excellence: Mechanical Pulps, Annual General meeting, Vancouver, May 29 - 31,(1999). Ingram, B .T . and Ottewill, R . H . , 'Adsorption of Cationic Surfactants at Interfaces' in 'Cationic Surfactants. Physical Chemistry', edited by Rubingh, D . N . and Holland, P . M . , 87 - 126, Marcel Dekker, New York, (1991). Janson, R . E . W . , 'Economic Driving Force in Electro-Organic Synthesis', in 'Electrochemical Cell design', edited by White, R . E . , Plenum Press, N Y , (1984). Jass 2.1, Factorial Experimental Design Software, Joiner Associates, Madison, WI , (1987). Jtbaker.com (http://www.jtha.ker. com/msds/t4706.html) Kalu, E . and Oloman, C , 'Simultaneous Electrosynthesis of Alkaline Hydrogen Peroxide and Sodium Chlorate', J. Appl . Electrochem., 20, 932-940 (1990). Keita, B. and Nadjo, L . , 'Catalytic Synthesis of Hydrogen Peroxide: An Attractive Electrochemical and Photoelectrochemical Route to the Reduction of Oxygen', J. Electroanal. Chem., 145, 431-437 (1983). Keita, B. and Nadjo, L . , 'Catalytic Synthesis of Hydrogen Peroxide: An Attractive Electrochemical and Photoelectrochemical Route to the Reduction of Oxygen', J.Electroanal.Chem., 145, 431-437 (1983). Kelsall, G .H. , and Thompson, I., 'Redox Chemistry ofH2S Oxidation by the British gas Stretford Process. Part III: Electrochemical behaviour of Anthraquinone 2,7 Disulphonate in Alkaline Electrolytes', J. Appl . Electrochem., 23, 296-307 (1993). Kinoshita, K . , 'Electrochemical Oxygen Technology', John Wiley & Sons, N Y (1992). Kirk-Othmer Concise Encyclopedia of Chemical Technology, M . Grayson Exec. Editor, p. 627, John Wiley & Sons, New York, (1985). Kislenko, V . N . and Berlin, A . A . , 'Kinetics and Mechanism of the Oxidation of Organic Compounds with Hydrogen Peroxide', Russ.Chem.Rev., 60, 470-488 (1991). Knarr, R . F . , Velasco, M . , Lynn, S. and Tobias, C.W. , 'The Electrochemical reduction of 2-EthylAnthraquinone', J.Electrochem.Soc, 139, 948-954 (1992). Kolthoff, L M . and. Lingane, J.J, 'Polarography', 2-nd ed., vol. 1, 184 - 188 and vol.2, 552 -554, Interscience Publishers, N Y , (1952). References 206 Koshitsuka, T., 'New Application of Hydrogen Peroxide in Chemical Pulp Bleaching', Japan T A P P I J., 52, 623-630 (1998). Kraft, F. in 'Pulp and Paper Manufacture', 2-nd ed., vol. 1, McDonald, R . G . and Frankin, J .N. editors, 723 - 724, McGraw-Hi l l , N Y , (1969). Lange's Handbook of Chemistry, 14-th ed., edited by Dean, J.A., McGraw-Hi l l , N Y , (1992). La Roche, R.D., Stadtherr, M . A . and Alkire, R . C . , 'Electrolytic Process Optimization by Simultaneous-Modular Flow sheeting', J.Appl.Electrochem., 24, 1206-1212 (1994). Letord-Quemere, M . M . , Coeuret, F. and Legrand, J., 'Forced Flow of a Two-Phase Electrolyte through a Porous Electrode of Expanded Metal: Mass Transfer at the Electrode - Application to the Reduction of Nitrobenzene', Electrochim.Acta., 33, 881-890 (1988). Lipkowski, J . , 'Ion and Electron Transfer Across Monolayers of Organic Surfactants' in 'Modern Aspects of Electrochemistry', No . 23, Conway, B . E . , Bockris, J. O ' M . and White, R E . editors, Plenum Press, New York, (1992). Mackay, R.A. and Texter, J., (editors), 'Electrochemistry in Colloids and Dispersions ' ,VCH Publishers, N Y , (1992). Manne, S. and Gaub, H E . , 'Molecular Organization of Surfactants at Solid-Liquid Interfaces', Science, 270, (1 December), 1480-1484, (1995). Manufacturing.net (http://www.manufacturing.net/magazine/purchasing/archives/2000/) Maron, S . H . and Prutton, C.F. , 'Principles of Physical Chemistry', 4-th ed., MacMil lan Co., London, (1971). Masliy, A.I., and Poddubny, N .P . , 'Influence of Solid Phase Conductivity on Spatial Localization of Electrochemical Processes in Flow-Through Porous Electrodes. Part I: Electrodes with Uniform Conducting Matrix', J.Appl.Electrochem., 27, 1036-1044 (1997). Matheson, I.B.C. and King, A . D . Jr., 'Solubilities of Gases inMicellar Solutions', J. Colloid Interface Sci., 66, 464-469 (1978). Mathur, I. and Dawe, R., 'Using on-site-produced Alkaline Peroxide for Pulp Bleaching', Tappi J., 82, 157-164 (1999). Merck Index, 11-th ed., S. Budavari editor, Merck & Co. Inc., Rahway, N J , (1989). Montgomery, D.C. , 'Design and Analysis of Experiments', 4-th ed., John Wiley & Sons, N Y (1997). References 207 Mousty, C. and Mousset, G . , 'Electrochemistry in the Presence of Cationic Surfactants. Influence of the Adsorpotion Mode on the Reduction of Acetophenone', New.J .Chem, 18, 1063-1070 (1992). Murphy, T.D., 'Design and Analysis of Industrial Experiments', Chemical Engineering, June 6, 168-182(1977). Myers, D., 'Surfactant Science and Technology', V C H Publishers, N e w York , (1988). Nagaoka, T. and Okazaki, S., 'Electron-Transfer Kinetics of Quinones at Solid Electrodes in Aprotic Solvents at Low Temperatures', J.Phys.Chem., 89, 2340-2344 (1985). Nekrasov, L . N . and Vykhodtseva, L . N . , 'Tetraalkylammonium Cation Effect on the Kinetics of Oxygen Electroreduction in Dimethyl Sulfoxide Solutions with Binary and Mixed Supporting Electrolyte', Russ.J. o f Electrochem., 3J_, 1139-1148 (1995). Newman, J.S., 'Electrochemical Systems', 2-nd ed., Prentice-Hall, Englewood Cliffs, N . J . , (1991). Nikitas, P., 'Aggregation phenomena occurring on an electrode surface in the presence or absence of aggregation in the bulk electrolyte solution', J. Electroanal. Chem., 425, 97 (1997). Nozaki, M . , 'On-site Alkaline Peroxide Generation System', Japan T A P P I J., 52, 616-622 (1998). Oloman, C , 'Electrochemical Processing for the Pulp and Paper Industry', The Electrochemical Consultancy, Romsey, U K , (1996). Oloman, C , 'Electro-oxidation of Benzene in a Fixed Bed Reactor', J.Appl.Electrochem., 10, 553-565 (1980). Oloman, C , 'Trickle Bed Electrochemical Reactors', J.Electrochem.Soc, 126, 1885-1892 (1979). Oloman, C , Matte, M . and Lum, C , Electronic Conductivity of Graphite Fiber Fixed-Bed Electrodes', J.Electrochem.Soc, 138, 2330-2333 (1991). Oloman, C. and Watkinson, A . P . , 'Hydrogen Peroxide Production in Trickle-Bed Electrochemical Reactors', J.Appl.Electrochem., 9, 117-123 (1979). Paren, A . and Tsujino, T., 'GreenOx Bleaching Process', Japan T A P P I I , 52, 630-635 (1998). Pekmez, K . , Can, M . and Yildiz , A . , 'Electrochemistry in Aprotic Solvents containing Anhydrous Perchloric Acid: Electroreduction behavior of Quinones', Electrochim.Acta., 38, 607-611 (1993). References 208 Perry's Chemical Engineers' Handbook, 6-th ed., edited by Green, D . W . and Maloney, J.O., McGraw-Hi l l Inc., N Y , (1984). Peters, M . S . , and Timmerhaus, K . D . , 'Plant Design and Economics for Chemical Engineers', 4-th edition, McGraw-Hi l l , N Y (1991). Pletcher, D. and Thompson, H . , 'A Microelectrode Study of the Influence of Electrolyte on the Reduction of Quinones in Aprotic Solvents', J.Chem.Soc. Faraday Trans., 94, 3445-3450 (1998). Ponce de Leon, C . and Pletcher, D . , 'Removal of Formaldehyde from Aqueous Solutions via Oxygen Reduction using a Reticulated Vitreous Carbon Cathode Cell', J.Appl.Electrochem., 25, 307-314 (1995). Posdorfer, J . , Olbrich-Stock, M . and Schindler, R . N . , 'Spectroelectrochemical Determination of the Heterogeneous and Homogeneous Rate Constants in the 9,10-Anthraquinone-System by Diffusion Layer Imaging', Z.Phys.Chem., 171, 33-52 (1991). Prentice, G . , 'Electrochemical Engineering Principles', Prentice Hall , Engelwood Cliffs, N J , (1991). Probstein, R .F . , 'Physicochemical Hydrodynamics: A n Introduction', Butterworths, Stoneham, M A , U S A , (1989). Radyushkina, K . A . , Zonina, E .O . and Tarasevich, M . R . , 'Oxygen Electroreduction at Carbon Materials in Nonaqueous Media', Russ.J.of Electrochem., 28, 736-744 (1992). Reid, R .C . , Prausnitz, J . M . , and Sherwood, T .K. , 'The Properties of Gases and Liquids', 3-rd edition, McGraw-Hi l l , N Y , (1977). Revenga, J. , Rodriquez, F. and Tijero, J., 'Study of the Redox Behavior of Anthraquinone in Aqueous Medium', J.Electrochem.Soc, 141, 330-333 (1994). Rieger, P.H., 'Electrochemistry', Prentice-Hall, Englewood Cliffs, N . J . , (1987). Roberts, J .L . Jr., Sugimoto, H . , Barrette, W . C . and Sawyer, D.T. , 'One-Electron Reduction of Anthraquinone by Hydroxide Ion in Aprotic Media', J .Am.Chem.Soc, 107, 4556-4557 (1985). Rudie, A.W., McDonough, T.J., Rauh, F., Klein, R.J . , Parker, J .L. and Heimburger, S.A., 'An Evaluation of Sodium Carbonate as a Replacement for Sodium Hydroxide in Hydrogen Peroxide Bleaching o / C T M P ' , J.Pulp and Paper Science, 19, J47-J52, (1993). Rusling, J .F . , 'Controlling Electrochemical Catalysis with Surfactant Microstructures', Acc. Chem. Res., 24, 75 (1991). References 209 Rusling, J.F., ' Electrochemistry in Micelles, Microemulsions and Related Microheterogeneous Fluids' in 'Electroanalytical Chemistry. A Series of Advances', vol. 18, Bard, A . J . , editor, Marcel Dekker, New York, (1994 I). Rusling, J.F., Electrochemistry and Electrochemical Catalysis in Microemulsions' in 'Modern Aspects of Electrochemistry', No . 26, Conway, B . E , Bockris, J. O ' M . , and White, R . E . , editors, Plenum Press, N Y , (1994 II). Sawyer, D.T. , 'Oxygen Chemistry', Oxford University Press, N Y , (1991). Schmal, D., van Erkel, J. and van Duin, P.J., 'Mass Transfer at Carbon Fibre Electrodes', J.Appl.Electrochem., 16, 422-430 (1986). seweb2.phillips66.com (http:// seweb2.phillips66.com/hes/msds.nsf) Shaw, D.J., 'Introduction to Colloid and Surface Chemistry', 4-th ed., Butterworth-Heinemann Ltd. , Oxford, U K (1992). Sigma Chemical Company Catalogue, 1502, (1991). Sigrist, L . , Dossenbach, O. and Ibl, N . , 'On the Conductivity and Void Fraction of Gas Dispersions in Electrolyte Solutions', J.Appl.Electrochem., 10, 223-228 (1980). Simonet, J . , 'Electrogenerated Reagents', in 'Organic Electrochemistry', 3-rd ed., edited by Lund, H . and Baizer, M . M . , Marcel Dekker, N Y , (1991). Socha, A . , Chrzescijanska, E . , Pietrzak, T . M . and Kwiatkowski, W. , 'Influence of the Hydroxyl Ion on the Electroactivity of the Quinone in the Disulfonated Derivatives of Anthraquinone', Polish J.Chem., 67, 505-516 (1993). Storck, A . , Latifi, M . A . , Barthole, G. , Laurent, A . , and J.C. Charpentier, 'Electrochemical Study of Liquid-Solid Mass Transfer in Packed Bed Electrodes with Upward and Downward Co-Current Gas-Liquid Flow', J.Appl.Electrochem., 16, 947-963 (1986). Sudoh, ML, Minamoto, K . , Makion, T., and Hakamada, H . , 'Production of Hydrogen Peroxide in Acidic Solutions by Electrodialysis of Peroxide Ions Generated by Electroreduction of Oxygen in Alkaline Solutions', J. Chem.Eng. of Japan, 24, 465-471 (1991). Swiatkowski, A . , Pakula, M . and Biniak, S., 'Cyclic Voltammetric Studies of Chemically and Electrochemically generated Oxygen Species on Activated Carbons', Electrochim. Acta, 42, 1441 (1997). Takahashi, K . and Alkire, R., 'Mass Transfer in Gas-Sparged Porous Electrodes', Chem.Eng.Commun., 38,209-227 (1985). References 210 Takahashi , K . M . and Alkire, R .C . , 'Mass Transfer from Dispersed Bubbles to Electrolyte Solution in Packed Beds', A I C h E J., 34, 1504-1510 (1988). Tatapudi , P. and Fenton, J. in 'Advances in Electrochemical Science and Engineering', vol. 4, edited by H . Gerischer and C.W. Tobias, V C H , Weinheim, Germany (1995). Taylor , R . J . and. Humffray, A . A , Electrochemical Studies on Glassy Carbon Electrodes. II. Oxygen Reduction in Solutions of High pH (pH > 10)', J. Electroanal. Chem., 64, 63-84 (1975). Tissot, P. and Huissoud, A . , 'Proton Effects in the Electrochemical behaviour of 2-ethyl-9,10-anthraquinone in the medium Dimethoxyethane-Tetrabutylammonium Hydroxide with and without Oxygen', Electrochim.Acta., 41_, 2451-2456 (1996). U l r i ch , G . D . , ' A Guide to Chemical Engineering Process Design and Economies', John Wiley & Sons, N Y , (1984). Vago, E . R . , de Weldige, K . , Rohwerder, M . and Stratmann, M . , Electroreduction of Oxygen on Octadecylmercaptan self-assembled Monolayers', Fresenius J.Anal.Chem., 353. 316-319 (1995). Vasudevan, D . and Wendt, H . , 'Electroreduction of Oxygen in Aprotic Media', J.Electroanal.Chem., 192, 69-74 (1995). Vetter, K . J . , Electrochemical Kinetics: Theoretical and Experimental Aspects', Academic Press, New York, (1967). Walas, S . M . , 'Rules of Thumb: Selecting and Designing Equipment', Chemical Engineering, March 16, 75-81 (1987). Walsh , F . C . , ' A First Course in Electrochemical Engineering', The Electrochemical Consultancy, Romsey, U K , (1993). W a n , J . , 'NeutralPeroxide Bleaching', Industry Workshop on Bleaching and Yellowing Inhibition '99, Queen's University, Kingston, O N , Apr i l 9-10, (1999). Webb , S.P. and Mclntyre, J., 'Generation of Hydrogen Peroxide in a Shorted Fuel Cel l ' , 10 t h International Forum on Electrolysis in the Chemical Industry, November 10-14, Clearwater, F L (1996). Wilson , T. , Hydrogen Peroxide Electrosynthesis: The Search for an Efficient Catalyst', Internal Report, Dept. of Chemical Engineering, University of British Columbia and National Research Council, Ottawa, (1996). Whiteside, J . , 'Greener Chemicals for a Whiter Pulp', Pulp and Paper International, 10, 48-50 (1998). References 211 Wrigh ton , M . S . , 'Surface Functionalization of Electrodes with Molecular Reagents', Science, 231. 32-37(1986). Yano , T., Tryk, D . A . , Hashimoto, K . and Fujishima, A . , 'Electrochemical Behavior of Highly Conductive Boron-Doped Diamond Electrodes for Oxgen Reduction in Alkaline Solution', J. Electrochem. Soc , 145, 1870-1876(1998). Y u a n , J . and Yuan, Z . , 'Alkaline Peroxide Generator', Chinese Patent, U15341A. January 24, (1996). Zagal , J . H . , Aguirre, M . J . , Paez, M . A . , 'O2 Reduction Kinetics on a Graphite Electrode Modified with Adsorbed Vitamin Bt 2', J. Electroanal. Chem., 437, 45-52 (1997). Nomenclature 212 Nomenclature a specific electrode surface area (m 2 m"3) acs area occupied per Aliquat 336 molecule (m 2 molecule"1) Ac superficial area per electrochemical cell (m 2) b Tafel slope (V) B' cost constant (Chapter 7) cu Ct concentration of species z ( M or mol m"3) C installed capital cost ($) (in Chapter 7 and Appendix L) Cf number of centerpoints in the factorial analysis (Appendix F) Cp heat capacity (kJ kg" 1 K" 1 ) C ' cost constant (Chapter 7) CE current efficiency (%) df mean fibre (e.g., graphite) diameter (u,m) D diffusion coefficient (m 2 s"1) Dd decanter diameter (Appendix L) Dsep diameter of the G / L separator (Appendix L) E electrode potential ( V vs. Ag /AgCl ) E ° standard potential ( V vs. S H E or V vs. Ag /AgCl ) Ej/2 polarographic half-wave potential ( V vs. SCE) Em electrode potential of the solid matrix in 3-D electrochemical reactors (V) EPzc potential of zero charge Es electrode potential of the solution phase in 3-D electrochemical reactors (V) / volume fraction of dispersed or organic phase in the emulsion F Faraday's constant (96,485.3 C equiv"1) Fv volumetric flow rate (m 3 s"1) Fm mass flow rate (kg s"1) G gas load (kg m"2 s"1) Gc cost constant ($ m" ) IIG, hj gas and ion parameter, respectively, for 0 2 solubility calculations /' current density (A m"2) / current (juA or A ) k second order rate constant (1 mol"1 s"1) ks standard heterogeneous rate constant (m s"1) kc pseudo-first order homogeneous rate constant (s"1) K aqueous / organic H 2 0 2 partition coefficient Ko single-phase mass transfer coefficient (m s"1) KD disproportionation equilibrium constant Km overall (e.g. 2-phase) mass transfer coefficient (m s"1) Kmc mass transfer capacity (s"1) KSj Setchenov's constant L liquid load (kg m" s" ) LD decanter length (m) (Appendix L) Nomenclature 213 Lhc hydrocarbon chain length in the surfactant, A3 3 6, molecule m mass (kg) (or scale exponent in eqn. [7.4]) nif total degrees of freedom in factorial analysis (Appendix F) n total number of electrons involved in the electrochemical reaction mechanism na number of electrons transferred in the rate determining step rii number of ions /', in the electrolyte stoichiometry rif total number of factors in the statistical experiments, Appendix F) N number of moles NA Avogadro's number (6.022x10 2 3 mol"1) Nf total number of factorial runs (Appendix F) Nhc number of hydrocarbon chains in the Aliquat 336 structure Nc total number of electrochemical cells Ns surfactant number (eqn. [3.14]) P pressure (kPa) Pr peroxide production rate (kmol s"1) Pw power consumption (kW) Q heat (kJ) R universal gas constant (8.314 J mol"1 K" 1 ) RE rectifier efficiency s2 pooled variance for the factorial experiments S cationic surfactant (A3 3 6) concentration in factorial experiments (mM) Sc sorptive capacity of the adsorption column packing (Appendix L) t thickness of the porous electrode; final (compressed) to thickness of the porous electrode; initial (uncompressed) tb adsorption column 'breakthrough' time (days) ts Student's statistics T temperature (K) Tr adsorption column throughput ratio in Appendix L uav average velocity of the emulsion stream (m s"1) (Appendix L) ut terminal settling velocity (m s"1) (Appendix L ) U overall heat transfer coefficient (W m"2 K" 1 ) (Appendix L) v scan rate in cyclic voltammetry ( V s"1) Vb volume of the adsorption bed (m 3) vgas velocity of 0 2 gas in the G / L separator (m s"1) (Appendix L) V volume (m 3) (or electric potential 'driving force' in Fig. 3.10) Vc voltage per cell (V) Vhc volume of the hydrocarbon chain in the surfactant, A3 3 6, molecule (m 3) Voc open circuit voltage (V) Vt total voltage (V) Xi coded variable in factorial experimental analysis (Appendix F) Xi factor (i.e., independent variable) in the statistical design (Appendix F) Xm>i main effect of factor Xt y factorial response (i.e., dependent variable) Nomenclature 214 z total number of electrons transferred in the electrochemical process z, charge of ion i Greek Symbols: a charge transfer coefficient B symmetry factor Bi liquid hold-up in the porous electrode s 3-D electrode porosity; final (compressed) So 3-D electrode porosity; initial (uncompressed) £b adsorption bed porosity (Appendix L ) Eeff effective matrix (bed) porosity sm generic solid matrix porosity sp pump efficiency (Appendix h) sr dielectric constant of water (78.5) (in Appendix Q sv permitivity of vacuum (8.85xl0" 1 2 C 2 J"1 m"1) (in Appendix C) 05 surface potential (V) y surface tension ( N m"1) K ionic conductivity (S m"1) KM electronic conductivity of the 3-D matrix (S m"1) rj overvoltage (V) A partition ratio of the adsorption column (Appendix L) A M molar ionic conductivity (S m"1 M" 1 ) ju dynamic viscosity (Pa s"1) v kinematic viscosity (m 2 s"1) as surface charge density (C m"2) rs surfactant adsorption density (mol m"2) p density (kg m"3) 6 three-phase contact angle in Figs. 3.5 and 3.6 0S surface coverage by the Aliquat 336 admicelle r residence time (s) rmax maximum electroactive bed thickness (m) co regression coefficient in factorial analysis (Appendix F) v|/ Nicholson's kinetic parameter Subscripts: aqu bulk eff aqueous phase bulk phase effective Nomenclature 215 em emulsion dissoc dissociated film the Aliquat 336 surface film E,ML maintenance and labour of the electrochemical plant IEC installed electrochemical cells I,E installed electrochemical reactor unit L mass transfer limited org organic phase p,p/2 cyclic voltammetry peak and half-peak, respectively PS power supply s surface phase (i.e. inner limit of the diffuse double layer) t total Abbreviat ions: A336 Aliquat 336® CIC total installed capital cost of the electrochemical plant COP total operating costs of the electrochemical plant DEB diethylbenzene DSA dimensionally stable anode GEP gross economic potential GF graphite felt EtAQ 2-Ethyl-9,10-Anthraquinone NEP net economic potential NHE normal hydrogen electrode ROI return of investment RVC reticulated vitreous carbon SCE standard calomel electrode SHE standard hydrogen electrode SDS sodium dodecyl sulfate STP standard temperature and pressure TBAP tetrabutylammonium perchlorate TBP tributyl phosphate Appendix A 216 APPENDIX A: Calculation of the Oxygen Solubility in Various Electrolyte Solutions The bulk O2 concentrations Cb,02, in the employed electrolyte solutions (e.g. 0.1 M Na2C03, 0.1 M H2SO4, 1 M Na2S04), were obtained from an improved version of the Setchenov equation [Hermann et al., 1995]: log c ~ ^H^S,jCsalt,j> KS]=Y,ihi+hoW [A-2] i where Cw,02 is the O2 concentration in pure water at 295 K (M), Ksj the Setchenov constant for electrolyte j (1 mol"1) [for a single electrolyte solution j =1], csaitjconcentration of electrolyte j (M), hi parameter for ion i, (e.g. H ^ : 0, N a + : 0.1079, S0 4 2 " : 0.1164, C 0 3 2 " : 0.1558, H C 0 3 " : 0.1019), ha parameter for gas (O2: 0), number of ions /' in the electrolyte. Table A - l Estimated O2 solubility in pertinent electrolyte solutions. Electrolyte Pressure Temperature O2 concentration (kPa) (K) (mM) 0.1 M N a 2 C 0 3 100 295 U 7 O . I M H 2 S O 4 100 295 1.24 M i x t u r e o f 0 . 5 M N a 2 C O 3 a n d 0 . 5 M N a H C O 3 350 295 2.14 l M N a 2 S 0 4 425 295 2.36 Appendix B 111 APPENDIX B: Calculation of Kinetic Parameters from Cyclic Voltammetry Data for Irreversible Charge Transfer under Mixed (Activation - Diffusion) Control For an irreversible charge transfer process under mixed control, e.g. 0 2 reduction in aqueous electrolytes (Figs. 5 .1a and b), the scan rate dependence of the cyclic voltammetry peak potential and current density are expressed by [Gileadi et al, 1975]: Ep=E°-b 1 fb} 0 . 5 2 - - l o g l o g * . 2 \ D ) s - - l o g v , [ B - l ] and ip = 3 . 0 1 x l 0 5 « | f 2.303RT^ 1/2 V bF j DmCbvm, [B-2] with b = r2.303RT^ V M J7 J [J3-3] where Ep peak potential (V), E° standard potential for O2 reduction [Ffoare, 1985], b Tafel slope (V), D O2 diffusion coefficient (1.9xl0" 9 m 2 s"1 at 295 K in dilute aqueous electrolytes [Kinoshita, 1992]), ks standard heterogeneous rate constant (m s"1), v scan rate ( V s"1), ip peak current density (A m"2), n total number of electrons transferred, C j bulk 0 2 concentration (mol m"3) (see Appendix A), a transfer coefficient, na nr. of electrons transferred in one act of the rate determining step, R universal gas constant (J mol"1 K" 1 ) , T temperature (K). Cyclic voltammograms such as those presented in Figs. 5.1 a and b, were recorded for nine different scan rates in the range of 0.05 to 0.5 V s"1. The Ep and ip values as a function of scan rate were obtained. From the slope of eqn. [ B - l ] (i.e. Ep vs. logv), the Tafel parameter b was determined. Next, employing eqn. [B-2] the total number of electrons transferred n, was calculated, to check the overall 2e" transfer mechanism of O2 reduction. Furthermore, using the intercept of eqn. [ B - l ] , the corresponding standard heterogeneous rate constant ks was calculated. The above parameters of O2 reduction in both 0.1 M Na2C03 and 0.1 M Ff 2S04 are given in Table 5.1. Appendix C 218 APPENDIX C : Theoretical Estimation of the Surface pH in the presence of Aliquat 336 Adsorption It was proposed (Section 5.1.2) that an increase of the surface p H induced by the cationic surfactant, might be the cause for the enhanced 0 2 reduction to H2O2, especially in acidic media. This is hinted by the fact that the peak potential Ep, for O2 reduction in 0.1 M H2SO4 in the presence of 1.7xl0"2 M Aliquat 336 is almost the same as Ep in 0.1 M Na2CC>3 in the absence of surfactant (compare Figs. 5.3 and 5.1 a). Therefore, the issue of surface p H needs to be addressed. The p H inside the electrical double layer pHs, can be calculated from the following equation [Rieger, 1987]: pHt=PHb+(&)loge [C- l ] where pHb is the bulk pH. For tps, referred to as the surface potential, one can consider the potential at the inner limit of the diffuse double layer, which is accessible numerically based on the Gouy-Chapman model [Rieger, 1987]; [Newman, 1991]. For a flat surface, <f>s is related to the charge density at the inner limit of the diffuse layer by [Newman, 1991]: cxs=+^2RTerevJ]Cub where cr s, charge density (C m"2), sr is the solvent dielectric constant (78.5), sv is the permitivity of vacuum (8.85xl0~ 1 2 C 2 J"1 m"1) and C,,i is the bulk concentration of the ionic species / (mol m"3). To solve the non-linear eqn. [C-2] for </>s, the charge density at the inner limit of the diffuse layer must be estimated, which requires a model for the electrical double layer in the presence of cationic surfactant adsorption. exp RT - 1 1/ z [C-2] Appendix C 219 It is accepted in the literature that at high concentrations of cationic surfactant (e.g. 1.7xl0"2 M considered in the present case) bilayer formation occurs on the solid surface [Rusling, 1994 I and II]. A model of the double layer based on this concept together with the corresponding potential distribution, is shown in Fig. C - l . 0 / ^ ~ S / 5 j / X y / - V © © HaO+ © 0 cf,sot s*"—'v© Aliquat 336 (cationic part)| Figure C - l The structure of the electric double layer in the presence of cationic surfactant bilayer (based on [Ingram and Ottewill, 1991]). From Fig. C - l , due to super equivalent adsorption at the inner boundary of the diffuse layer a charge reversal occurs, causing a positive charge density as and a positive potential </>s [Ingram and Ottewill, 1991], Furthermore, as is related to the surfactant adsorption density rs by: [C-3] Appendix C 220 The surfactant adsorption density rs (mol m"2) at the inner limit o f the Helmholtz plane is given by: 0. a„N< with [C-4] a = Vhc = 27.4 + 26.9# t e (w c - 1 ) 1 Q - 2 0 ( 2 ) [Rusling, 1994 II] [C-5] cs L h c 1.5 + 1.26n„ where acs is the area occupied per surfactant molecule (m"2) , Vhc is the volume of the hydrocarbon chain region per surfactant (m 3) , Lhc is the chain length (m), nc number of carbon atoms in the chain, Nhc number of hydrocarbon chains, NA Avogadro's number and 0„ is the fraction of the total surface covered by admicelle. A 6S of 0.75 was assumed here [Nikitas, 1997]. Substituting the numerical values in eqns. [C-3] - [C-5] gives: acs = 51. l x l 0 " 2 0 m 2 , rs = 3.25x10"6 mol m"2 and as = 0.235 C m"2. These values compare fairly well with the literature data on cationic surfactant adsorption [Ingram and Ottewill, 1991; Nikitas, 1997]. Once <js has been calculated, the non-linear equation [C-2] can be solved for </>., provided that the ionic concentrations are known. Regarding the ions accumulated at the outer plane of the admicelle (i.e. inner limit of the diffuse layer, Fig. C - l ) two limiting cases were considered. First, it was assumed that only the cationic surfactant (i.e. quaternary ammonium ion and its co-anion, CI") is present at the outer plane of the admicelle. This assumption implies exclusion of the H 3 0 + ions, and yields a theoretical maximum value for the surface pH, i.e. pHS) max. For the second limiting case, it was assumed that all ions of the electrolyte (i.e. quaternary ammonium, CI", H.30 + and SO4 2") are accumulated unhindered at the outer plane of the admicelle and they are contributing to fo. This limiting case gives the theoretical minimum surface p H in the presence of surfactant, pHs, min-Substituting into eq. [C-2] the bulk ionic concentrations based on the above assumptions, and solving for </>s, one obtains: 0.494 V (assumption 1) and 0.065 V (assumption 2), respectively. Finally, from eqn. [ C - l ] , with a bulk p H of 0.9 (i.e. 0.1 M H2SO4), the surface p H values are: pHSi m a x = 9.4 andpHs, m i n is 2.0. Appendix D 221 APPENDIX D: Estimation of the Intra-Admicellar O2 concentration for the Aliquat 336 Surface Film The O2 concentration in the A3 3 6 layer Cozflim, was estimated based on literature studies of the intra-micellar solubility of O2. Employing micellar cetylthrimethlyammonium bromide (CTAJ3) solutions, it was found experimentally [Matheson and King, 1978] that the intra-micellar O2 concentration was 2.8 times the O2 concentration in the bulk. The same ratio was adopted in the present work, to describe the intra-admicellar O2 concentration. Furthermore, the bulk 0 2 concentration, Cozbuik, in the presence of Aliquat 336 is slightly higher than the 0 2 concentration in the 'pure' electrolyte (i.e. without surfactant, see Appendix A). For the cationic surfactant concentration range used in the present work, a factor of 1.1 is recommended by the literature [Matheson and King, 1978]. Hence, the bulk O2 concentrations in the presence of Aliquat 336 are 1.38 m M in 0.1 M H 2 S 0 4 and 1.29 m M in 0.1 M N a 2 C 0 3 . From the latter values together with the 2.8 times factor (see above), Co2,fum in acid and alkaline media was estimated as: 3.86 m M and 3.61 m M , respectively. Appendix E 222 APPENDIX E: Structural Characteristics and Estimation of the Mass Transfer Limited Current Density of 0 2 reduction to H 2 0 2 for Graphite Felt (GF) and Reticulated Vitreous Carbon (RVC) 1. Structural Characteristics of G F and R V C Scanning electron microscopy images are presented for graphite felt and reticulated vitreous carbon 100 ppi, Figures E - l and E-2 respectively, to reveal the structural features of these electrode materials. Figure E - l a, shows the randomly distributed bundles of fibres comprising the graphite felt. On the 30 urn magnification scale (Fig. E - l b) it can be seen that each fibre is composed of several strands, which are mostly parallel to each other. The reticulated vitreous carbon is characterized by a 'honeycomb' structure with well defined pores (Figures E-2 a and b). The dissimilar structural features of G F and R V C induce different fluid dynamics and mass transfer characteristics, which in addition to certain differences in surface functional groups, has important consequences on the electrochemical systems under study. Appendix E 223 Figure E - l : Scanning electron microscopy images of graphite felt. Magnification: a) 600 um; b) 30 um. Appendix E 224 Figure E-2 : Scanning electron microscopy images of reticulated vitreous carbon 100 ppi. Magnification: a) 600 urn; b) 150 urn Appendix E 225 2. Estimation of the Mass Transfer Limited Superficial Current Density for O2 reduction to H 2 0 2 The procedure and the corresponding equations for estimation of the mass transfer limited superficial current density in porous electrodes is described in Section 3.4, eqns. [3.20] -[3.25]. Here a case study is presented for 0 2 reduction at 350 kPa in 0.5 M N a 2 C 0 3 / 0.5 M N a H C 0 3 (pH 9.6) using the graphite felt cathode. General physico-chemical data: Oxygen concentration (mol m"3) Electrolyte conductivity (S m"1) Liquid load (kg m"2 s"1) Gas load at STP (kg m"2 s"1) Maximum allowable potential difference (V) 2.14 (see Appendix A ) 7.9 (measured) 4.5 0.16 0.2 Graphite felt cathode: Uncompressed bed thickness (mm) 7.25 (Table 4.1) Compressed bed thickness (mm) 4.5 ibidem Uncompressed bed porosity 0.95 ibidem Compressed porosity 0.92 ibidem Graphite fiber diameter (m) 2x10"5 ibidem Compressed specific surface area (m 2 m"3) 16000 ibidem Liquid hold-up 0.70 eqn. [3.25] Effective bed porosity 0.64 eqn. [3.22] Effective electrolyte conductivity (S m' 1) 4.3 eqn. [3.22] Mass transfer capacity (s"1) 6.7 eqn. [3.24] Electroactive bed thickness (mm) 0.8 eqn. [3.20] Mass transfer limited superficial current density (A m2) 2200 eqn. [3.23] Appendix F 226 APPENDIX F: Elements of Factorial Experimental Design (based on [Montgomery, 1997] and [Murphy, 1977]) Factorial design is a very useful experimentation technique for determining both the individual and synergistic effects of several factors on selected responses. A common embodiment of this technique involves w/ factors, each at two levels (i.e., ' low' (-) and 'high' (+)). The levels may be either quantitative or qualitative. A complete analysis for such a design requires 2" / experiments, plus a number of replicated observations. However, when « / i s greater than five, usually fractional factorial designs (e.g., half or one-third) are performed to avoid unnecessary experiments leading to superfluous results. Furthermore, to screen for potential non-linear system behaviour, so-called centerpoints could be added to the design (e.g., 2" / +1). The levels of the factors for the centerpoint is approximately set at midway between the ' low' and 'high' levels (the notation for it is '0'). Usually three types of effects are distinguished, main, interaction and curvature effects. a) MAIN effect The main effect is defined as the change in response brought about by a change in the level of the factor (eqn. [F-l]). Main effect of variable Xt Xmj = [ I (responses a t '+ ' X,) - 1 (responses a t ' - ' Xi)] I (half the total number of factorial runs). [F-l] The confidence interval for the main effect is: [F-2] where s2 pooled variance of the response based on m/total degrees of the freedom (/w/equals to the total number of replicated runs in the design minus one), t$ Student's statistic with ntf degrees of freedom at the desired confidence level (e.g., 95%), Nf total number of factorial runs in the design. Appendix F 227 b) INTERACTION effect The interaction or synergistic effect between two factors is a reflection of the fact that the impact of a given factor might not be the same at all levels of another factor. In other words, the effect of a factor A depends on the level of factor B. B y definition, the interaction effect AB is the average difference between the effect of A at the 'high' level of B and the effect of A at the ' low' level of B. c) CURVATURE effect The curvature effect is estimated as the difference between the average of the centerpoint responses and the average of the factorial points. Thus, a strong curvature effect is a reflection of a non-linear system behaviour. The confidence interval for the curvature effect is calculated as: [F-3] where C total number of centerpoints. Once the main, interaction and curvature effects were calculated, the results of the factorial design can be expressed in terms of a regression model (eqn. [F-4]): n n n n y factorial = «0 + £ + Z Z ^ i j X i X j +COcYl Xi 1 Serror • [F"4] i=l i=l j=l i=l In eqn. [F-4], yfactonai is the estimate for response y based on the factorial regression model and a>o is the average of all the factorial responses (i.e., Qo = yavg)- The regression coefficients OH are one-half of the corresponding main effects, the coefficients c% are one-half of the corresponding interaction effects, and coc is one-half of the overall curvature effect. 8err0r is the error estimate. The variables x,-, are the so-called coded variables (or coded factors) given by: ( X ^ ijow \ 2 J fx i.high x ^ ijow 2 J [F-5] where Xf are the factors (i.e., independent variables) of the factorial design (e.g, pressure, temperature etc.). Appendix G 228 APPENDIX G: Determination of the Aqueous / Organic Partition Coefficient for H2O2 The partition coefficient was determined for five organic solvents: 1,2-dichloroethane (DCE), tributyl phosphate (TBP), a 15/85 v mixture of T B P and diethylbenzene ( T B P / D E B ) , propylene carbonate (PC) and 2-undecanone, using three different total H 2 0 2 concentrations, i.e. 0.59, 1.2 and 1.6 M . A H2O2 solution of known concentration was added to the emulsion composed of a 1 to 3 volume ratio of organic solvent and aqueous phase (0.1 M H2SO4 or 0.1 M N a 2 C 0 3 ) to obtain the desired total peroxide concentration per emulsion. The emulsion was mixed for 10 min. at 300 K to achieve steady-state H2O2 distribution. Next the two phases were separated, samples were taken from both phases and analyzed for peroxide content by permanganate titration (Section 4.4.3). The peroxide decomposition rate under the employed conditions and the experimental error were also determined. The partition coefficient for 0.1 M N a 2 C 0 3 is given by: c N iv IN -N -N -C v )/v _^H2Q2,AQU _ H2Q7,AQU 1 r AQU _XY t i v H2Q2,decomp i v H2Q2,dissoc H202.ORG ORG/ AQU [ G i l ~ C ~ C ~ C ' ^H202,ORG K-H202,ORG y-H202,ORG where NT total moles of H 2 0 2 added, Nmo2,dissoc moles of peroxide dissociated, CH202,ORG peroxide concentration in the organic phase, VORG and VAQU are the corresponding phase volumes (5 and 15 ml, respectively). The moles of dissociated peroxide Nmo2,dusoc is expressed as: l y H202,dissoc l u lyH202,AQU- L \ wherepK a is equal to 11.64 [Lange's, 1992] and the p H is 11.5 (for 0.1 M N a 2 C 0 3 ) . Substituting eqn. [G-2] into eqn. [ G - l ] , the partition coefficient for H2O2 into 0.1 M N a 2 C 0 3 becomes: Appendix G 229 if 1 (iV f NH202decomp CH2020RGVORO)IVAQU l + 1 0 M a ) ' c „ [ G " 3 ] •'H202.ORG In eqn. [G-3] both CH202.ORG and NH202,decomP were determined experimentally and the corresponding results are presented in Fig. 6.1 b, of Section 6.1.1. For acid (0.1 M H2SO4) aqueous phase, the partition coefficient is given by: C C _ ^H2Q2,AQU _ ^H2Q2,AQU R^ .-, ~ C ~ (N -C -V )/V ' ^H202,ORG V i ¥r ^H202,AQU AQU ' ORG where CH202AQU was determined experimentally. The resulting K values are presented in Fig. 6.1 a, of Section 6.1.1. Appendix H 230 A P P E N D I X H: Spectrophotometric Investigation of 2-Ethyl Anthraquinone Electroreduction Based on the 'square-scheme' concept (see Fig. 3.7), the reduction of 2-ethyl anthraquinone (EtAQ) can yield a variety of products depending on the composition of the organic solvent (e.g. proton and supporting electrolyte concentration) and on the applied potential. The identification of the major species obtained during E t A Q reduction using both the acid (0.1 M H2SO4) and alkali (0.1 M Na2COs) emulsions, can help to elucidate the mechanism of the mediated O2 electroreduction. To study the above aspects, spectrophotometric experiments were performed (Section 6.2). For both the acid and alkali emulsions, the organic media consisted of 0.1 M E t A Q dissolved in tributyl phosphate (TBP) with 0.2 M T B A P and 0.06 M A336. The initial, pale-yellow organic phase containing 0.1 M E t A Q , exhibits a straightforward absorption spectrum with a maximum at 423 nm (Fig. H - l ) . Subjected to electrolysis, however, the spectrum of the organic phase becomes intricate (Figs. Ff-2 and H-3) and it is dependent on various factors such as emulsion pH, current density and the presence of quaternary ammonium ions ( T B A P and/or A3 3 6). 0) o d CTj .0 b CO < 2.5 2.0 1.5 -1.0 0.5 -i i i i i i i i i | i i i i i i i i i | i i i i i i i i i | i i i i i i i i i | i i i i i i i i IJ 423 nm 0.1 M 2-EtAQ/TBP Ml I I 0.0 [ i i i i n l l l | l l i i i l l l l | l l l l l l l l l | l l l l l l l l l | i i i n i l i T j 400 500 600 700 800 900 Wavelength / nm Figure H - l Absorption spectrum of the 'blank' organic phase, i.e. 0.1 M E t A Q in T B P . T: 295 K . N2 purge. Appendix H 231 O tz ro xi i o X ) < 2 H 0 + l l l i • » « i » i i i i i i i I I | I I I I I I I I I [ | | | | | ! , , , | , | , , , , , , T B A P = 0 .2 M ; A 3 3 6 = 0 .06 M I I ' I I I I I | | | | t | ! • | | ( t l l | | , 400 500 800 900 600 700 Wavelength / nm Figure H-2 Absorption spectra of the organic layer for the acid (0.1 M H2SO4) emulsion as a function of current density and organic phase composition after 10 min. of electrolysis. Current d e n s i t y : 1 0 4 A m"2; 377 A m' 2 . Organic: T B P with 0.1 M E t A Q . N 2 atm. T: 295 K . 2 H CD O c CO X ! 1— o If) .Q < I I I <- I I I I I I I I I I I I I I I 1 I I I I I I I I I I I I I I I I I [ I I I I I I I I I a) 2 min.; TBAP = 0.2 M; A336 = 0.06 M I 1 I I I I I ' ' l ~ [ * ' l'f I T t 'I I *i I I ' I I I I I I I I I I I I I I I I I I 1 1 ) 1 1 { 1 1 1 400 500 600 700 800 900 _] I I I I I I I I I | I ' ' I ' I • ' ' I ' ' I I I t I I I | I I I I I I I I I j I I I I I I ! I 1 b) 10 min., TBAP = 0.2 M; A336 = 0.06 M 4 I ' ' I ' ' ' ' ' | I I I I I I I I I ) I I I I I I I I I [ I I I | I I I I I | 1 | I , ! t 1 , 1 400 500 600 700 800 900 Wavelength / nm Figure H-3 Absorption spectra of the organic layer for the alkali (0.1 M Na 2 C03) emulsion as a function of current density and organic phase composition after: a) 2 min and b) 10 min of electrolysis. Current density: 104 A m"2; 377 A m"2. Organic: T B P with 0.1 M E t A Q . N 2 atm. T: 295 K . Appendix H Table H - l Absorption maxima for Figs. H-2 and H-3. 232 Figure Absorption Corresponding Species Reference Maxima (nm) H-2; H-3 423 E t A Q (yellow) Fig. H - l H-2 468 E t A Q 2 - or possibly E t A Q H 2 (dark orange) 463 nm A Q 2 " [Posdorfer et al, 1991] 470 nm A Q 2 " [Babaeiefa/., 1997] H-3 a H-3 b 450 470 (broad band) E t A Q 2 -(dark orange) idem H-3 a H-3 b 521 560 (broad band) E t A Q H " ' (purple) .542 - 545 nm A Q H " ' [Posdorfer et al, 1991] [Babaei etal, 1997] H-3 b 630 (weak band) E t A Q 2 " (dark green-blue) 622 nm A Q 2 " [Posdorfer et al, 1991] H-3 b 842 nm (broad band) [without current applied is short lived (ca 5 min)] unidentified (dark green) unreported in the literature Appendix H 233 The absorption peaks for Figs. H-2 and H-3, together with the corresponding species are summarized in Table H - l . Figs. H-2 and H-3 in conjunction with Table H - l are discussed in Section 6.2.4. Interestingly, besides the electrochemical reduction of E t A Q , it was observed that by mixing the organic phase with 0.1 M Na2CC>3 under N 2 atmosphere without current applied, the organic layer turns from pale yellow into light red. The absorption spectra of the organic phase after 10 minutes of mixing with 0.1 M N a 2 C 0 3 shows indeed, a peak with a maximum at 496 nm (Fig. H-4). This indicates the reduction of E t A Q by OH", where the hydroxide ion acts as an electron donor [Roberts et al, 1985; Socha et al, 1993]. The mechanism of anthraquinone reduction by OH" involves, in the first step, the formation of an adduct, i.e. EtAQ(OH)", which can react further with E t A Q in a slow electron-transfer step, to form the anion-radical EtAQ"* [Roberts et al, 1985; Socha et al, 1993]. Since there was no peak around 520 - 540 nm that would be characteristic for the radical-anion (Fig. H-4), it is concluded that under the employed conditions, the adduct EtAQ(OH)", is the main reaction product and it is solely responsible for the peak at 496 nm. 3.0 I I I I I I I I I ['I I I I I I I I I [ I I I 1 T T T I I | I I I I I I 1 1 1'| I I I I i I I I I 2 . 5 ^ 2.0 A o to 1.5-3 496 nm < 1.0 A 0.5 A 0.0 400 900 Wavelength / nm Figure H-4 Absorption spectrum of the organic phase after 10 minutes of mixing with 0.1 M N a 2 C 0 3 under N 2 purge. N o current applied. Organic phase composition: 0.1 M E t A Q , 0.2 M T B A P and 0.06 M A336 in TBP.295 K . Appendix I 234 APPENDIX /: Cyclic Voltammograms and Related Experimental Data for the Calculation of the Standard Heterogeneous Rate Constant ks, for 2-Ethyl Anthraquinone Reduction Section 6.2.1 of the main text, discussed the equations and the calculation procedure for obtaining ks. Appendix / gives the cyclic voltammograms and related experimental data. a) Aqueous phase: 0.1 M H2SO4 Cyclic voltammograms for E t A Q were recorded for seven scan rates, in the range of 0.02 and 0.3 V s"1 (Fig. 1-1). The cathodic - anodic peak potential difference AEP for each scan rate was obtained from the voltammograms and Nicolson's kinetic parameter W, was calculated (eqn. [6.11], Section 6.2.1). The corresponding values are given in Table 1-1. Using the l v a l u e s of Table 1-1, was calculated for each scan rate from eqn. [6.10]. The average ks together with the standard deviation are also reported in Table 6.2 of Section 6.2.1. Table 1-1 Experimental data used in the calculation of ks. Scan rate ( V s"1) \ ^ P \ = E , c - E p , a (mV) ks x l O ^ f m s - 1 ) k ns, average x l O - ^ m s - 1 ) 0.02 456 0.01969 3.30 (neglected) 0.05 480 0.01749 4.64 0.10 550 0.01258 4.72 0.15 574 0.01127 5.18 4.97 ±0.26 0.20 600 0.01001 5.31 0.25 636 0.00849 5.04 0.30 660 0.00759 4.93 Appendix I 235 Figure 1-1 Cyclic voltammograms as a function of scan rate used for the calculation of ks for E t A Q reduction. Aqueous phase: 0.1 M H 2 S 0 4 . Organic phase: T B P , 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336. N 2 sat. Electrode: G C . 295 K . Scan rate ( V s"1): 1: 0.02, 2: 0.05, 3: 0.1, 4: 0.15, 5: 0.2, 6: 0.25, 7: 0.3. Appendix I 236 0 -0.8 -1.6 E / V vs. Ag/AgCl Figure 1-2 Cyclic voltammograms as a function of scan rate used for the calculation of ks for E t A Q reduction. Aqueous phase: 0.1 M N a 2 C 0 3 . Organic phase: T B P , 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336. N 2 sat. Electrode: G C . 295 K . Scan rate ( V s"1): 1: 0.02, 2: 0.05, 3: 0.1, 4: 0.15, 5: 0.2. Appendix I 237 b) Aqueousphase.0.1MNCI2CO3 Repeating the above procedure, but for both peaks (Fig. 1-2), the results are summarized in Table 1-2. The standard rate constant is also reported in Table 6.2 of Section 6.2.1. Table 1-2 Experimental data used in the calculation of ksJ and kSt2, respectively. Scan rate | A E I ; |A£ I 1 PU>\ p\2 ks,j, ks2 ks,], average ks,2,average (V s-1) (mV) xlO" 5 (m s"1) xlO" 5 (m s"1) x l O ' 5 (m s"1) 0.02 70; 80 1.9139; 1.0815 3.03; 1.68 0.05 100; 110 0.5418; 0.4223 1.29; 1.04 0.10 110;120 0.4223; 0.3410 1.49; 1.18 1.42 +0.09 0.99 ±0.15 0.15 120;150 0.3410; 0.2051 1.48; 0.872 0.20 130; 160 0.2826; 0.1783 1.41; 0.876 Appendix J 238 APPENDIX / : Experimental Data for the Calculation of the Homogeneous Rate Constant kc The pseudo-first order homogeneous rate constant for E t A Q oxidation kc was obtained from eqn. [6.18] of Section 6.2.3. To apply eqn. [6.18], the scan rate dependence of both the limiting redox regeneration current IL,T and the diffusion current in the absence of regeneration IP4, must be determined (Section 6.2.3). Cyclic voltammograms of E t A Q , similar to those presented in Figs. 6.9 and 6.10, were recorded for several scan rates in both 0 2 and N2 saturated electrolytes. Fig. J - l exemplifies the voltammograms obtained with 0.1 M N a 2 C 0 3 as aqueous phase. To obtain the 'true' h,r from the limiting current values given by Fig. J - l a, for each scan rate the contribution of O2 electroreduction was subtracted (Fig. 6.8). Tables J - l a and b summarize the Ii,r and Ip,d values used for least-square fitting determination of kc based on eqn. [6.18], Appendix J 239 Table J - l Summary of experimental data used for the determination of kc using eqn. [6.17] (Section 6.2.3). a) 0.1 M N a 2 C 0 3 Scan rate Ip.d h,r 1 Ip,d ( V s"1) (uA) (uA) 0.05 8.09 1.05 7.71 0.10 6.32 1.50 4.21 0.20 5.69 2.10 2.71 0.30 5.14 2.60 1.98 0.40 3.34 3.00 1.11 b) 0.1 M H 2 S 0 4 Scan rate h. Ip,d h.r 1 Ip.d ( V s-1) (uA) (uA) 0.002 0.92 0.325 2.83 0.005 0.70 0.46 1.50 0.010 0.31 0.600 0.52 Appendix J 240 i 30 0 -0.7 -1.4 E / V vs. Ag/AgCl Figure J - l Cyclic voltammograms for E t A Q used in the calculation of kc with 0.1 M N a 2 C 0 3 as aqueous phase, a) N 2 saturated; absence of redox regeneration, gives IPi±. b) 0 2 saturated; 'redox regeneration', gives ILy, Organic phase: T B P with 1.1 m M E t A Q , 0.1 M T B A P , 9.6 m M A336. Scan rate ( V s"1): 1: 0.05, 2: 0.1 — 7 : 0.6 (increments of 0.1), 8: 0.8, 9: 1.0. Electrode: G C . 295 K . Appendix K 241 APPENDIX tf: Estimation of the Organic-Liquid to Solid Mass Transfer Capacity for Co-current, Upward, Three-Phase (L/L/G) Flow in Graphite Felt, R V C 100 ppi and R V C 30 ppi The acid emulsion, composed of a 2.1 to 1 volume ratio of organic to aqueous (1 M Na2S0 4 ) phases (Section 6.4.1), was treated as one entity, characterized by an overall conductivity o f 4 S m"1 (assuming bicontinuous emulsion Fig. 6.4 a), average viscosity of 2.23xl0" 3 Pa s* and a linear velocity o f 4xl0" 3 m s"1 (based on the emulsion flow rate). Furthermore, it was assumed that the specific surface area available for the organic to solid mass transfer as,org, is proportional to the volume fraction of the organic phase, i.e., as,org=f-a [ K - l ] where / organic volume fraction (0.32), a total specific surface area of the porous electrode (Table 4.1 for G F and 4.2 for R V C s , respectively). The one-phase liquid to solid mass transfer coefficient for the 30 and 100 ppi R V C , was estimated with the correlation reported by Fenton and Alkire for a 45 ppi R V C [Fenton and Alkire, 1988], i.e., Sh = 11.0 R e 0 3 0 ; where [K-2] s h = ; a n d R e = «„ ; w h i t h aS,orgF)EUQ as,orgV£eff KO,RVC liquid (organic) to solid ( R V C ) , one-phase flow, mass transfer coefficient (m s"1), eeg effective bed porosity (eqn. [3.22]), aSi0rg specific surface area available for the organic phase (m m" ), DEUQ diffusion coefficient o f E t A Q (1.5x10" m s" ), u0 linear velocity of the emulsion (4xl0" 3 m s"1) and v kinematic viscosity of the emulsion (2.23xl0" 6 m 2 s"1). Note that the calculations are performed for 295 K . C a l c u l a t e d as a f u n c t i o n o f the p u r e - c o m p o n e n t v i s c o s i t i e s a n d v o l u m e f r ac t ions , u s i n g L o b e ' s c o r r e l a t i o n fo r l i q u i d m i x t u r e s [ R e i d , et al, 1977] . Appendix K 242 For graphite felt, the mass transfer correlation presented by Schmal et al. [1986] was used, i.e., Sh = 7 R e 0 4 ; w i t h [K-3] Sh = — : and Re = . DEtAQ v In eqn. [K-3] K0,GF liquid (organic) to solid (GF), one-phase flow, mass transfer coefficient (m s"1), df single fiber diameter (2xl0" 5 m; [Oloman et al., 1991], The rest of the variables have the same meaning as in eqn. [K-2]. To approximate the real L / L / G flow situation from the cell, the liquid - solid mass transfer coefficients have to be corrected to take into account the influence of the gas ( 0 2 ) phase, which might affect the liquid to solid mass transfer rate by boundary layer disturbance phenomena. For porous vitreous carbon electrodes operated in bubble flow regime*, Takahashi and Alkire reported that the presence of the gas phase increased the mass transfer coefficients over those in single-phase flow by a factor of ca 1.7 [Takahashi and Alkire, 1985], Extrapolating the two-phase flow data for graphite felt presented by Hodgson and Oloman [1999] to the liquid and gas loads employed in the present experiments indicates also an enhancement of the single-phase mass transfer coefficient by a factor of ca. 1.5-1.7. Therefore, K = 1 . 7 £ [K-4] where KM overall mass transfer coefficient taking into account the presence of the gas phase (m s"1), K0 mass transfer coefficient estimated based on one-phase flow eqns. [K-2] and [K-3] (m s"1). Thus, the estimated mass transfer capacity KMC, is expressed as: Kmc ~ as.orgKm • [K"5] The pertinent numerical values for KM and KMC, are given in Table 6.4 (Section 6.4.1). * The same flow regime was employed in the present experiments (see Section 6.4.1). Appendix L 243 APPENDIX L: Design and Cost Calculations for the Auxiliary Equipment employed in the Acid Emulsion Mediated Peroxide Electrosynthesis Process The calculations in the present Appendix are based on the level 2 flowsheet given in Fig.7.2, using the operational conditions described in Section 7.3. The following equipment is employed: feed tanks, pumps, heat exchangers (coolers), G / L separator, L / L decanter and adsorption column. The uninstalled costs of the process equipment obtained from mid-1980s and early 1990s data, were updated using the Marshall-Swift index corresponding to the 1-st quarter of 2000 (Table L - l ) . Table L - l Relevant Marshall-Swift Indexes (basis: M & S Index = 100 for 1926) Year Marshall - Swift Reference Index 2000 (1-st quarter) 1,076.2 [Chem.Eng., 2000] 1990 924 [Peters and Timmerhaus, 1990] 1985 813 ibidem A) Calculation Procedure (based on [Ulrich, 1984] and [Perry's, 1984]) 1) Feed Tanks Feed tanks are designed to hold fluid volumes for the specified flow rates, assuming a typical residence time of 1800 s. The feed tank for the emulsion catholyte is also equipped with a mechanical agitation (e.g. turbine impeller, see Fig. 7.2). A height to diameter ratio of 1.75 was used for the vertically oriented catholyte feed tank. For the anolyte feed tank (no mixing) the length to diameter ratio is 3 (horizontal orientation). For flow rates of 80 m 3 h"1 (emulsion) and 30 m 3 h ' 1 (anolyte), the required volumes are: 3 3 40 m and 15 m , respectively. Using the aspect ratio, the dimensions of the tanks are: 5.4 m (height) x 3.1 m (diam.) for catholyte and is 5.6 m x 1.9 m for anolyte. Table L - 2 shows the Appendix L 244 corresponding capital costs for vessels made of carbon steel as construction material. For the catholyte mixing-feed tank there is a corresponding electrical energy cost (Table L-2) . 2) Pumps The capital and operating cost of a pump is a function of the required shaft power Ws, [Uhlrich, 1984] W s = ^ , w i t h ^ l - F / 0 2 7 , [ L - l ] where Fv volumetric flow rate (m 3 s"1), AP pressure differential (10x10 s N m" 2 , sp pump efficiency. 3) Water Cooling of the Electrochemical Cells To dissipate the Joule heating, cooling by 'chilled' water (i.e., inlet temperature 10 °C*) is provided to the electrochemical cells. The required cooling water flow rate was estimated from the overall energy balance on the electrochemical cells. The thermal load of vaporization and other heat losses were neglected. Q Joule Q anolyte Q catholyte Q col.water [L-2] where Qj0Uie-^Nc-i-Ac.^c-v0X(m Q a n o i y t e = F , a C p , a ( T a : O U t - T a J Fma ( k g s - 1 ) , ^ =4.18kJkg- 1 K" 1 Q c a t h o i y t e = F m , c C p , c ( T C i O U l - T c . n ) Fmc (kg s"1), C p fi = 1.92 kJ k g ' K " 1 (for explanation see further) Qcoi water = F m i W C P A T » , o u t - T w , , n ) Fmw (kg s"1), C p w = 4.18 kJ kg ' 1 K " 1 . * The typical industrial cooling water is available at 30 °C and is returned to the cooling tower at 45 °C. Appendix L 245 The specific heat for the emulsion catholyte Cpc, was obtained from the mixture molar heat capacity, which was calculated as the mole-fraction average of the organic and aqueous phase molar heat capacities [Reid et al, 1977]. The molar heat capacity of the organic phase (i.e., TBP) was estimated using the Physprop® software of G & P Enginnering [1999]. The software uses the Rowlinson - Bondi corresponding-states method. Thus, the molar heat capacity for T B P was found to be 515.8 kJ kmoi" 1 L - 1 . The inlet and outlet temperatures for the process and cooling water streams are given in Fig. L - l . 313 K 313 K catholyte 3 0 £ K 283 K 'chilled' cooling water anolyte 303 K 303 K Figure L - l . Inlet and outlet temperatures for the process and cooling water streams in the electrochemical cell. From eqn. [L-2] the mass flow rate of cooling water F m , w , was calculated. The cost of cooling water was estimated using a price of 0.03 $ m"3. The reactor cooling was incorporated in the mathematical model of the operating cost associated with the electrochemical unit (see Section 7.5). Appendix L 246 4) Heat Exchanger for Catholyte Stream To reduce the rate of homogeneous peroxide decomposition, the catholyte stream from the electrochemical cells with a temperature of 313 K is cooled to 293 K using a shell and tube, co-current heat exchanger, operated with chilled water of 278 inlet and 283 K outlet temperatures (Fig. 7.2). The design equations are: Q ~  Fm,cC piC(Tc,in,hex ~ ^cfiul.hex) — U • A - ATm — F m w C p w ( T w o u t h e x — Tw i n h e x ) , [L~3] where U overall heat transfer coefficient (1250 W m"2 K" 1 [Perry's, 1984]) and ATm, the mean temperature difference for co-current flow expressed as: A T : = (T - T )-(T v c,in,hex wsinshex / V c. - T ,out,hex w,out,hex > In c,in,hex w,in,hex T \ c,out,hex T. [L-4] w, out, hex J Substituting into eqns. [L-3] and [L-4] one obtains a total required heat transfer area of 11.2 m . The corresponding cost data is given in Table L-2 . 5) Heat Exchanger for the Anolyte Stream As shown in Fig. 7.2, the anolyte is recycled. The outlet temperature of the anolyte from the electrochemical unit is 313 K , therefore, it has to be cooled to a reactor inlet temperature of 303 K (Fig. L - l ) . The chilled cooling water has a temperature range of: 283 K inlet and 293 K outlet. Using the same type o f shell and tube heat exchanger as above and substituting into the pertinent design equations [L-3; L-4] , the required overall heat transfer area is 17 m 2 . The cost data is given in Table L-2 . Appendix L 247 6) G / L Separator The role of the gas / liquid separator is to release the excess O2 from the catholyte stream. Typically, it is a vertical column with packing in the upper part to separate the entrained liquid droplets. The column was designed for an O2 gas flow rate of 120 m 3 h"1 (at STP) (see Section 7.3) and an aspect ratio of 3. The separator was sized using eqns. [L-5] - [L-6] [Walas, 1987], vgas= 0.03041 -P em - 1 KPgas j - l l / 2 and D. F V,gas sep 0 7 5 -v [L-5] [L-6] where vgas velocity of 0 2 in the separator (m s"1), pem density of emulsion (kg m"3), pgas density of 0 2 (kg m"3), FVigas volumetric flow rate of O2 (m 3 s'1) and Dsep diameter of the G / L separator (m). Substituting in eqns. [L-5] and [L-6], the required diameter is 0.3 m. The corresponding separator height, including 0.3 m of disengaging space both ahead and above the mesh pad, is 1.5 m. The cost data is given in Table L-2 . 7) L / L Decanter The L / L separator was designed for a settling velocity of 8.5xl0" 4 m s"1 [Walas, 1987] and an aspect ratio (length to diameter) of 3. The diameter of the decanter is given by: 4R V,em 0.5 with —4- > 2 [L-7] where Fv,em L / L emulsion flow rate (m 3 s"1), uav average velocity of the emulsion stream (m s"1), ut terminal settling velocity (m s"1), Ld, Dd length and diameter, respectively . Substituting the numerical values in eqn. [L-7], the dimensions o f the decanter are: 7.1 m length and 2.4 m diameter. Appendix L 248 8) Adsorption column The column was designed to remove 99% of the T B P with a 'breakthrough' time* h of 7 days at a throughput ratio Tr o f 0.95. The throughput ratio for the present case is defined as: V -F v Tr= aq" " [L-8] stoic where Vaqu aqueous solution volume that has entered the bed since the start up to the breakthrough time (i.e. Vaqu = Fv,aqu • h), vb bed volume, Sb bed void fraction (0.5) and Vstoic the theoretical aqueous solution volume containing enough organic to saturate the bed volume completely. Vmoic and Vb are related by the partition ratio A , i.e. * ^ C = A - V [L-9] Furthermore, the partition ratio A can be expressed as a function o f the sorptive capacity of the packing Sc, and the desired purification (i.e. 99%). Thus, [L-10] where /% dry bed density (650 kg m"3 ), Sc sorptive capacity of the adsorbent (0.5 kg kg" 1 ), cin, cout inlet and outlet solute (TBP) concentrations (kg m"3). In eqn. [L-10], cin - cou, = 0.99c,„. The solubility o f T B P in water is 0.59%*, therefore, cin = 5.9 • 10" 3p a q u, w i t h p m u 1130 kg m"3 (1 M N a 2 S 0 4 , 20 °C). Substituting the above numerical values into eqn. [L-10] yields a partition ratio A of 49.2. i.e. the time after which the column cannot achieve the desired purification and needs to be regenerated. Appendix L 249 Using A and substituting eqn. [L-9] into eqn. [L-8], the required bed volume v& is expressed by: V> = Tr7'tb ' [L"11] Tr-A + sb From eqn. [ L - l l ] , for the specified conditions, the required bed volume is 53 m 3 (e.g., three columns in series, each of 4.5 m height and 2.2 m diameter, using an aspect ratio of 2). The installed capital cost of the adsorption unit is given in Table L-2 . Comparing the costs of the different auxiliary equipment (Table L-2) , as expected, the separation (adsorption) is the most expensive unit operation. The total installed capital cost for all auxiliary, non-electrochemical, equipment is 725,000 $. The corresponding operating cost (i.e. utilities plus labour and maintenance cost which is 5%Peryear of the total installed capital cost) is 7.8 $ h"1. Appendix L 250 B) Summary of Design Calculations for the Auxiliary Equipment Table L-2 Capital and utilities costs for the auxiliary equipment employed in the acid emulsion mediated peroxide electrosynthesis process (flowsheet given in Fig. 7.2) Equipment Uninstalled Cost (US$per200o) Installation Factor Installed Capital Cost (US$) Power Consumption (kW m 3) Electrical Energy / or other Utility (US h-1) 1. Feed Tank (for e m u l s i o n ca tho ly te ) 15,000 3 45,000 0.2 0.24 2. Feed Tank (for ano ly te ) 12,000 3 36,000 3. Pump (for e m u l s i o n ca tho ly te ) 7,000 5 35,000 33 1 4. Pump (for ano ly te ) 8,000 4.5 36,000 15 0.45 5. Heat Exchanger (for ca tho ly te ) 4,800 3 14,400 1.4 (for 48 m 3 IT 1 c o o l i n g water ) 6. Heat Exchanger (for ano ly te ) 6,600 3 19,800 1.0 (for 34 m 3 h " 1 c o o l i n g water) 7. G / L Separator 16,950 4 67,800 8. L / L Decanter 7,000 3 21,000 9. Adsorption Column 90,000 5 450,000 T O T A L 725,000 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0058669/manifest

Comment

Related Items